Novel BP neural network calculation method and flexible factor optimization design method thereof

Abstract

The invention relates to the technical field of flexible factor optimization, in particular to a novel BP neural network calculation method and a flexible factor optimization design method thereof. Acalculation method of a novel BP neural network comprises the steps of increasing momentum items and adjusting the learning rate of the network, and an optimization design method for flexible factorsbased on the calculation method of the novel BP neural network comprises the steps of simulation modeling and analysis, data acquisition and database establishment, database import and actual effect verification. According to the method, a plurality of groups of sample pairs different from a training set are used as test samples to verify the calculation capability and generalization performance of the novel BP neural network, wherein vibration information in the test set samples is used as input of the BP neural network, and then vibration information of a rigid body structure added with flexible factors is calculated through a trained BP neural network model. Therefore, a relatively good calculation result is obtained.

Description

technical field [0001] The invention relates to the technical field of flexible factor optimization, in particular to a calculation method of a novel BP neural network and a flexible factor optimal design method thereof. Background technique [0002] Artificial Neural Network (ANN), referred to as neural network, its fundamental purpose is to realize the intelligence of the computer, so that the computer can quickly learn and memorize new knowledge and new laws. Its main research is how to let the computer simulate and realize the self-learning and thinking ability of human beings, and can mine the potential connection between physical objects and data from limited samples. Neural network is an important branch of the field of artificial intelligence. It mainly derives and reveals the invisible relationship between variables in unknown data samples by learning and storing potential data relationships, inference rules, probability distributions and other information in known...

AI Technical Summary

Problems solved by technology

[0009] (1) The number of training times lacks on-site theoretical basis, and too many training times will reduce the learning efficiency of the algorithm and affect the network convergence speed;

[0010] (2) It is easy to form multiple minimum values, and the error surface is easy to generate flat areas

[0011] (3) There is no theoretical basis for the number of neurons in the hidden layer. At present, it mainly depends on the designer's experience or repeated operation and parameter adjustment, which affects work efficiency.

This design method often lacks actual data and experiments as the basis and support

Images