Search

A new approach using Dirichlet process mixture model (DPMM) and particle filtering (PF) method to predict remaining discharge time (RDT) of ongoing discharge processes of lithium-ion batteries is proposed. Different voltage trajectory patterns are proposed to describe the discharge process at different periods of a battery’s life. Each pattern is represented by the same empirical model based on the physical discharge behavior of lithium-ion batteries, with different parameters to distinguish itself. A DPMM is developed to automatically discover these voltage trajectory patterns from historical monitoring data, without specifying the number of patterns in advance. The trajectory parameters for each pattern can also be learned simultaneously, which are used as the initial parameters for online prognostics. The developed DPMM is able to discover new patterns as more trajectory data become available. During online prognostics, voltage trajectory pattern is constantly identified with new voltage data, then initial parameters for identified pattern and PF-based method are combined to predict the RDT of the ongoing discharge process. A case study demonstrating the proposed approach is presented. It also demonstrates that this approach improves accuracy of RDT prediction compared with benchmark PF-based method.

In order to improve the testing ability of traditional dedicated test system, this paper proposes and designs a distributed test system with the cross-platform, generalization, and publish / subscribe middleware. The distributed test system is developed by B/S architecture, which consists of the system website deployed on the test master server and a number of test service nodes that perform data interaction through Data Distribution Services (DDS) middleware. The test resources on each test service node can be fully used by the test system to maximize the utilization rate of test resources. This test system completes the testing tasks and improves the testing efficiency in the test application of Distributed Integrated Modular Avionics (DIMA) system.

Aircrafts, especially the Unmanned Aerial Vehicle (UAV) are always operating in environments filled with expectations and unforeseen occasions. In addition, hardware and software failures caused by design defects are inevitable [1]. In order to avoid catastrophic failure, we must locate the failure immediately in working condition. That means real-time diagnosis. Bayesian inference has been well established in the area of diagnosis and health management. In this paper, we come up with a method to take advantage of FPGA to implement Bayesian inference, which has the features of hardware programmable and the ability of parallel computing. Therefore, we compile the Bayesian Network into arithmetic circuit, and then evaluate the arithmetic circuit on FPGA.

A new data acquisition (DAQ) instrument based on LAN eXtensions for Instrumentation (LXI) bus and synthetic instruments has been proposed to deal with tricky testing problems in complex electrical system. This instrument is equipped with multi-core ARM processor and large capacity FPGA to achieve best DAQ and data processing performance. One important feature is multi-functional measuring, it can acquire amounts of data coming from different kinds of signals simultaneously. In addition, the instrument is portable for mobile testing on account of its compact mechanical components and low weight. The software system refers to the concepts of IVI, SCPI and ATML, using a hierarchical and modular design method. Furthermore, the whole system has a strong ability of data processing and is able to realize parallel DAQ on many channels at the same time. And the online reconfiguration make it more possible to meet the requirements coming from different targets, so the equipment is universal and can be applied to a variety of tasks. Practices prove that this instrument works well with easy operation and meets the requirements of design.

An OMAP-based real-time can-label monitoring system has been designed to deal with the recognition problem of cans in a strongly-interfered industrial condition. The dual-core processor OMAP3530 and binocular structure are utilized to guarantee the effective capture and analysis of cans' images. Firstly, the monitoring system captures several images of the cans, then preprocesses these images and outputs the binarized images. Secondly, the binary images are normalized to make sure the images are in the same size. After the image normalization, the monitoring system classifies these images and outputs a number of recognition results by the different features. Lastly, by fusing these results, the system outputs the final recognition result. The experiment results show that the system can resist the high accuracy and robustness as well as overcome the effect of illumination changes and shadows.

Nowadays in industrial processes, whether producers or users think highly of performance reliability and robustness of equipments. Therefore, the FDI (Fault detection and isolation) and maintenance techniques have become hot topics for health management of industrial units, as a safety guarantee indeed. As a consequence, researchers have made great efforts to develop, verify and refine diverse diagnosis techniques, meanwhile compare and screening them in order to apply them properly in practice. And then, NASA Ames has built the Advanced Diagnostics and Prognostics Testbed, a real-world system as a general platform for verification and validation (V&V) of diagnosis techniques. Until now, many researchers have developed effective diagnosis algorithms specially applied to this system. In this paper, we introduce the ADAPT and the diagnosis competition around the system, and we review a variety of diagnosis methods divided mainly in three types, model-based, optimization-based and artificial intelligence-based methods, while elaborating the first type in detail by two sorts of model: physical and graphic, of which the second attracts more and more attention of scientists in actual research. Finally, we make a comparison among them based on simplified metrics of qualification, which plays an important role in choosing appropriate methods for diagnosing a special problem.

