Search

In order to improve the detection accuracy of the navigation line by the unmanned automatic marking vehicle (UAMV) in the complex construction environment. Solve the problem of unqualified road markings drawn by the UAMV due to inaccurate detection during construction. A navigation line detection algorithm based on and improved YOLOv4 and improved Haar-like feature named YOLOv4-HR is proposed in this paper. Firstly, an image enhancement algorithm based on improved Haar-like features is proposed. It is used to enhance the images of the training set, make the images contain more semantic information, which improves the generalization ability of the network; Secondly, a multi-scale feature extraction network is added to the YOLOv4 network, which made model has a stronger learning ability for details and improves the accuracy of detection. Finally, a verification experiment is carried out on the self-built data set. The experimental results show that, compared with the original YOLOv4 network, the method proposed in this paper improves the AP value by 14.3% and the recall by 11.89%. The influence of factors such as the environment on the detection effect of the navigation line is reduced, and the effect of the navigation line detection in the visual navigation of the UAMV is effectively improved.

For the problem of simplifying pattern-based modeling procedures, an improved pattern-based modeling method is put forward via pattern classification for a class of complex processes. It is a pure data-driven modeling method using statistical attributes of the processes. At the beginning of the paper, a method of system dynamics description based on pattern moving is introduced. Then, an improved method of pattern-moving-based prediction modeling is put forward, and it simplifies the pattern-moving-based modeling method by integrating an initial model and a classification mapping. It consists of two parts: system pattern construction and pattern classification. And a constructive classification neural network (CCNN) is designed to describe system dynamics by classifying the system pattern, and its generalization is discussed. Finally, simulations using data of an actual production process demonstrate the feasibility of the proposed modeling method, and the effectiveness of the CCNN is verified using comparison experiments.

Due to the existence of a kind of industrial production process which is too complex to determine the model structure and identify the system parameters, a data-driven control method is proposed based on pattern moving. First of all, the way of system dynamics description based on pattern moving is introduced. Then, according to the mentioned way a pattern-moving-based model free adaptive control (PMFAC) algorithm is put forward, and pseudo orders of the PMFAC algorithm are identified by using the conditional entropy of output classes. At last, simulation results demonstrate the feasibility and effectiveness of proposed data-driven control method.

With the development of high-speed roads and increase of lighting, ventilation and monitoring devices, it is necessary to study a real-time and robust voltage stability monitoring method. In this paper, a new method of voltage monitoring is proposed, and this method relies on voltage dynamical entropy and its deviation. At the beginning of the paper, the voltage is symbolized by using a maximum-entropy-based clustering algorithm (MECA), and the MECA is put forward in this paper. Then, a dynamical pattern is constructed, and voltage dynamical entropy is calculated of a short-term signals. The entropy is used to describe the voltage fluctuation and voltage dynamical properties. Subsequently, the entropy and a voltage deviation consist of a voltage feature. And a fuzzy classifier is constructed and used for monitoring a voltage stability status. At last, a simulation is given to demonstrate feasibility and effectiveness of the proposed method.

In order to improve the harsh working environment such as high temperature and soot faced by the operators of hot in-place recycling heating vehicle, and to increase the intelligent level of heater control, an intelligent driving system composed of a differential GPS positioning system, a path planning system, a vehicle control system and a management center was designed It uses coordinate transformation and the least square algorithm to achieve high-precision path fitting. The Newton method is used to calculate the deviation from the point to the fitting curve, and a double closed-loop PID controller is used to achieve accurate path tracking. The field test results show that the vehicle can be controlled by intelligent driving system to walk along the pre-construction route, and the lateral control accuracy is within ±30mm, which meets the requirements of the hot in–place recycling construction process.

