Research On Particle Swarm Optimization Algorithm And Its Application In Space Shuttle Engine Health Management

Particle Swarm Optimization(PSO)algorithm is an intelligent optimization one with advantages of simple structure,few parameters and strong global optimization ability.It is widely used in many fields.Apart from the research of the theory and parameters and performance of PSO algorithm,this paper also constructs an improved particle swarm optimization algorithm and combines it with neural network and support vector machine to establish two models of liquid rocket engine fault diagnosis and uses them in aerospace engine health management.In the improved PSO algorithm,this paper introduces the similarity between particles to determine the degree of aggregation in the particle swarm,and uses the latter one to measure the diversity of the particle swarm.Then based on the degree of diversity in the particle swarm,it makes use of random mutations to diversify the particles.And it re-adjusts the position of the particles to avoid the particle swarm algorithm falling into local optimum.Based on the research of PSO algorithm,this paper adopts the improved PSO to optimize the weight and threshold of BP neural network,and finds the optimal position to assign to BP neural network in order to obtain the optimal network structure.Then using the selected training data of samples to train the fault detection model.Finally,the model is adopted to test the data so as to establish a fault detection model of Liquid-Propellant Rocket Engine(LRE)based on improved particle swarm optimization BP neural network.As the fault samples of actual diagnosis are rare,this paper combines the improved particle swarm optimization algorithm with the Support Vector Machine that suits for detecting the data of small sample,and optimizes kernel width parameters and penalty factors which are key parameters in deciding generalization and stability of the model and then establishes the fault detection model of the least squares support vector machine(LSSVM),which is applied to optimize the particle swarm and the model is applied to the fault detection of the Liquid-propellant rocket engine.Simulation tests proves that the above LRE fault detection algorithms are equipped with high accuracy.In fault detection,they can reduce false alarms and thus improve the safety of space launches.Therefore,it is of great significance.Finally,the paper constructs a space engine health management test platform.After integrating the fault detection model established in this paper,it can complete the fault detection while the aerospace engine health parameters are collected and managed in real time.