A new fault diagnosis method based on multi-signal flow graph (MSFG) has been proposed to deal with online complex system diagnosis. This method has three steps. The first step is to develop a multi-signal graph model for the object system and generate a dependency matrix as diagnostic knowledge. The second step is to generate conflict sets by the dependency matrix and the system observation vector, and then the problem of finding the minimal diagnosis set is transformed into a problem of integer programming. In the final step, the optimal solution of the problem of integer programming is calculated by using a branch-and-bound algorithm for diagnosis. In this way, explosion problem caused by enumeration can be avoided. In order to facilitate the application of MSFG modeling and on-line diagnosis in real problems, an Intelligent Diagnostic Platform (IDP) that provides a MSFG modeling environment and employs the branch-and-bound algorithm for Complex System diagnosis is developed. The structure of the software IDP is presented. Finally, a case study for a fuel system of aircraft is described in detail to illustrate the procedure of using IDP for diagnostic problems and to demonstrate the effectiveness of the proposed method.

The health modeling for aerial system is a difficult problem, especially when the system is composed of identical or almost identical system, the system parameters and sensors measurements are subject to uncertainty, and the health state of system evolves over time. In order to solve this problem, a novel object-oriented modeling approach is proposed. This approach is based on the Bayesian theory and can provide corresponding algorithms for different levels of system in constructing the dynamic Bayesian network of entire system health step by step in a bottom-up fashion. The effectiveness and robustness of the resulting dynamic Bayesian network are verified through experiments on a particular aerial fuel delivery system.

Verification and Validation (V&V) technology has been identified as a key step in quantifying and building confidence before integrated health management (IHM) system deployment to inform about equipment or component's health status and to support operational and maintenance decisions. However, due to generally no rigorously universal methods and no statistically adequate data available, V&V technology remains a challenging open problem. The recent developments appeared in academic journals, conference proceedings and technical reports for V&V technology in IHM field are systematically reviewed in this paper, focusing on V&V definitions and standards, V&V metrics, and V&V methods as well as the future development trends and challenges of V&V technology.

The techniques of remaining useful life (RUL) estimation are playing a more and more important role in aircraft safety and condition based maintenance. This paper gives an overview of RUL prognostic estimation approaches applied to aircraft subsystems or components. Existing RUL estimation approaches are categorized into three types, namely model-based approaches, data-driven approaches and fusion approaches and their characteristics are comprehensively introduced. Moreover, three common and promising methods: particle filtering, neural network and relevant vector machine as well as their advantages and disadvantages are discussed in details. Finally, the future challenges concerned with RUL prediction are also presented.

This paper presents an approach to inference in Bayesian networks, which meets the uncertainty and real time challenges in diagnosis. The process can be described as compiling multi-linear function of a Bayesian network into Arithmetic Circuit for online inference. There are three advantages of this approach. First of all, offline compiling process decreases online inference time. Our method differs from the original approach of compiling Bayesian networks using variable elimination by omitting step of representing Bayesian network with Algebraic Decision Diagrams, which simplifies offline compiling process. Then, local structure reduces Arithmetic Circuits and results in less space consumption and online inference time. Finally, differential approach of inference has high efficiency as it calculates answers to multiple queries simultaneously without repeating compiling. Simplified differential approach of inference for particular situation, when evidences in fixed, decreases the searching time and improves efficiency of online inference.

Capacitive pressure sensor has many advantages, such as high sensitivity, low temperature drifting and low power consumption. In this paper, capacitive pressure sensor is used as pressure sensor. This paper analyzes SiC MEMS capacitive sensor's features and introduces its conditioning circuit. This sensor has widely commercial exploitation because of its small size and outstanding chemical stability as well as its mechanical character for harsh industrial environment application. Capacitance of this sensor varies at fF or pF degree. As a result, exploitation of such high sensitive capacitive pressure sensor is inhibited by parasitic capacitance in conditioning circuit. Integrated conditioning circuit is introduced for those high-sensitivity capacitive pressure sensors. Measures taken to reduce parasitic capacitance are described in the paper. Experience results give the range and linear area of this sensor.

