Search

Cutting tool wear prediction plays an important role in the machining of complex aerospace parts, and it is still a challenge under varying cutting conditions. To overcome the limitations of the existing methods in generalization ability when dealing with cutting conditions changing largely, this paper proposed a novel cutting tool wear prediction method based on continual learning. A meta-LSTM model is firstly trained for specific cutting conditions and can be easily fine-tuned with very small number of samples to adapt to new cutting conditions. Specifically, the meta-model could be continuously updated as machining data increase by using an orthogonal weights modification method. The experiment results show that the proposed method can realize accurate prediction of tool wear under different cutting conditions. Compared with existing methods including meta-learning methods, the range of adapted cutting conditions could be expanded as the task distribution of new cutting conditions is continuously learned by the prediction model.

Deformation resulting from residual stress has been a significant issue in machining. As allowance allocation can directly impact the residual stress on part deformation, it is essential for deformation control. However, it is difficult to adjust allowance allocation by traditional simulation methods based on residual stress, as the residual stress cannot be accurately measured or predicted, and many unexpected factors during machining process cannot be simulated accurately. Different from traditional methods, this paper proposes an allowance allocation method based on dynamic approximation via online inspection data for deformation control of structural parts. An Autoregressive Integrated Moving Average (ARIMA) model for dynamic allowance allocation is established so as to approach the minimum deformation, which is based on the in-process deformation inspection data during the alternative machining process of upside and downside. The effectiveness of the method is verified both by simulation cases and real machining experiments of aircraft structural parts, and the results show that part deformation can be significantly reduced.

An appropriate machining sequence can benefit the deformation control of structural parts. However, the existing machining sequence optimization methods can only take single-sided parts into consideration, which is not sufficient to the deformation control of the double-sided parts. To this end, this paper proposes a double-sided collaborative optimization method of machining sequence for deformation control of structural parts. Off-line numerical analysis and on-line deformation monitoring are combined in the proposed method, i.e., the relative residual stiffness of the workpiece in different machining sequence is obtained by off-line numerical analysis, and the residual stress distribution is reflected by on-line deformation monitoring data, respectively. Then, a fuzzy comprehensive evaluation model aiming to decrease the overall deformation is established to realize the collaborative optimization of machining sequence. The machining efficiency is also considered in this model by optimizing the number of alternate flip-over, where the machining features of the part with two sides are machined alternatively according to the optimized sequence. The experiment shows that the proposed collaborative optimization method of machining sequence can decrease the machining deformation of double-sided parts effectively.

Large scale parts with freeform surfaces used in aerospace products usually need to be measured during the machining process with high accuracy and spatial resolution. Touch probe and laser scanner are two kinds of measuring sensors widely applied in manufacturing industry, but each of them is not up to this task independently due to the high requirements in both accuracy and efficiency. This paper proposes a multi-source integrated fusion (MSIF) framework for surface measurement. Sufficient measurement data from multiple datasets generated by laser scanner and few measurement data from touch probe are fused by residual approximation to eliminate the systematic error; then, initial modified models are integrated to reduce the random error. Two algorithms are developed according to the proposed framework, named MSIF-W (multi-source integrated fusion by weighting) and MSIF-S (multi-source integrated fusion by stacking). Both synthetic simulation and actual measurement on three real aircraft skin surfaces are conducted to verify the performance of the proposed method. The results imply that the proposed MSIF can effectively improve the accuracy and precision in fusing multi-sensor datasets.

Deformation due to residual stress is a significant issue during the machining of thin-walled parts with low rigidity. If there are multiple processes with deformation during machining, some process suitability issues will appear. On this occasion, the actual geometric state of the deformed workpiece is needed for process adjustment. However, it is still a challenge to obtain the complete geometry information of deformed workpiece accurately and efficiently. In order to address this issue, a time-varying geometry modeling method, combining cutting simulation and in-process measurement, is proposed in this paper. The deformed workpiece model can be reconstructed via transforming the deformed workpiece with only a small amount of the measurement points by superimposing material removal and workpiece deformation simulation according to a time sequence, which takes advantage of the proposed Curved Surface Mapping based Geometric Representation Model (CSMGRM). Machining experiment of a typical structural part has shown that the deformed geometry model of the whole workpiece can be reconstructed within the error of 0.05mm, which is less than one tenth of the finish machining allowance in general cases, and it is sufficient to meet the accuracy requirements for interference or overcut/undercut analysis and process adjustment.

In order to ensure machining stability, curvature continuity and smooth cutting force are very important so as to meet the constraints of both cutting force and kinematics of machine tools. For five-axis flank milling, it is difficult to meet both of the constraints because tool path points and tool axis vectors interact with each other. In this paper, multiple relationships between tool path points and tool axis vectors with cutting force and kinematics of machine tools are established, and the strategies of corner-looping milling and clothoidal spirals are combined so as to find feasible solutions under both of the constraints. Tool path parameters are iterated by considering the maximum cutting force and the feasible range of the tool axis vector, and eventually a curvature continuity five-axis flank milling tool path with smooth cutting force is generated. Machining experimental results show that the conditions of cutting force are satisfied, vibration during the process of machining is reduced, and the machining quality of the surface is improved. Keywords: Curvature continuity, Cutting force, Five-axis flank milling, Machining, Tool path generation

The concept of additive–subtractive hybrid manufacturing provides a new idea for the manufacturing of high-precision complex structural parts. Currently, under the five-axis additive–subtractive hybrid manufacturing mode, existing research work concerned with sequence planning issues have limitations. This article presents a sequence planning method for hybrid manufacturing of complex structural parts with high precision. The initial printing direction of parts was determined based on an iterative search method and the initial hybrid manufacturing sequence was constructed by part volume decomposition, which solved the coupling problem of printing direction decision and machinability calculation. Under the constraint of tool accessibility, the whole planning of the hybrid manufacturing sequence was realized based on greedy algorithm. This method has achieved highly effective planning of the alternative sequence in the process of hybrid manufacturing, thus greatly reduced the number of tool changes required and laid a foundation for the realization of highly efficient hybrid manufacturing.

