Research On Fusion Method Of For Hydraulic Power System's Multi-source Fault Information Based On Neural Network

In the light of the characteristic and development of the hydraulic system fault diagnosis,this paper,based on the information fusion theory,presents an intensive to the acquisition,analysis and fusion of information from multi-sensor applied to the fusion technique of hydraulic power system's multi-source fault information.The research covers the following essential contents and conclusions:(1)The paper first summarized the actuality & development of modern hydraulic system fault diagnosis technology,stressed the importance of multi-source information fusion,introduced the fusion configuration and algorithm of multi-sources fault information,which is applicable for fault monitoring,alarm and diagnosis.(2)Detailed introduced the multi-source fault information collection and analysis of experimental system of hydraulic power system.Finally,through the examples used to verify that the system is practical.(3)According to the application features of multi-source information monitoring technique and the operation parameters of hydraulic system,and then bring forward the multi-source information monitoring system model operation status of hydraulic system; The multi-source information of hydraulic system was analyzed such as current signal and pressure signal;In addition,it analyzed choices of bridge type while monitoring pressure signal with resistor strain gauge,effects of pipeline shape to pressure signal and selections of monitoring signals.(4)On basic theory and characteristics of Counter-propagation Networks,and fusion the method of spectrum analysis used for typical fault classifications to hydraulic power system;Compared with BP neural network,CPN showed its advantage. (5)The fusion technique of multi-source fault information,based on the fuzzy CPN neural network,solved the problems such as inadequate information offered by single-source and misjudgment induced by effects among faults;The multi-source fault information fusion model applied to hydraulic system has been given and fuzzy membership function has been built.Effectiveness has been authenticated by application examples at last,at the same time,it showed that it is necessary to combine neural network with other fusion methods.