For a class of production processes, a modeling method based on pattern recognition is introduced, and state feedback stabilization of the system is studied in a pattern-moving state space. At the first part of the article, the modeling method of the system is introduced, which is based on pattern moving, and the system dynamics are presented by using the so-named pattern class variable(PCV) instead of a system output or state variable. A discrete-time and discrete-state nonlinear state-space model (DNSM) is given in Euclidean space based on the modeling method. And soon a new control and stability analysis method is put forward. Finally, validity of the stabilization method is demonstrated by using data of an actual production process.

A new type of modeling method is put forward based on pattern recognition (PR) technology for some industrial production processes. The proposed method is a pure data-driven modeling method since the model is independent of the controlled plant, and it is based on the measured input and output (I/O) data of the controlled plant in a closed loop. Different from the traditional modeling method, the system dynamics is described by I/O classes, which are obtained from raw I/O data through partitioning of the data space respectively and I/O orders of the model resort to the conditional entropy. The covering algorithm based pattern classification (PC) is used to establish the mapping between input and output of the proposed model in metric spaces. The experimental results illustrate the feasibility of the modeling method.

This paper proposes a control scheme for a kind of complex systems whose dynamics follow statistical law. First, inspired by data mining technology, the samples are clustered into several classes reflecting the working pattern by modified ISODATA method. Second, based on the clustering result, the cell state space is constructed within global and local target cells respectively by Cartesian product, and the cell mapping further is created via the analysis of pattern time series. Third, by the reverse searching algorithm, global optimal paths to the target cell are found and the information provided by the searching paths constitute the optimal controller at a coarse-grained level so as to bring the system into the target cell quickly, once enter into the target cell, another controller which consists of the subcell mapping with the optimal cost increment in refined granularity level is applied in order to reduce steady-state error. Finally, a simulation in the dynamic trajectory of a sintering process is given to demonstrate the feasibility of the proposed approach.

For a kind of complex production processes, a new control method is proposed based on pattern recognition technology. It avoids structure design or parameter identification of a system model or controller. First of all, the moving trajectory is predicted through the classification of system pattern samples constructed by input-output classes. And then it is approximately described by a straight-line model. According to the model the system input of a class is obtained. At last, experimental simulations demonstrate the validity of the proposed control method and analysis of parameters affecting control performance is presented.

Aiming at the characteristics of non-linear, time-varying and wide-ranging working conditions of heating furnace, with the improvement of control requirements for actual system prediction, single model modeling has the problems of large amount of calculation and poor accuracy. A multi-model modeling method is proposed in this paper. This method first divides the actual data of the heating furnace system into training set, validation set and test set, and uses FCM clustering to divide the training set into different working conditions; The Elastic Network (ElasticNet) and support Vector Machine regression (SVR) models are established in each local condition, and the optimal model of each local condition is selected from the two models by the validation set; use genetic algorithm (GA) to obtain the optimal weight of each local model, finally construct a model suitable for the global. This modeling method has a good global adaptability to the identification process. The veracity of the model is verified on the test set, and good results are obtained.

This paper proposes a new type of modeling framework for complex systems. This kind of systems has the characteristics: it is difficult to establish the mechanism model and its dynamics obey the statistical law. Firstly, in order to extract the statistical attributes, an iterative self-organizing data analysis techniques algorithm (ISODATA) of data mining is used to separate process data into a number of classes. Based on these classes, a pattern class space in which each class is taken as an element is established. Next, to describe the system dynamic with statistical law, a pattern class variable is defined in the pattern class space. To further solve the incalculable problem of the pattern class variable, a cell space is introduced where each pattern class variable is measured by cell centers. Based on this measure, an improved controlled autoregressive moving average (CARMA) model described by the pattern class variable is obtained, and model parameters are identified by an improved quantum-behaved particle swarm optimization (IQPSO) algorithm. The attractive of the algorithm lies in enhancing the ability of the global development and local exploration. Based on the model, the corresponding cell mapping model is derived by the state space representation, which avoid the expansion of categories. Then, the global performance of the pattern class space is analyzed via cell mappings. Finally, a simulation in the dynamic trajectory of a sintering process is given to demonstrate the feasibility of the proposed approach.