The traditional 2.5-axis volume printing process purely relies on planar and parallel slicing layers, which imperatively requires the support structure when dealing with overhanging features on the part. The advent of multi-axis additive manufacturing inaugurates a brand new type of printing process with an adjustable build direction, based on which the support structure can be successfully reduced (if not completely eliminated) upon a proper process planning. Presented in this paper is a curved layer based process planning algorithm for multi-axis printing of an arbitrary freeform solid part. Given a freeform solid model represented as a watertight mesh surface, our algorithm starts with the establishment of a surface embedded field, whose value at any particular point is exactly the geodesic distance to the specified bottom of the model. Any iso-level contour induced from this field is first flattened, filled by a Delaunay triangular mesh, and then mapped back to 3D space through the Harmonic mapping to interpolate the original 3D contour, thus generating a curved layer. After the entire model is decomposed into curved layers by the proposed adaptive slicing strategy, the multi-axis printing paths are then generated on these layers in a contour-parallel fashion. Finally, following the strict increasing order of iso-levels, the contours are printed one by one till the final formation of the part. Preliminary tests in both computer simulation and physical printing of our algorithm have given a positive validation on its effectiveness and feasibility in eliminating the need of support structure.

Data driven approach has been an effective means to realize intelligent manufacturing, and it is very important to get really valued data. It is difficult to realize real-time closed-loop control of complex machining conditions by only using monitoring signals without considering the geometry and process information of the workpiece which is being machined. Therefore, the synchronous association of geometry, process and monitoring information is the key to realize data-driven intelligent manufacturing. However, geometry, process and monitoring information is multi-source, multi-modal and heterogeneous data, and both static and dynamic information is involved, so the effectively synchronous association of the information is still a challenge. In order to address the issue mentioned above, the real-time synchronous association relationship among the three kinds of information is established by using the theoretical discrete cutter position points and the actual sampling cutter position point as linkages. In addition to that, time sequence matching and coordinate matching are combined for avoiding miss match of coordinate. Experimental results show that the proposed method can realize synchronous association of geometry, process and monitoring information in real time, and the average lag time for synchronous association at each cutter position point is far less than 0.1 ms, which is shorter than the minimum time period for signal acquisition.

Pre-deformation of the workpiece in the elastic range is an effective deformation control method, as the machining deformation can be balanced by the pre-deformation. Different from the traditional pre-deformation control method based on experience or inaccurate prediction before machining, an innovative idea of online active pre-deformation control based on deformation monitoring data is proposed in this paper. The deformation monitoring data are used to analyze the workpiece state and predict the deformation of the next layer based on a cubic spline curve fitting approach, and then Kalman filter algorithm is used to control the error between the prediction deformation and the actual deformation so as to ensure the pre-deformation control accuracy. A real aircraft structural part is used to validate the proposed method, and the results show that the part deformation can be significantly reduced.

Chatter suppression is a classic research topic for improving productivity in real manufacturing industry. One of the most challenging tasks in chatter suppression is to predict tool tip dynamics in the whole workspace, especially the multimode conditions. Traditional prediction methods based on finite element analysis or kinematic modeling are either time-consuming or inaccurate. To address this subject, this paper proposes a transfer learning based multimode tool tip dynamics prediction method. A tool with multimode dynamics at the tool tip is first selected as the source tool and sufficient impact tests are carried out within its workspace to construct the source tool tip dynamics dataset. For a new tool, namely the target tool, the workspace is first divided according to the modal orders of the FRFs in the source dataset. In each of the sub-workspaces, only few impact tests are required to establish a target dataset for training the tool tip dynamics prediction model assisted by the source dataset with a transfer learning algorithm. Finally, the tool tip dynamics prediction model of the target tool in the whole workspace can be constructed by integrating the prediction models in all sub-workspaces. To demonstrate the performance of the proposed method, a five-axis machine tool along with three different cutting tools are selected for providing a detailed experimental validation.

This paper inaugurates a brand new five-axis tool path generation method for freeform surface machining, which is solely based on digital image processing. By utilizing image representation, a freeform surface can be digitized into a uniform grayscale image. With this novel representation, we come up with a region based five-axis tool path scheme with remarkably enhanced utility towards freeform surface machining process. The image representation possesses sufficient geometric data for us to evaluate the surface normal distribution and split the surface area accordingly via image processing algorithms. Since each partitioned region contains a limited surface inclination range, one rotary axis is locked with a certain angle during the processing of this region. In addition, by further exploiting the potential of the image representation, the cutter contact (CC) curves are extracted purely from the image of each partitioned region. Preliminary results show that this method is more than adequate for the tool path computation. The computed tool path outperforms the benchmarks by reducing the machining time for as much as 40%, as verified in the physical cutting test.

