Summary of FMS fault diagnosis technology in the h

Summary of research on Fault Diagnosis Technology of flexible manufacturing system (FMS)

Abstract: analyze the characteristics of FMS and the difficulty of fault diagnosis. On the basis of a large number of documents, this paper summarizes the main research contents, development status and research results in this field at home and abroad, and points out the characteristics of the current research work and the main problems existing in the existing research. The future development is prospected. It is pointed out that the research and development direction of FMS diagnosis system is the integrated intelligent decision system with integrated diagnosis, multi-sensor information fusion, multi-method comprehensive application, and networked remote diagnosis service

key words: flexible manufacturing system; Fault diagnosis; Intelligent diagnosis system

the market environment determines the production mode of enterprises. Manufacturing enterprises need to meet the product needs of different customers and the requirements of social sustainable development with the fastest listing speed, the best quality, the lowest cost, the best service and the cleanest environment. Driven by this goal, a variety of advanced manufacturing technologies (AMT) have been proposed, and have been focused on research and development. Flexible manufacturing system (FMS) is the product of the development of ATM. It has been widely studied and widely used in manufacturing enterprises. FMS usually includes several numerical control equipment, central tool magazine, material transportation device, computer control system and other sub equipment or sub-systems. The control network organically combines multiple equipment, so that each equipment can be uniformly scheduled, coordinated and jointly complete the production and processing tasks, and can be flexibly adjusted according to the changes of manufacturing tasks or production environment. This flexibility refers to the flexibility of the system. Flexibility is the biggest feature of FMS. It has the characteristics of good flexibility, high productivity, and adaptability to medium and small batch production

1 FMS characteristics and the difficulty of fault diagnosis

flexible manufacturing system (FMS) as a kind of complex electromechanical system, its complexity, behavior state and working environment are very different from the traditional manufacturing system. The more obvious is:

(1) FMS is the organic integration of multiple heterogeneous systems in function and structure, which belongs to a complex large-scale system

(2) the system emphasizes high automation and requires a high degree of intelligence

(3) compared with the automatic production line, the dynamic behavior of the system is more complex, the rigid control is weakened, and the flexibility is more obvious

(4) the system has fault-tolerant control. When a certain sub equipment or subsystem fails, the operation process control (i.e. scheduling) of the system can be reconstructed to ensure the integrity of the overall function of the system

(5) the behavior or failure of a single device or system is no longer limited to its own scope, which often affects related and connected devices or systems in function or region

fms system has the above characteristics, resulting in its fault diagnosis not only has the characteristics of general equipment diagnosis, but also more complex and special performance

(1) the high flexibility of FMS inevitably requires a high degree of flexibility within the system and the diversity of operation modes. The negative impact is to increase the uncertain factors of the system and the high possibility of failure in the process of mode conversion

(2) the system equipment is diverse and complex, the processing aims at flexible multitasking, and the processing types, processes and working conditions are diverse. Therefore, it is difficult to comprehensively collect a priori samples and mode samples of various normal and abnormal states, that is, it is difficult to obtain diagnostic knowledge

(3) the discontinuity, paroxysm, fuzziness, relevance and timeliness of process status and faults are more obvious, which makes it more difficult to obtain fault symptom information and equipment status information, and it is more difficult to locate faults quickly

(4) the dynamic linkage and discreteness between the components of the processing equipment make the propagation of faults and the dispersion of fault sources more obvious

(5) the influence of random interference factors such as workpiece size and even misoperation is increased, which makes the possibility of misdiagnosis and missed diagnosis of the diagnostic system greater, and the accuracy of diagnostic reasoning and the reliability of conclusions are reduced

(6) the amount of information in the processing process is large and complex, and the information resources suitable for monitoring, diagnosis and early warning need to be mined, which poses a challenge to the monitoring strategy, fault feature extraction, diagnosis knowledge base management and other links

(7) during the operation of FMS, it is often lack of human on-site monitoring, so it is difficult to detect faults early; Instantaneous information of on-site faults, especially sensory information, is often unable to be captured, and such information is extremely valuable for rapid fault location (reasoning)

from the perspective of practical application, the increase in the complexity of the diagnostic object may lead to the geometric exponential increase in the complexity of the diagnostic system. From the practical application of many FMS research and application units, FMS operation faults occur frequently, and the existing diagnosis system is difficult to cope with the requirements of a variety of complex fault rapid location

2 current research content and development status

as one of the key and bottleneck technologies of FMS theoretical research and practical application, fault diagnosis technology has been the focus of research in the manufacturing field at home and abroad, and has achieved some research results. Analyze and summarize many innovative research results in different research directions, and summarize and classify them to form the basic direction of FMS diagnostic technology research as shown in Figure 1. It can be clearly seen that around FMS, an automated manufacturing system with complex structure and composition, the research on diagnosis technology is mainly carried out in the following four directions:

(1) research on diagnosis system architecture

(2) research on intelligent diagnosis methods

(3) research on FMS fault mechanism and fault model

(4) research on system integration technology

based on the above four major research directions, many studies started from different focuses, and finally formed more detailed research branches. On the whole, the research on FMS diagnosis technology is divergent, and the technology related to each link of the diagnosis process is deepened step by step

