Relay life prediction method based on improved RAO algorithm and BP neural network

Abstract

The invention relates to a relay life prediction method based on an improved RAO algorithm and a BP neural network. The method comprises the following steps: step 1, establishing material performance parameters required by a training sample; 2, performing data normalization processing to obtain sample data; step 3, determining a topological structure of the BP neural network; 4, setting an initialization parameter of the improved RAO algorithm, and obtaining an optimal solution through the improved RAO algorithm as a network initial connection weight and a node threshold; 5, selecting training data, inputting the training data into the neural network model, and training a BP network by using an improved BP algorithm; and step 6, reducing the error between the output quantity and the actual quantity by repeatedly adjusting the weight and the threshold value until the precision requirement is met. According to the invention, parameters such as weights and thresholds of the BP neural network are optimized by using the RAO algorithm, the neural network is trained through the improved BP algorithm, error analysis is carried out on the obtained result and a true value, and the precision requirement is met by repeatedly adjusting the parameters of the BP neural network.

Description

Technical field [0001] The present invention relates to a relay life prediction method, specifically, to a prediction based on the remaining life of the relay RAO Improved BP and network fusion method. Background technique [0002] The basic task of the relay contact is good electrical conductivity, good insulation can be separated. The main reason for the failure of the relay fault relay contacts, and the contacts of the various performance indicators reflect the effect of contact with the contacts, and therefore is a measure of the performance parameters of the relay relay product performance, quality and reliability of key indicators, also be life prediction the main basis. In specific areas such as aviation and military, not only for the relay products in terms of reliability and more stringent requirements, and requirements specific occasions in service can not relay any failure, if failure occurs, the accident would not avoid, can also cause serious major security incidents...

AI Technical Summary

Problems solved by technology

[0005] In order to solve the problem that the prediction accuracy of the existing relay remaining life prediction method is not high, the present invention proposes a relay remaining life prediction method based on the fusion of the improved RAO algorithm and BP neural network. The prediction method uses the RAO algorithm to optimize the weight of the BP neural network. Value and threshold and other parameters to improve the accuracy of the BP neural network model, and then train the neural network through the improved BP algorithm, analyze the error between the obtained result and the real value, and then repeatedly adjust the BP neural network parameters to meet the accuracy requirements

Images