Carbon fiber reinforced plastic self-resistance electric (SRE) heating has been conceived as an alternative to out-of-autoclave technology due to its characteristics of uniform heating, fast heating/cooling, low energy consumption, and low equipment investment. In this work, a series of SRE heating experiments were conducted, in which the temperature distribution field, energy consumption, and curing time of SRE curing process were characterized. Comprehensive mechanical tests and microscopic characterization were carried out. The experimental results exhibit that the rapid heating rate of SRE curing process resulted in a weaker matrix performance because of the insufficient time of void elimination, which finally leads to an inferior compression and flexural strength for the composite part, while the fiber preferential heating effect can significantly improve the fiber-resin interfacial strength, because the naturally formed temperature difference along the interfacial area enhanced the adhesive strength of the resin around the interface, which improved the macroscopic tension and interlaminar shear strength.

For the machining of freeform surface parts with large size and thin walls, the actual shape of the blanks is always inconsistent with the desired shape, unexpected overcut or undercut will be introduced by using the tool paths generated according to the nominal part model. According to the online inspection results of large-sized and thin-walled freeform surface part, we found that the error distribution of the blank has different magnitude and form over the whole surface of the part. In this paper, the error distribution is mathematically defined based on online inspection data, and new tool paths adaptive to the actual shape of the workpiece are generated on the basis of the tool paths of nominal model, different tool path transplantation strategies are adopted according to the error distribution for different surface areas. A typical aircraft skin part with dimensions of 3867 mm × 1905 mm × 6 mm is used as a case to validate the proposed approach, the actual cutting experiment results show that the wall thickness error ranges from –0.089 mm to + 0.162 mm, and the shape error is within 0.46 mm, which can both meet the machining requirement of the aircraft skin parts.

Freeform surfaces are used more and more widely in products of aerospace, automotive and watercraft due to excellent aerodynamic and hydrodynamic performance. Could manufacturing provides a promising solution to satisfy the increasing demands in high quality freeform surface machining capability but significantly increase the shop-floor uncertainties, which necessitates adaptive planning and execution of machining services. To address this issue, this paper first introduces the concept of surface dynamic machining feature (SDMF) which considers feature dynamics across different machining steps and uncertain machining resources. Then a machining service modeling method for feature-based freeform surface machining is put forward by combing SDMF with function blocks (FB). Based on the event-driven mechanism of SDMF-FB, the adaptive response of the machining service to machining resource changes could be started from feature model adjusting for a real optimal machining result. At last, a surface part from the real industry is selected as the case to show the advantages of the proposed method.

Composite microwave curing technologies have been researched and given great expectations for decades to cutting down the long curing cycle and enormous energy consumption during traditional curing process. However, this good vision was stopped since the microwave can hardly penetrate and heat multidirectional carbon fiber reinforced polymer composites . In this paper, the mechanism that multidirectional composite can hardly be heated was revealed. New method was found to stimulate Vertical Penetrating Microwave (VPM) by using metal strips, and the multidirectional carbon fiber reinforced polymer composites can be heated directly. The theory and model of VPM’s heating depth of multidirectional composite were established and verified. The results indicated that 2.3 mm thickness multidirectional composite can be cured effectively by using the VPM.

The resin micro-flow between fibers influences bubble evacuation and fiber impregnation during the processing of fiber reinforced polymer (FRP) composites. The trapped bubbles and unimpregnated regions within a laminate will develop into voids after the processing, which will impair the structural performance severely. In this study, a new sensing principle of the step-changing difference of capillary pressure was proposed for the micro-flow measurement between reinforced fibers in FRP composites. A highly sensitive and calibration-free micro-flow sensor was designed by simply introducing an optical fiber with super-infiltration polyimide coating along the reinforced fibers. The FRP composite around the optical fiber was regarded as part of the sensing structure. The key resin flow states between fibers were first detected continuously during composite processing.

Carbon fiber reinforced plastics have been widely applied in aircraft structures, which require more rapidly repairing process when carbon fiber reinforced plastics structures are damaged. A novel method of self-resistive electrical heating to rapidly repair carbon fiber reinforced plastics parts was investigated, which utilizes the electrical current passing through the carbon fiber to generate heat and directly cure the whole carbon fiber reinforced plastics patches. A portable self-resistive electrical repairer with automatic temperature control unit was developed. Repair experiments of self-resistive electrical and the electrical blanket heating were conducted, in which the heating temperature distribution field was observed by the thermal infrared imager. The degree of cure of different samples was characterized, and the cross-section morphology and void content of different samples were assessed. Flexural and inter-laminar shear tests were carried out. The experiment results showed that the flexural and interlaminar shear strength recovery ratio of self-resistive electrical repaired samples attained to 85.48% and 65.89%, which was the same or even slightly higher level as the blanket repairing process, but only cost 62% repairing period and 58% repairing energy of the later one. Finally, from the experimental results, the heat transfer behavior of different processes was analyzed.

Machining chatter has a great influence on the producing efficiency and surface quality. Frequency response function at the tool tip is a crucial input for constructing accurate milling stability model. However, the tool tip dynamics usually change with the continuously varying postures of the machine tool axes during the whole machining process. How to predict the pose-dependent tool tip dynamics precisely has become one of the most challenging tasks in chatter suppression in both research and industry. Compared to traditional finite element analysis or kinematic modeling based methods, this paper proposes a data-driven method using transfer learning to predict the pose-dependent tool tip dynamics for different tool-holder assemblies. Firstly, a tool-holder assembly is selected as the source tool and its pose-dependent tool tip dynamics are obtained as the source data by sufficient impact tests. For a new tool-holder assembly, namely the target tool, only few impact tests are required to measure the tool tip dynamics as the target data. Then both the target and source data are used to train a regression model for predicting the target pose-dependent tool tip dynamics based on transfer learning by integrating domain adaptation and adaptive weighting. Furthermore, a detailed experimental validation with a five-axis machine tool is carried out to verify the accuracy and efficiency of the proposed method.

