Rotatory Unit Concurrent Fault Diagnosis Based On Artificial Immune System And Evidence Theory

Rotating machinery is the main production tool in various industrial branches, it is the mechanical device accomplished by the rotary movement, usually is an Unit which is composed of many complex mechanical parts. With the development of science and technology, equipment performance is better and better, their structures are more and more complex. Rotating machinery equipment trends to be the large, high-speed and complex equipment. Accordingly, the probability of the concurrent fault is increased and more fault trends complex. At present, concurrent fault diagnosis fault has become a common problem. When multiple concurrent failures occur, its manifestations are diverse, but failure and symptoms are not one-to-one correspondence, different characteristics of fault are mixed with each other, it shows more coupling and ambiguity. Some faults are not simple linear addition with several single faults, it is difficult to describe accurately by the mathematical model, and also difficult to diagnose entirely with deterministic criteria, it brings extreme difficulty to the diagnosis. As the complexity of concurrent failures, we proposed many multiple intelligent diagnostics ideas to build diagnosis mechanism for multi-concurrent fault. The simulation results show that it brings breakthrough for the accuracy of fault diagnosis.In recent years, our research team used the Bionic mechanism of biological immune system, as dimensionless indices is not sensitive to the characteristics of complicated defects, in view of the immune network of ideas, we has explored a artificial immune system for rotating machinery fault diagnosis based on dimensionless and immune algorithm. System provides the type and range of indicators for the diagnosis of concurrent fault, giving new evidence for concurrent fault. We obtained the time domain waveform and other signal data of fault through experimental sensor, normalizing the various parameters, generating immune detectors online according to the negative selection, getting fault information space according to the immune learning mechanism, comparing with the fault coverage, and determined fault type finally. Artificial immune system analyses dimensionless indicators of time domain for concurrent failures, has drew a widespread law based on non-dimensional index for concurrent fault diagnosis.The research for fault diagnosis based on artificial immune still has many defects. Aim at the diagnosis ambiguity of the concurrent fault and accuracy of fault diagnosis which can be improved, the evidence theory is introduced to the comprehensive fault diagnosis of rotating machinery, through the study on the home and abroad research results. Considering the characteristics and difficulties of the large rotating machinery fault diagnosis, we proposed system model and application methods which is used for rotating machinery fault diagnosis. This paper put forwards the concurrent fault diagnosis based on artificial immune system and evidence theory. The diagnostic strategies of comprehensive diagnostic module have been simulated. In order to overcome the limitation of non-dimensional index on fault judgment, we bring up a new and optimized dimensionless index according to genetic algorithm, verifying the diagnostic results, the ability to identify concurrent fault was improved. Since there are different judge capacities of different fault for each non-dimensional index, In the following experiments, we tested the sensitivity and diagnostic capability of various non-dimensional index for the composite fault gradually, acquiring the weighted coefficient。Therefore, we carried on the information fusion used evidence theory and obtained final decision for each fault. The simulation results show that the accuracy of concurrent fault diagnosis is improved greatly, the diagnosis are more and more clarified.