In military and industrial sectors, automatic test system remarkably reduces maintenance cost of certain equipments. As is depicted in the thesis, the highly adaptive automatic test software architecture is presented with multiple common data interfaces, which not only enables rapid prototyping of automatic test procedure, but also collectively facilitate information sharing and data exchange in the whole lifecycle.

Integrated Health Management (IHM) is an advanced technology which integrated artificial intelligence with advanced test and information technologies. Having gone through fault detection, isolation and reconfiguration and immerged with the state of arts reasoning technologies, IHM monitors and controls the function of critical systems and components in order to ensure safe and efficient operation. An IHM system usually comprises seven functional modules, namely data acquisition, signal/feature extraction, condition assessment, diagnostics, prognostics, decision reasoning and human interface. Among them, fault prognostics are not only the core of IHM, but also an important guarantee to reduce the costs of life-cycle maintenance, and to improve system security. Fault prognostics is the process to project the current health state of equipment into the future taking into account estimates of future usage profiles. It may report health status at a future time, or may estimate the remaining useful lifetime (RUL) of a machine given its projected usage profile. In recent years, fault prognostics are under unprecedented attentions. And it is becoming the most challenging research area which is so-called crystal ball of IHM. Based on the theory, methods and routes adopted in the practical application, fault prognostics is generally fallen into three main categories, namely model-based approaches, knowledge-based approaches and data-based approaches. Then, based on the analysis of some typical applications on each approaches, the strengths and weaknesses of each approach are further discussed. Finally, according to the current research situation at home and abroad, the future development trend of fault prognostics is also presented.

The efficiency of diagnosis and the accuracy of diagnosis result have been improved by amalgamating and combining the advantage of fault tree analysis and expert system. The visible modeling of fault tree model has been realized by introducing the technique of Visio components development, which not only makes the user interface more intuitional and friendly, but also corresponds to the specification of fault tree model greatly and makes the checking of the consistency and validity of fault tree model more convenient. The modular and hierarchical thought, which is combined with the information description of object oriented, has been applied in different diagnosis objects, which not only implements the representation of knowledge base for expert system, but also enhances the system's universal character. Finally, the framework of C/S has also been built in this system by the technique of Database, which improves the system's performance of open, maintenance and transplant to a large extent.

The main idea of condition based maintenance (CBM) is to monitor the health of critical machine components and system almost continuously during operation and maintenance actions based on the assessed condition. If done correctly, CBM has the benefits such as reducing catastrophic failures, minimizing maintenance and logistical cost, maximizing system security and availability and improving platform reliability. A CBM system usually has four major functional modules, namely feature extraction, diagnostics, prognostics and decision support. Among them, fault prognostics is the most important enabling technology. It is the most challenging research area which is so called crystal ball of CBM. But it has the potential to be the most beneficial. This paper presents a fault prognostic algorithm based on a generic wavelet neural networks (WNN) architecture. Its training process based on genetic algorithm is described in detail. Finally, the fault prognostic algorithm has been verified using a simulation experiment, and the results are very satisfactory.

Integrated health management (IHM) technology is the emerging paradigm in system supervision and maintenance area, and it is the key to improving platform reliability, security and availability, reducing catastrophic failures and reducing operational and logistical cost. Mainly through as early as possible reporting incipient faults, predicting the remaining useful life of fault components, identifying diagnosis across subsystems accurately and making maintenance decision advisably, IHM system can achieve the goals mentioned above. Those functions, which can be formulated as inference problems, are important aspects of the IHM program. Therefore, it is very important to research the reasoning techniques to promote the IHM technology improvement. A review about reasoning techniques implementing IHM is conducted with the emphasis placed on the layered architecture (supporting both distributed and hierarchical reasoning, information fusion as well as interoperability and interchangeability of reasoning algorithms) and reasoning approaches for diagnosis, prognostics and maintenance decision-making, and the future challenges concerned with reasoning techniques are also presented.