Fibre reinforced filament fabrication is a potential additive manufacturing method for certain load-bearing parts in aerospace and automotive products. In current practice, printing paths are planned from part geometry without considering loading conditions. This paper presents a new method for optimizing printing paths to align with the principal stress field of parts in use. Because of the powerful processing capabilities, for the first time grayscale image was adopted to represent the irregular vector field, which can be robustly processed into sub-regions for generating regular printing paths. Preliminary bending test achieved promising increase in tensile strength compared with conventional methods.

Accurate, rapid and automated tool wear inspection is critical to manufacturing quality, cost and efficiency in smart manufacturing systems. However, manual inspection of tool wear is still a common industrial practice which is inefficient, prone to human errors and not suitable for digitized manufacturing. Previously reported automatic tool wear inspection methods were inaccurate because they only used the remaining worn boundary (i.e., the partial-absence original tool boundary) to approximate tool wear. The authors discovered the association principle between the change law of the cutting edge grayscale and the relative position of the original and worn boundary, which was used to establish the probability functions to accurately reconstruct the curved original tool boundary via Bayesian Inference. The experiment results reported in this paper proved higher efficiency and accuracy than previous automatic tool wear inspection methods.

The rapidly increasing demand and complexity of manufacturing process potentiates the usage of manufacturing data with the highest priority to achieve precise analyze and control, rather than using simplified physical models and human expertise. In the era of data-driven manufacturing, the explosion of data amount revolutionized how data is collected and analyzed. This paper overviews the advance of technologies developed for in-process manufacturing data collection and analysis. It can be concluded that groundbreaking sensoring technology to facilitate direct measurement is one important leading trend for advanced data collection, due to the complexity and uncertainty during indirect measurement. On the other hand, physical model-based data analysis contains inevitable simplifications and sometimes ill-posed solutions due to the limited capacity of describing complex manufacturing process. Machine learning, especially deep learning approach has great potential for making better decisions to automate the process when fed with abundant data, while trending data-driven manufacturing approaches succeeded by using limited data to achieve similar or even better decisions. And these trends can demonstrated be by analyzing some typical applications of manufacturing process.

Monitoring and predicting tool wear is an important issue in dynamic process control under changing conditions, especially for machining large-sized difficult-to-cut materials used in airplanes. Existing tool wear monitoring and prediction methods are mainly based on given cutting conditions over a period of time. This paper presents a novel method for accurately predicting tool wear under varying cutting conditions based on a proposed new meta-learning model which can be easily trained, updated and adapted to new machining tasks of different cutting conditions. Experiments proved a substantial improvement in the accuracy of predicting tool wear compared with existing deep learning methods.

Naphthalimides, such as amonafide and mitonafide in clinical trials, have been developed as antitumor agents for orthotopic tumor. However, the serious side effects in cancer patients limit their applications. Herein, a new class of polyamine-based naphthalimide conjugates 5a-5c, 7a-7b, and 11a-11b with and without the alkylation of the distant nitrogen in the polyamine chain were synthesized and the mechanism was determined. Compared with amonafide, dinitro-naphthalimide conjugate 5c with a 4,3-cyclopropyl motif preferentially accumulates in cancer cells and minimizes side effects in vitro and in vivo. More importantly, 5c at the dosage of as low as 3 mg/kg (57.97%) displays better antitumor effects than the positive control amonafide (53.27%) at 5 mg/kg in vivo. And a remarkably elevated antitumor activity and a reduced toxicity are also observed for 5c at 5 mg/kg (65.90%). The upregulated p53 and the apoptotic cells (73.50%) indicate that the mechanism of 5c to induce apoptosis may result from its enhanced DNA damage. Further investigation indicates that in addition to target DNA, 5c can modulate the polyamine homeostasis by upregulating polyamine oxidase (PAO) in a different way from that of amonafide. And also by targeting PTs overexpressed in most of cancer cells, 5c downregulates the contents of Put, Spd, and Spm, which are in favor of suppressing fast-growing tumor cells. Our study implies a promising strategy for naphthalimide conjugates to treat hepatic carcinoma with notable activities and reduced toxicities at a low dosage.

Pocket milling has long been a popular means for machining pocket features in structural parts and skins in the aviation industry. Recent advanced milling technologies pose new challenges for pocket milling path which existing contour-parallel path generation schemes cannot overcome. For high-speed machining, pocket milling path is desired to be smooth and with no tool retractions during the process, while the path stepover should be kept within a prescribed range to achieve relatively constant cutting load. These geometric constraints are also vital in the application of aircraft skin mirror milling in order to guarantee a correct and consistent thickness signal reception for real-time adjustment of the process. Traditional path optimization based on local modification can only meet a few of these constraints while others are being violated. Therefore, we propose a novel contour-parallel path generation scheme that respects all these process constraints by utilizing the idea of image morphology. The two-step scheme first generates an initial path by propagating from the rectified medial curve of the pocket shape. The initial path is then treated as a binary image being iteratively deformed and projected back into the pocket region via quadratic optimization. Experimental results show that our developed scheme can generate a smooth, tool retraction free and stepover-guaranteed path for various shapes of pocket.

Deformation prediction is the basis of deformation control in manufacturing process planning. This paper presents an on-line part deformation prediction method using a deep learning model during numerical control machining process, which is different from traditional methods based on finite element simulation of stress release prior to the actual machining process. A fourth-order tensor model is proposed to represent the continuous part geometric information, process information, and monitoring information, which is used as the input to the deep learning model. A deep learning framework with a conventional neural network and a recurrent neural network has been constructed and trained by monitored deformation data and process information associated with interim part geometric information. The proposed method can be generalised for different parts with certain similarities and has the potential to provide a reference for an adaptive machining control strategy for reducing part deformation. The proposed method was validated by actual machining experiments, and the results show that the prediction accuracy has been improved compared with existing methods. Furthermore, this paper shifts the difficult problem of residual stress measurement and off-line deformation prediction to the solution of on-line deformation prediction based on deformation monitoring data.

