Improved RBF flight control system fault diagnosis network training method

Abstract

The invention belongs to the technical field of aircraft fault diagnosis, and particularly relates to an improved RBF flight control system fault diagnosis network training method. The method adopts population initialization based on a training sample, and performs crossover operation on a network parameter gene based on a differential evolution operator and an arithmetic crossover operator. The population initialization based on the training sample comprises that: at the time of population initialization, the training sample is used to initialize a neuron center of an RBF network, and the network hidden layer neuron number and the radial basis expansion constant are randomly initialized. The crossover operation based on the differential evolution operator and the arithmetic crossover operator comprises that: in the crossover operation, the crossover operation of the network parameter gene part is based on the commonly used arithmetic crossover operator, and the differential evolutionoperator is used at a certain probability to enable individual parameter genes of an offspring individual to gain richer diversity.

Description

technical field [0001] The invention belongs to the technical field of aircraft fault diagnosis, and in particular relates to an improved RBF flight control system fault diagnosis network training method. Background technique [0002] The flight control system is the core system of the aircraft. The failure of its system components not only affects the performance of the flight control system, but also poses a great threat to the flight safety of the aircraft. Intelligent fault diagnosis technology is applied to the failure of the aircraft flight control system In the diagnosis, assisting the maintenance personnel to eliminate the faults of the flight control system in time, improving the maintenance efficiency of the aircraft, and ensuring the safe flight of the aircraft are currently urgently needed research contents. [0003] The commonly used methods of fault diagnosis of flight control system are model-based fault diagnosis, signal processing-based fault diagnosis and k...

AI Technical Summary

Problems solved by technology

Hierarchical genetic algorithm is a kind of evolutionary algorithm, because the hierarchical structure of chromosomes can represent the network structure and network parameters of RBF network at the same time, and the prior knowledge of data is relatively low during training, so it is widely used in the training of RBF network However, because the genetic algorithm is a heuristic search algorithm, the training time of the algorithm is greatly prolonged compared with other non-evolutionary training algorithms, which restricts its application in RBF network training. The traditional hierarchical genetic algorithm generally includes parameter coding, initial There are six elements including population setting, fitness function design, genetic operation design, control parameter setting, and constraint conditions. The initial population affects the number of iterations of the algorithm, that is, the running time. Traditional hierarchical training algorithms usually use random The way of initialization; selection, crossover and mutation constitute the genetic operation, among which the crossover operation is the main operation of the hierarchical genetic algorithm to enhance the diversity of the population, which has an important impact on the optimization ability and the number of iterations of the algorithm

Images