Intelligent Diagnosis for Aero-engine Based on Multi-Population Immune Algorithm

Intelligent diagnosis for aero-engine is a hotspot issue in aviation field. To improve the efficiency of immune algorithm, a new immune algorithm based on multi-population is proposed. The process of MPIA (Multi-population Immune Algorithm) fault diagnosis is divided into two stages: the establishment of MPIA and the implementation of MPIA. Firstly, several normal sub-populations and one memory sub-population are randomly generated and trained by multi-threading of fault samples simultaneously to achieve the concurrent evolution of multiple populations. Moving-in and moving -out operations can improve the the efficiency of the algorithm. Finally, the main sub-population become the mature population that can be used for fault diagnosis. At the end of the paper, four important parameters of the algorithm are analyzed, including the number of sub-populations, the scale of each sub-population, the inhibition threshold of the memory population and the diagnostic threshold. Considering the convergence rate and the accuracy of the fault diagnosis, the optimal parameters are determined. The results of samples diagnosis demonstrate that this method can effectively recognize aero-engine gas path components faults, and coordinate the accuracy and speed of the immune algorithm.