Conventional tool path generation algorithms are usually dependent to the surface geometry and thus case-sensitive. A specific tool path planning method for regular parametric surfaces cannot directly handle a triangular mesh surface, and vice versa. Presented in this paper is a unified digital image representation of a surface, along with a general process of tool path generation and optimization directly based on such representation. Regarding different objectives and utilities, three typical digital images are introduced to represent a projectable surface. Each digital image induces a uniform discrete scalar/vector field indicating the surface geometric property, which can be further applied to the cutter contact curve generation, tool orientation determination, and the cutting simulation tasks. Preliminary examples give a sneak peek of the capability of the proposed method. It is observed that, instead of utilizing geometry-based algorithm, some global optimization task of the tool path is transformed into finding a proper convolutional kernel and its parameter for the image processing. Though more investigation is worth spending in the future, the proposed approach inaugurates a standard and effective way to facilitate the tool path generation and optimization task for surface machining.

Contour-parallel tool path has been a popular means in milling 2D pocket regions. Traditional contour-parallel path suffers from unsmooth transition and uncut problem around corners, which requires dedicated local geometric treatment. In addition, these two issues contradict with each other; solving one may jeopardize the other. Motivated by these problems, we present a digital image approach to deal with the contour parallel tool path optimization. By initially converting the pocket boundary into a binary image, a greyscale image is constructed via an inward-marching algorithm, such that the greyscale value of every pixel inside the pocket indicates the shortest Euclidean distance (SED) towards the boundary. The contour-parallel tool path can be easily derived from this SED image as the iso-contours. Distinguished from the geometry-based treatments towards unsmooth tool path and uncut regions, the proposed framework takes advantage of various image processing techniques (such as Gaussian blur and unsharp masking) to the SED image. By adaptively optimizing the parameters for image blurring and sharpness, the corresponding iso-SED tool path eliminates the uncut residues while still satisfies a specified level of smoothness. A quantitative analysis together with an image optimization scheme is also presented in this work.

In traditional machining process of large-scaled structural parts, the process-planned pre-machining process cannot respond to the unpredictable factors during machining process, and therefore the ...

Machining features have been deemed as an effective way to accumulate and reuse machining process knowledge. The research gap for machining feature based method is how to define machining features, as the geometric shapes and machining processes of the same kind of machining features are only similar but not seriously unique. In order to address the issue mentioned above, a machining feature definition approach based on historical data for process knowledge reuse via two-times unsupervised clustering is proposed in this paper. Machining feature definition is realized based on learning machining feature patterns using unsupervised clustering by taking advantage of historical data. The feasibility of the proposed approach is validated by some aircraft structural parts, which provides an important theoretical reference for process planning and process reuse.

For thin-walled parts, uniform allowance to each machining surface is allocated by the traditional machining method. Considering the sequence of the adjacent machining features, it may cause poor stiffness for some side walls due to a minor wall thickness, which may cause the deformation of the final formed parts to be large, or deduce machining efficiency for some machining features due to too thick remains. In order to address this issue, a non-uniform allowance allocation method based on interim state stiffness of machining features for the finishing of thin-walled structural parts is proposed in this paper. In this method, the interim state model of machining features is constructed according to the machining sequence of the parts, and the stiffness of the side wall is taken as the evaluation index to allocate reasonable allowance value to the corresponding machining surface to ensure the stiffness requirement of the parts in the machining process. According to the finite element simulation results, the non-uniform allowance allocation method proposed in this paper can effectively improve the stiffness of the parts and reduce the deformation of the parts, when compared with the traditional uniform allowance machining method.

In freeform surface machining, the cutter should move along the feed directions with maximal machining strip widths (MSW) as much as possible to improve machining efficiency. However, since the optimal feed directions usually vary among different regions over a complex freeform surface, only local optimal machining result could be achieved by most existing tool path generation methods which consider the whole surface as one machining region. This research proposes a tensor-based approach to generate regional tool paths for more globally optimal machining result. A rank-two tensor is first constructed to enable the evaluation of the MSW over the whole feed direction space at the cutter contact point (CCP). To make this MSW tensor usable for generic APT cutters, the concept of effective cutting surface (ECS) is raised. The tensor is then obtained by modeling the geometric approximation between the ECS and the part surface around CCP. A MSW tensor field is then induced over the entire part surface. The regional distribution of the optimal feed directions is associated with the degenerate points within the tensor field. Then, the surface may be divided into several machining regions by constructing inside boundaries starting from the trisector degenerate points. At last, tool paths will be separately calculated in each sub-surface. Since the optimal feed directions in each sub-surface vary continuously, the cutter movements can follow the optimal feed directions closely based on the generated regional tool paths which refer to shorter tool path length and machining time. Two parts with freeform surface are selected from the real industry as test cases and the comparisons to other traditional methods are also provided.

Recently, indirect microwave curing technology was developed to process multidirectional carbon fiber reinforced composite materials with high efficient and energy saving purpose. The aim of this paper is to solve the problem of large mid-plane heat generation in manufacturing this kind of materials with large thickness. The difference between the traditional thermal curing and indirect microwave curing processes was analyzed. A Multiphysics model was constructed to reflect the composite curing behavior in a cost-effective manner, which agreed well with the experimental results. On this basis, a new indirect microwave curing cycle was designed, and relevant process parameters were determined. Compared with the manufacturer’s recommended cycle, the degree of mid-plane heat generation was greatly relieved, and a reduction of 34.6% in composite residual strains was achieved. Moreover, the curing cycle was reduced by 38%, while the interlaminar shear strength of the composite was improved by 1.38 times. Corresponding reinforcement mechanisms were explored through the observation of composite cross-sections with optical microscopes.

