Based On Genetic Optimization BP Neural Network Fault Diagnosis Analysisand Research Of Electro-hydraulic Servo Valve

Electro-hydraulic servo valve in electro-hydraulic servo system in a core position, its performance is good or bad direct influence on the performance of the system control precision, reliability and so on; In iron and steel, metallurgical and other industries, the electro-hydraulic servo valve is the main part of the failure and accident. Electro-hydraulic servo valve is set in a machine, electricity, liquid, and its characteristics of high precision, high integration makes the electro-hydraulic servo valve fault type is complicated, and most of the electro-hydraulic servo valve fault type cannot be directly measured, this gives the fault diagnosis of electro-hydraulic servo valve caused great difficulties. Therefore, developed a kind of effective electro-hydraulic servo valve fault intelligent diagnosis method, the electro-hydraulic servo valve may be down for accurate and rapid intelligent diagnosis is very important.In this paper, with a double nozzle flapper two-stage electrohydraulic servo valve as the object electro-hydraulic servo valve fault intelligent diagnosis research, the main research content is as follows:(1) The characteristics of electro-hydraulic servo valve and the fault mechanism research. And analyzes the common fault type electro-hydraulic servo valve, the electro-hydraulic servo valve is fixed on one side of the throttle mouth blockage, nozzle and nozzle baffle block, the main valve core wear, valve core end limit when under the common fault type on the static and dynamic characteristic of the external characterization of features.(2) Electro-hydraulic servo valve characteristic test measurement and control system research. Studying the characteristics of electro-hydraulic servo valve static and dynamic test principle and method, and the NI- PXI data acquisition equipment and Labview programming test software for measurement and control system design. In the process of data collection is inevitable to introduce many kinds of noise, therefore, to study comparing several digital filter methods commonly used in industry, to find an efficient and accurate static and dynamic data filtering method, to get accurate data for later experiment.(3) The BP neural network and genetic optimization theory research. Study the structure of the BP neural network model and the learning algorithm, and combined with the traditional BP neural network is slow training speed, easy to fall into minimum, shortcomings and so on the initial weights matrix dependence is larger and parallel search ability of genetic algorithm with global advantage proposed genetic algorithm was applied to the design method of the BP neural network.(4) Genetic optimization accuracy analysis of the BP neural network. Design five kinds of typical fault type electro-hydraulic servo valve, valve and acquisition under the five kinds of fault type of characteristic curve, the curve as the fault sample input to the design of the fault intelligent diagnosis for fault diagnosis in the program, based on genetic optimization test results of BP neural network and the traditional BP neural network, this paper compares and analyzes validate genetic optimization of BP neural network of character of rapidity, accuracy and robustness.