Research On The Key Technologies Of Fault Diagnosis And Detection Based On Artificial Immune Theory For Generating Plant

Fault diagnosis is the science of finding fault source and making corresponding decision according to the equipment running information.It is necessary to study all kinds of advanced fault diagnosis technology and apply it to the fault diagnosis of advocate complementary equipment in power plant,which not only conforms to the development direction of fault diagnosis theory,but also is urgently needed in the production process.At present,there is a widespread data generating equipment fault diagnosis,lack of fault samples,diagnostic methods,the lack of continuous learning ability lead to abnormal trend of equipment failure is not easy to determine and low detection efficiency.The artificial immune system based on the biological immune mechanism provides a new approach to solve these problems.This paper focuses on the negative selection algorithm in artificial immune system,and studies the two core problems of "self-space determination" and "generation of detector" in the negative selection algorithm.AiNet artificial immune network is mainly studied on the data distribution optimization and combination of the Voronoi diagram and immune detector.Respectively,the aiNet artificial immune network model of negative selection algorithm and optimization of negative selection algorithm based on Voronoi detector was put forward.There is an uneven distribution of boundary,data redundancy,difficult to identify and complex problems.In this paper,an aiNet artificial immune network model is proposed to compress self-space,which will reduce or increase the data density to the appropriate degree,and the data is distributed roughly evenly in the self-space.Then the boundary data of the data set is described as the self-space.According to the above theory,a negative selection algorithm for the optimization of aiNet artificial immune network model was proposed,that the description of autologous space was simplified,and the calculation time of detector generation was reduced.On this basis,a negative selection algorithm for the optimization of aiNet artificial immune network model is proposed.The new algorithm simplifies the description of self space and reduces the calculation time of detector generation.For the general artificial immune detector,there are problems with low coverage,"holes" and redundancy.This paper presents the concept of Voronoi detector and the negative selection algorithm based on Voronoi detector.The algorithm takes advantage of the Voronoi diagram of spatial partitioning advantage,at the same time avoiding the detector "holes" and redundancy and redundant detector of computing time.Therefore,the coverage of the detector is improved.The simulation experiment is based on the Iris data of UCI data and the actual production data of the power plant,compared with existing RNSA algorithm and V-Detector algorithm.Simulation results show that the algorithm based on Voronoi detector has higher accuracy and faster recognition of abnormal data.Finally,using Qt development framework,the simulation platform of generating plant fault detection system based on artificial immune principle was implemented.On the engineering point of view,the application of immune algorithm to fault detection is verified.