Microwave curing technologies have many advantages in manufacturing fiber reinforced polymer composite materials used in aerospace products, compared with traditional autoclave curing technologies. However, multidirectional carbon fiber reinforced polymer composites can hardly be penetrated and heated by microwave directly, which has become a major obstacle in industrial application worldwide. In this paper, an indirect microwave curing method was proposed to solve this problem. The microwave absorption performance of the indirect microwave heating medium was systematically optimized by evaluating its dielectric properties and reflection loss according to the transmission line theory. On this basis, the microwave susceptive mold was carefully designed and manufactured. Subsequently, the multidirectional carbon fiber/epoxy composite was successfully cured with indirect microwave heating, which was demonstrated by the observation of the curing process with infrared thermal imager and differential scanning calorimetry analysis of the final products. Compared with the traditional thermal curing method, the curing cycle and energy consumption were reduced by 42.1% and 75.9% respectively. Results of further characterization experiments indicated that the mechanical properties of indirect microwave cured specimens were slightly higher than those of the thermally cured counterparts.

Due to the geometric complexity, tool orientations usually change dynamically during freeform surface machining with 5-axis machine tool. As the kinematic performance of the rotary axes is usually weaker than that of the linear axes, the real cutting speed is difficult or even impossible to reach the desired level, which further leads to low machining efficiency. This paper presents a region-based 3 + 2-axis machining toolpath generation method with 5-axis machine tool. The surface is first divided into several preliminary sub-surfaces using K-means clustering algorithm. A post processing procedure is then carried out to optimise the preliminary sub-surfaces to ensure the machinability. For each sub-surface, gouging-/collision-free tool orientations are first calculated and then the optimal combination of the fixed tool orientation and the feed direction is determined by maximising the average machining strip width for toolpath generation. The proposed method is tested by a case surface and the comparisons to some other traditional methods are also provided.

Microwave curing technologies have many advantages in manufacturing fiber reinforced polymer composite materials used in aerospace products, compared with traditional autoclave curing technologies. However, the uneven electromagnetic field of microwave in the cavity of the curing chamber results in uneven temperature on the surface of composite laminates during curing, which has been a major obstacle in industrial applications worldwide. Existing methods attempted to solve the problem by the random superposition of uneven electromagnetic fields, but the results were still not satisfactory to meet the high quality requirements of aerospace parts. This paper reveals the one-to-one correspondence between heating patterns of composite parts and microwave curing system settings, and reports a new concept to solve this problem by continuously monitoring and compensating the uneven temperature distribution in real-time. Experimental results from both fiber optical fluorescence sensors and infrared thermal imagers showed significant improvement in temperature uniformity compared with existing methods.

Cutting tool state recognition plays an important role in ensuring the quality and efficiency of NC machining of complex structural parts, and it is quite especial and challengeable for complex structural parts with single-piece or small-batch production. In order to address this issue, this paper presents a real-time recognition approach of cutting tool state based on machining features. The sensitive parameters of monitored cutting force signals for different machining features are automatically extracted, and are associated with machining features in real time. A K-Means clustering algorithm is used to automatically classify the cutting tool states based on machining features, where the sensitive parameters of the monitoring signals together with the geometric and process information of machining features are used to construct the input vector of the K-Means clustering model. The experiment results show that the accuracy of the approach is above 95% and the approach can solve the real-time recognition of cutting tool states for complex structural parts with single-piece and small-batch production.

Adaptive machining with responsive fixtures is an effective way to release and eliminate deformation during the machining process of large-scale structural parts. While a proper locating principle is a key challenge for the responsive fixture based machining due to its two contradictory requirements of maintaining the machining datum and deformation releasing sufficiently. For traditional machining methods, locators are placed as far as possible from each other to maintain workpiece stability, which would significantly limit the release of workpiece deformation. In order to address the issue mentioned above, different from traditional locating method, a novel 6+X locating principle is proposed in this paper, where the workpiece is divided into a fixed region and a floating region to facilitate the new locating principle. The fixed region is optimized by considering dynamic mass centers with a Genetic Algorithm to maintain workpiece stability, and the 6° of freedom of the workpiece are restrained by the fixed region to keep machining datum. The floating region is supported by X floating units so as to keep the workpiece stable during machining process, while the floating units can adjust the workpiece posture in accordance with the machining deformation during the machining intervals. Algorithm experiments are applied to verify the convergence of the proposed optimization model. Finally, a case study with real machining experiment is carried out to verify the proposed method.

Part deformation prediction and control is a crucial issue for obtaining tight dimensional accuracy so as to ensure product quality with high performance, and deformation prediction is the fundamental of the deformation control. However, existing machining deformation prediction methods are based on the prediction or measurement of residual stress and suffering from two challenges: (i) the measurement accuracy of residual stress field is limited by physical principle and (ii) low prediction in accuracy. In order to address these issues, this paper presents a method for predicting part machining deformation based on deformation force using the proposed Physics-informed Latent Variable Model involved physics knowledge. Deformation force is introduced to represent the inner unbalanced residual stress state of the workpiece, and it is a much easier and more accurate signal compared with residual stress. Machining deformation is predicted by fusing the data-driven method and the prior knowledge of deformation mechanical relationship by taking advantage of the latent variable. The proposed method was verified both in simulation and actual machining environment, and accurate machining deformation prediction has been achieved. The proposed method can be readily extended to the prediction problems involved with difficult-to-measure physical quantities.