Figure 1 research direction and classification of FMS diagnosis technology

2.1 diagnosis system architecture

according to the characteristics of FMS, the current diagnosis system architecture design mainly has two forms: centralized and distributed. On the basis of the two basic methods, in order to give consideration to the requirements of real-time diagnosis and precision diagnosis, the system has a modular structure combining real-time diagnosis and off-line precision diagnosis

Huazhong University of technology has carried out pioneering research in diagnosis methodology and architecture, and further research on FMS fault diagnosis of Zhengzhou Textile Machinery Factory; Beijing University of technology takes Changchun bq-fms as the research object. The diagnosis system adopts the combination of simple real-time diagnosis and offline precision diagnosis. The system has been applied to field operation, but its overall architecture is still centralized

Internet based remote fault diagnosis technology is the latest development trend of complex equipment fault diagnosis. Some research institutions such as Stanford University in the United States and National University in Singapore have established open remote diagnosis and support centers, forming a dynamic agile diagnosis channel for multi-user and multi device between equipment users, research institutions (domain experts) and equipment manufacturers, realizing the sharing of diagnosis resources, The diagnostic efficiency, success rate and reliability of diagnostic results are greatly improved. Xi'an Jiaotong University, Shanghai Jiaotong University and Northwestern Polytechnic University have successively established remote fault diagnosis service centers, which has taken a gratifying step in the remote networking of large-scale complex equipment diagnosis. At present, Huazhong University of science and technology has also carried out research on distributed remote collaborative diagnosis, and has established a prototype test system with certain functions

The theory and method of agent and MAS (multi-agent system) is one of the most revolutionary achievements of computer software engineering. The application of MAS theory to fault diagnosis hopes to solve two problems, one of which is to establish a distributed diagnosis architecture from the perspective of distributed problem solving. Using multi-agent system to build a software system with flexible configuration, high flexibility and good expansibility has great advantages. Maria Athina and other scholars analyzed the details of the application of intelligent agent technology in distributed device fault management, and gave a simple system design method. In the research of FMS real-time monitoring, German researchers used multi agent mechanism to solve the distributed problem of monitoring, and gave the monitoring agent model and function packaging. The University of Manchester, UK, has studied and designed an integrated fault diagnosis system based on multi agent for typical FMS systems. At present, the research is gradually deepening. The Intelligent Engineering Laboratory of Edmonton University in Canada has proposed an integrated distributed intelligent system structure for fault monitoring and diagnosis of complex chemical equipment. Under the integrated framework based on MAS, it effectively realizes the comprehensive utilization of a variety of diagnostic tools, and can easily integrate the original tools and systems. Osla also applies the multi-agent method in the monitoring of power supply system. The multi-layer model and multi diagnosis algorithm (software) agent collaborative solution method proposed by osla are worthy of reference and adoption. The commercial fault diagnosis software compass developed by the man-machine system research center of Berlin University of technology in Germany is completely based on multi agent architecture and has good openness. Users can easily expand functions and knowledge through API interface to realize fault diagnosis of specific equipment. Researchers from the National University of Singapore proposed and constructed the architecture of Remote Fault Diagnosis System Based on multi agent, gave the self-learning method of the system, and carried out two case tests in Java environment. Lubaochun et al. Built a distributed multi-agent diagnosis system structure for manufacturing process monitoring, and studied the multi-agent fuzzy correlation model and the diagnosis and decision-making problems based on it; Aiming at the multi-agent fault diagnosis prototype system, Tsinghua University focuses on the distributed task decomposition and control strategy of equipment diagnosis based on multi-agent theory and the coordination and cooperation mechanism between agents, and puts forward the serial and parallel and hybrid control strategies of diagnosis tasks

2.2 intelligent diagnosis methods

at present, developing intelligent diagnosis is a research hotspot in the field of diagnosis, and there are many corresponding achievements. Figure 2 summarizes the application of intelligent diagnosis methods. Generally speaking, FMS fault diagnosis technology mainly focuses on intelligent diagnosis, especially expert system (ES), artificial neural net (ANN) and their combination with fuzzy theory. The research and application in this field are the most common. Lacic et al. Of the international advanced technology center developed the exmax expert system model to realize the fault diagnosis and maintenance of FMS mechanical system. Beijing University of Aeronautics and Astronautics, in cooperation with Beijing Institute of Aeronautical Technology and other units, designed and built the Beijing flexible manufacturing system experimental center, and preliminarily studied and applied the applicable diagnostic expert system. However, the shortcomings of expert system, such as the "bottleneck" of knowledge acquisition, the "combination explosion" of rules, the low efficiency of reasoning process, and the strong dependence on machine system, limit its wider and more perfect application. Fault diagnosis is still a pattern recognition problem fundamentally. People have successfully applied neural networks to solve many problems such as real-time condition monitoring, fault classification, fault prediction and so on

Figure 2 Application of intelligent diagnosis method

from a large number of applications, Ann is only a tool for information soft processing, and has obvious advantages in local problem processing, but from all aspects