A prediction method for crankshaft fatigue life based on genetic nerve network

Abstract

The invention discloses a prediction method for crankshaft fatigue life based on a genetic nerve network, belongs to the field of crankshaft fatigue life testing of an internal combustion engine. The purpose of the method is to solve the deficiency of a DC resonant-type crankshaft fatigue life testing machine for possessing destructiveness to the crankshaft testing and having long testing time. The principle of the method is to normalize the historical data of the crankshaft testing by utilizing the conventional DC resonant-type crankshaft fatigue life testing machine to obtain a training sample set; to optimize a BP artificial nerve network model through genetic algorithm; to carry out iteration training to the genetic algorithm-based and optimized BP artificial nerve network by utilizing the training sample set to obtain a trained BP artificial nerve network prediction model; to use the prediction model to carry out rapid prediction for the crankshaft fatigue life. The method optimizes the BP artificial nerve network based on the genetic algorithm, avoids the "over fitting" problem of the single BP nerve network, improves the training speed and prediction precision effectively; rapidly predicts the crankshaft fatigue life in a short time without destroying the crankshaft quality, is capable of carrying out a batch testing for the crankshafts of a whole production batch.

Description

technical field [0001] The invention relates to a method for predicting crankshaft fatigue life based on a genetic neural network, which belongs to the field of fatigue life testing of internal combustion engine crankshafts. Background technique [0002] The safety and reliability of mechanical products are important indicators of the technical level. The reliability and failure issues within the product design life have an important impact on human production and life, so they have been widely concerned by people. The crankshaft is a key bearing part of the internal combustion engine, and one of the main contents of its reliability is its fatigue life under the action of high cycle fatigue stress. [0003] On the one hand, due to the complexity of the crankshaft fatigue life problem, there are still many limitations in the current technical level through purely mathematical derivation and calculation analysis of the reliability of the crankshaft in actual work. For example,...

AI Technical Summary

Problems solved by technology

[0006] The present invention provides a crankshaft fatigue life prediction method based on genetic neural network, using the historical data tested by the crankshaft bending fatigue testing machine as a training sample set, training the genetic neural network model, and quickly predicting the crankshaft fatigue life to solve the existing DC resonance The crankshaft fatigue life testing machine is destructive to the crankshaft test, and the test time is long. It can quickly and timely predict the fatigue life of the crankshaft.

Images