Microwave cuing technology is a promising alternative to conventional autoclave curing technology in high efficient and energy saving processing of polymer composites. Dielectric properties of composites are key parameters related to the energy conversion efficiency during the microwave curing process. However, existing methods of dielectric measurement cannot be applied to the microwave curing process. This paper presented an offline test method to solve this problem. Firstly, a kinetics model of the polymer composites under microwave curing was established based on differential scanning calorimetry to describe the whole curing process. Then several specially designed samples of different feature cure degrees were prepared and used to reflect the dielectric properties of the composite during microwave curing. It was demonstrated to be a feasible plan for both test accuracy and efficiency through extensive experimental research. Based on this method, the anisotropic complex permittivity of a carbon fiber/epoxy composite during microwave curing was accurately determined. Statistical results indicated that both the dielectric constant and dielectric loss of the composite increased at the initial curing stage, peaked at the maximum reaction rate point and decreased finally during the microwave curing process. Corresponding mechanism has also been systematically investigated in this work.

Continuous Fiber Reinforced Thermoplastic (CFRTP) lattice truss core structures are potential to aerospace engineering because of their high weight reduction efficiency and multifunctional application. A novel free-hanging 3D printing method was firstly presented by the authors to manufacture CFRTP lattice truss core structures. The free-hanging printing path generation strategy was established to realize the undercut and overhanging truss structure without any support. Based on the free-hanging 3D printing method, the typical lattice topologies and more complex structures like the integral variable-thickness wing were printed. The cross sectional morphology of samples printed by using the free-hanging printing method were analyzed. The investigations on the parameters such as relative density, fiber volume content and truss angle were conducted to reveal the relationship between the process parameters and out-of-plane compressive properties. The experiment results demonstrated that the average structural error was about 1.89% compared with the theoretical geometry. The compression strength of the printed sample was 224% higher than that of the pure thermoplastic resin counterpart. The specific compressive strength of free-hanging printed samples was in competitive status compared with currently existed CFRTP lattice structures in available literatures. Keywords: Free-hanging 3D printing, Continuous carbon fiber, Lattice truss core, Thermoplastic composites

Deep transfer learning (DTL) has received increasing attention in smart manufacturing, whereas most current studies focus on the situation of marginal distribution shift in classification. We observe a new regression scenario in machine health monitoring systems (MHMS) with conditional distribution discrepancy across domains and try to propose a general theoretical approach for broader applications. In this paper, we propose a DTL framework CDAR, namely conditional distribution deep adaptation in regression. As only few labeled target data is available, in addition to only considering the prediction accuracy of individual samples, CDAR aims to preserve the global properties of the conditional distribution dominated by the target data. Thus, a hybrid loss function is constructed by combining the mean square error (MSE) and conditional embedding operator discrepancy (CEOD) in CDAR, and the target model is able to be finetuned by minimizing the designed loss function through back-propagation. The performance of the proposed CDAR is compared with two classical marginal distribution adaptation algorithms, TCA and DAN, and a specific method of DTL, FA. Experiments are carried out on two real-world datasets and the results verify the effectiveness of our method.

The accurate prediction of chatter stability in milling operations is a continuous pursuit of manufacturing engineering. Physical-based stability analysis methods suffer from inaccurate model parameters, while data-driven methods lack generalisability and physical interpretability. This study proposes a physics-informed Bayesian inference framework for milling stability analysis. The framework leverages experimental data to infer the distribution of model parameters for probabilistic stability lobe diagram (SLD) computation, thus maintaining the generalisability and interpretability of the physical model in a hybrid-driven manner. By defining a novel likelihood function based on the Floquet theory, the underlying connection between the model parameters and experimental cutting data is established. The uncertainties and variations of the model parameters can then be represented by the inferred probability density function, which can be used to generate a reliable probabilistic SLD. The experiments indicate that the proposed method has significant potential for improving the accuracy of milling stability prediction.

In a typical multi-pass machining process of freeform surfaces, especially at the roughing stage, the constant Z level strategy is typically used. Under this strategy, the inevitably formed islands where the tool has to travel between increase the machining time as the tool has to air-cut between them. Moreover, the uneven intermediate geometry left on the islands leads to sharp changing of chip load and jeopardizes the finish surface quality. To avoid the staircase-like intermediate geometry of the islands and most of all for the ultimate goal of maximizing the machining efficiency, in this paper we present a method of variable-depth of cut multi-pass tool path generation for multi-axis machining of mesh surfaces. By parameterizing the cut volume to a unit uvw cubic parametric domain, the multi-pass process planning task turns to finding an optimal sequence of w level parameters. To solve the optimization problem, we introduce a machining efficiency indicator—the average material removal rate (AMRR). The w level parameters are obtained by iteratively solving the maximization problem regarding AMRR. The preliminary computer simulation and real physical cutting experiments show that substantial savings in total machining time could be achieved by the proposed method as compared to the traditional constant Z level method.

Based on a cyclic heating-cooling treatment method, a temperature-independent silica optical fiber evanescent wave sensor is proposed. The cyclic heating-cooling treatment process could significantly release the residual stresses in the taper, achieving a high measurement accuracy in the temperature range of 20–200 °C. After the treatment, the relative light intensity 1 (RLI) variation was 0.03 dB in the temperature range of 20–200 °C. The RLI variation of the tapered silica fiber-optic evanescent wave 2 (TSFEW) sensor was reduced by 86% compared with the untreated TSFEW sensor. Two kinds of typical treatment processes were studied by contrast experiments to further explore effective treatment process of temperature-independent TSFEW sensors. The effectiveness of the temperature-independent TSFEW sensor was validated using sucrose solutions at different temperatures. The mechanism of cyclic heating-cooling treatment method for eliminating the effect of temperature on TSFEW sensors was analyzed.