In the production process, the complex conditions which have a lot of disturbance, noise and random fluctuations, greatly affected the accuracy of modeling. To slove this problem, a combined forecasting model based on empirical modal decomposition(EMD), extreme gradient boosting (XGBOOST), regularized least squares estimation (RLSE) is proposed. In this paper, the actual data of the sintering process is taken as an example to illustrate the construction process of the model. Firstly, the intrinsic mode functions(IMFs) of bellows temperature is obtained by empirical mode decomposition(EMD). Then, the XGBOOST model and RLSE model of each IMF are respectively constructed on the train set. On the basis of this, the combination of IMF models is determined by RMSE of the IMFs on validation set, and ultimately add them together to build a combination model. Finally, through the evaluation on the test set, the prediction accuracy of the combination model is verified.

For many industrial production processes obeying certain statistical laws, a new method of establishing a SISO control model is put forward based on pattern recognition technology. Firstly, k-means clustering algorithm is used to partition input and output data collected into several classes respectively. Secondly, the distance between two classes is described by the distance of the two class centers. Then the mapping relationship between input and output of the control model is established based on pattern classification technology. And input and output orders of the model are identified by conditional entropy. Lastly experimental results illustrate the feasibility of the modeling method using the data collected from an actual industrial process.

For the limitation of current license plate location algorithms, present an algorithm of the license plate location based on image energy and its practical formulas. Introduce the algorithm of extraction and methods to verify region of license plate. It takes full advantage of characteristics of the license plate, such as complicated textures, sharp-cut contrast and so on. It also has the advantages to avoid the difficulty to choose a threshold and the influence of colors changes due to light. A large number of experiments demonstrate that the algorithm is of higher accuracy and of fast speed.

License plate recognition (LPR) is the hot research topic in Intelligent Transportation Systems. This paper focuses on license plate location which is based on image energy and license plate character recognition based on pattern matching, and then gives experiment results of license plate recognition system. And experiment results verify the feasibility of the LPR system.

This article presents a new method based on CCD images to measure the speed of vehicles. The working principle is taking twice instantaneous tandem photographs.Then based on the location of the vehicle by calculating image energy and CCD, with reference to the advance calibration rectangle, we can obtain speed of vehicle. Present a method to locate vehicle which is based on image energy and a new formula for calculating the vehicle speed. Simulations verify the effectiveness of the given method.

This paper proposes the definition of the tendency curve with saturation line and gives the range of the parameters and the necessary and sufficient condition when the cubic polynomials satisfy the definitions, which is based on the method of tendency surface constructed by cubic polynomial (TSCP) proposed in our former paper. The simulation experiment results, compared with the general fuzzy control, are given, which verify the effectiveness of the given method.

A new control method of tendency surface based on cubic polynomial (TSCP) is proposed; the definition of the tendency of control output is given; the cubic polynomial correlation coefficients are determined to meet the requirements of tendency of control output. At the same time, the procedure to expand the univariate polynomial curves into surface is suggested. Further, tendency surface is optimized by genetic algorithm. Simulation results verify the effectiveness of the given method.

A new control method by employing optimized tendency surface based on cubic polynomial (TSCP) is proposed herein. The definition of termed control output tendency is given; and the cubic polynomial correlation coefficients are determined to meet the requirements of the tendency. At the same time, the procedure to expand the univariate polynomial curves into a three dimensional surface is suggested. Furthermore, the optimization process is utilized as automatic construction procedure of the tendency surface. The effectiveness of the proposed method has been verified by simulation results of a fuzzy controller.

This article presents a new method based on video images to measure the speed of vehicles. The working principle is taking two instantaneous tandem photographs. Then based on the location of the vehicle and CCD, with reference to the advance calibration lines, we can obtain speed of vehicle. Further, this paper analyses two measurement errors, and discusses how to choose the picture filming location. Simulations verify the effectiveness of the given method.