Research On Intelligent Fault Diagnosis And Maintenance System Based On Service-Oriented Architecture And Its Application To Rotating Machinery

All kinds of equipment of process industry are becoming more and more large-scaled, high-speed, automated and intelligent, especially such significant machines as high-speed turbine drivers, large-sized centrifugal compressors, fanners, strainers, etc, which are closely associated and interacting with the production engineering and form a large-scaled system. Once faults occur in such a system, it tends to give rise to severe accidents and great economic losses. With the development of studies on theories and techniques of condition monitoring, early warning, fault diagnosis and analysis, and maintenance, and with the extension of their applications to engineering, the above techniques will play significant roles in ensuring the secure and stable running, improving the using efficiency, extending the life-span of machines.Under the nation’s guideline of using information technology in recasting traditional industries, all enterprises are increasing the investment into their informationization construction. However, the informationization construction in the process industry should not be simply regarded as the pile-up of all the above techniques and their engineering applications. Instead, an architecture of equipment safety and risk management that can organically integrate all the above technologies within it, based on the combination of the characteristics of the process industry and the management mode of enterprise equipment, should be established. Moreover, a complete solution, which includes primary functions of the basic management, running and maintenance management of equipment, and integrates various software of monitoring and specialized analysis, embodies an advanced concept of management, actually improves the managing level of equipment and the maintaining reliability and reduces running risks and maintaining costs of the enterprise equipment, should be made.This subject is under the direction of the theory of diagnostic engineering, with the object of rotational machinery in the process industry, and treats the intelligent maintaining of rotational machinery as the main thread, with the target of improving the real-time performance and robustness of early warning systems on the large-scaled rotational machinery, boosting the exactness of fault diagnosis on the large-scaled turbine-driven centrifugal compressor and the scalability of the framework of expert systems on fault diagnosis, as well as advancing the planning and reliability of maintenance and reducing the maintaining costs. Researches involved in the paper are carried out, and a "Risk-Based Dynamic Intelligent Maintaining Platform" has been developed based on related theories and found its applications in large smelting enterprises. Detailed researches are briefed as follows.1) Based on the Deming circle "PDCA", i.e. Plan-Do-Check-Action, a process of intelligent maintenance and security and risk management in enterprises of the process industry is proposed. A new kind of enterprise system integration software architecture based on the Service-Oriented Architecture is studied, based on which a Risk-Based Dynamic Intelligent Maintenance Platform is developed. The platform can organically integrate all kinds of systems in charge of condition monitoring, fault diagnosis and analysis, risk analysis, maintenance management and asset management, etc., that are serving the enterprise. In addition, the management of engineering assets and that of equipment security risk in the enterprise are combined in use of the business process and functional services offered by the platform. Businesses are gradually integrated through the process, and a management system of equipment security risk will be established. 2) The limitations and defects of the traditional reliability-centered maintenance theory when used in the process industry are analyzed, and the theory and method of risk-based dynamic intelligent maintenance of rotational machinery are proposed.3) The fault early-warning system is developed and released. Referencing to the characteristics of typical faults found in the rotational machinery and from the aspect of fault diagnosis, the monitoring and warning characteristic parameter set is extracted purposely and classified. An intelligent warning system with the regular warning, fast warning and slow warning organically combined is devised. Meanwhile, targeting at problems currently existing in the fault early-warning system, such as difficult selection of the characteristic warning threshold, large percentage of error warning and low sensitivity, a self-learning system of eigen-value warning thresholds is studied and established, which realizes the real-time catch on the minimal change of the characteristic parameter set.4) Targeted at the limitations and problems existing in the diagnosis expert system on the fault of the turbine-driven centrifugal compressor, the architecture of the diagnosis expert system on the fault of the rotational machinery based on the theory of the D-S evidence is studied and devised. Taking advantage of the SOA, the system architecture realizes the multi-dimensional knowledge acquisition, and thus improves the reliability of diagnosis. Based on the enterprise service bus, a topological structure of the fusion of data and knowledge of the fault diagnosis is proposed. On the basis of analyzing the fault mechanism of the turbine-driven centrifugal compressor, the fault space and symptom space used for the fault diagnosis and analysis are summarized, and the reasoning rules and algorithms with the combination of the evidence body based on the D-S evidence theory and the target proposition.