The blank of large complex surface parts such as aircraft skin parts is formed by roll–bending process, which always introduces a big shape deviation between actual blanks and the desired ones. The cutting tool path for the blank machining planned based on the nominal digital model of the part may scrap the part due to the shape deviation. In order to address this issue, laser inspection method is used to inspect the blank before machining to obtain the actual shape of the blank which can be used as a reference for cutting tool path generation in addition to the nominal digital model of the part. While the laser inspection device posture adjustment is an important factor for inspection precision, it should be intensively focused. The difficulty of the laser inspection device posture adjustment optimization is that it can only be performed based on the nominal digital model of the part, while the real inspection object is the actual blank whose real shape is unknown, which may lead to the inspection precision disqualification. In order to address the above difficulty, the actual blank profile is first constructed with a few of laser inspection points obtained by rough inspecting. An inspection path planning method is proposed based on the actual blank profile. A genetic algorithm is used to optimize the pendulum angles of the laser inspection device based on the inspection path and the actual blank profile, and the optimized postures of the actual blank can be calculated based on the pendulum angles. The proposed method of this article is verified by the inspection of an aircraft skin part, and the experiment results show that this method can ensure the final inspection precision.

The tool-part interaction of composite materials during microwave curing is different from conventional thermal heating, mainly because of the rapid and selective heating mechanism of microwave. Novel measurement and analysis approaches of microwave curing induced tool-part shear stresses were established and verified in this article. The optical fiber sensors were placed in the groove of tool surface to measure the strain values, which were converted to the tool-part shear stresses. A comprehensive friction coefficient was calculated by the shear stresses, and it was presented by considering the thermal expansion mismatch of composite/tool and the chemical shrinkage of resin. In order to validate the accuracy of model, the shape dimension of microwave cured scaling down leading edge of the composite part was measured to compare with simulation model. The results exhibited that the presented model can significantly decrease the prediction error of composite deformation compared with other methods during microwave curing.

The continuous carbon fiber reinforced polylactic acid composite was manufactured by the rapid prototyping approach of three-dimensional (3D) printing. In order to analyze the improvement of this new method, the comparison experiments of printed samples with or without preprocessed carbon fiber bundle were performed. The mechanical strength and thermodynamic properties were measured by using the electronic testing machine and Dynamic Mechanical Analyzer (DMA). The novel nozzle and path control methods were designed to satisfy the demands of continuous carbon fiber printing. Through the experiment and analysis, the preprocessed carbon fiber with polylactic acid sizing agent could effectively increase the interfacial strength between carbon fiber and resin. The experimental results demonstrated that the tensile strength and flexural strengths of modified carbon fiber reinforced composites were 13.8% and 164% higher than original carbon fiber reinforced samples. The Scanning Electron Microscope (SEM) scan results indicated that the preferable bonding interfaces were achieved of modified carbon fiber reinforced composite.

During the machining process of large-scaled and thin-walled parts such as aircraft structural parts, the deformation is a relatively common phenomenon which seriously affects the machining quality of the parts and may lead the parts to be scrapped. In this paper, interim machining states of workpiece are considered in addition to final machining states for deformation control so as to improve the machining quality and part correct rate. It is very important for large-scaled and thin-walled parts to consider the interim machining states, as considerable deformation has always occurred in interim machining states. The difficulties of deformation control of interim machining states contain two aspects: (1) how to assess whether interim machining states can satisfy process requirements for further machining and (2) how to adjust the tool paths adaptively for further machining so as to make the final machining states correct. In order to address the above difficulties, an adaptive machining approach based on in-process inspection of interim machining states for large-scaled and thin-walled complex aircraft structural parts is proposed in this paper. The actual interim machining state is obtained based on in-process inspection of machining states during the machining process; the essential idea of this paper is that the final machining state of the workpiece can be guaranteed by adjusting the tool path based on the inspection of interim machining states, which is realized by coordinating the dimensional tolerance and geometrical tolerance. In order to realize the new idea, the criterion for determining whether the interim machining state is suitable or not for further machining and the concept of expected final state are introduced. Eventually, the large-scaled and thin-walled complex aircraft structural parts can be machined adaptively according to process requirements. A typical large-scaled and thin-walled complex aircraft structural part is used as a case to validate the proposed approach. The results show that the dimensional error is 0.10 mm and the profile error is 0.06 mm, which can meet the machining requirement of large-scaled and thin-walled complex parts.

Carbon fiber-reinforced polymer composites have been widely used in the aerospace industry. However, they are extremely sensitive to crack initiation, propagation and interlaminar delamination which severely reduce their service life. This paper demonstrated that the Mode-I interlaminar fracture toughness could be significantly improved in carbon fiber/bismaleimide composites using a microwave curing process. An increase of about 53.5% in critical load and an increase of approximately 133.5% and 61.2% in fracture toughness and fracture resistance have been achieved, respectively. The microwave manufacturing cycle for composites was cut to 44% of the thermal processing cycle. Dynamic mechanical thermal analysis was performed to investigate the enhanced interfacial strength in microwave-cured composites. The improvement in fracture toughness was attributed to a better interfacial adhesion between resin and fiber, which was investigated by the observation of fracture surfaces with optical microscopes.