A Performance Evaluation Method for Complex Industrial Products Based on Hybrid Neural Network Algorithm

Abstract

The invention discloses a performance evaluation method for complex industrial products based on a mixed neural network algorithm. First determine the HNN structure parameters; secondly, initialize the connection weights and membership function feature points of each layer of neurons, set the error limit, iteration number, learning rate and momentum factor; then quantify and normalize the input fuzzy sample data, quantify The values ​​are normalized; in repeated iterations, the learning error derivatives of each layer of neurons are calculated, the connection weights are corrected, and a gradient descent method is used to adjust the feature points of the degree of membership; repeated iterations until the set error is reached, and finally the HNN algorithm gives the Quantitative performance prediction results for complex industrial products. The HNN algorithm provides a solution with good prediction accuracy for complex industrial performance evaluation problems in actual situations based on qualitative, quantitative, and qualitative-quantitative combination of three types of data input, and the dimensions of input data items may be different.

Description

technical field [0001] The invention belongs to the fields of computer simulation and algorithm optimization, and relates to a performance evaluation method for complex industrial products based on a hybrid neural network (HNN) algorithm. Background technique [0002] A complex industrial product is usually a comprehensive system that performs a specific function, or a combination of some cooperative working units. With the development of science and technology, the complexity of products increases, and the corresponding research and production costs also increase. Typical complex products include aircraft, complex mechanical and electronic products, automobiles, weaponry systems, etc. In the development and production of industrial products, how to effectively evaluate the performance of products (especially complex industrial products) has great practical and economic significance. Through product performance evaluation, products with better performance and less cost can...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to solve the problem of different types and uncertain quantities of product input data in the performance evaluation process of complex industrial products, and propose a method for evaluating the performance of complex industrial products based on the HNN algorithm

Images