New energy automobile electric control system fault prediction method based on working condition data

Abstract

The invention provides a new energy automobile electric control system fault prediction method based on working condition data. On the basis of collecting fault data, such as data of fields of a controller output deviation, controller response data, an instrument display deviation and the like, and fault samples of a new energy automobile electric control system, the data are used as support of anew energy automobile fault prediction database, the fault samples are used as learning samples, a neural network-based electric control system fault prediction model is established in combination with a neural network prediction method, a final probability prediction matrix is obtained through matrix operation and conversion of a softmax layer, and a fault with the maximum probability is selectedas a final prediction result. A training method of the model is a random gradient descent method, the neural network model which can be used for actual electronic control system fault prediction is finally formed by continuously iterating until the error is less than a threshold or the number of iterations is greater than a set value, and the prediction accuracy rate of the fault is more than 96%.

Description

technical field [0001] The present invention relates to the field of new energy vehicle detection fault prediction, in particular to the field of fault prediction of neural network and vehicle electronic control system. Background technique [0002] At present, a relatively complete testing process and system have been formed for the announcement testing and delivery testing of new energy vehicles, and a considerable number of industry standards have also been formulated. [0003] At present, there are 42 new energy vehicle standard projects (including 6 electric motorcycles) that have been released in terms of generality, safety, interchangeability, technical conditions and test methods of new energy vehicles and key parts and components, of which There are 35 national standards and 7 automotive industry standards. According to the technical route, the standard includes 11 pure new energy vehicles, 6 hybrid vehicles, 4 fuel cell vehicles, 6 electric motorcycles, 8 power ba...

AI Technical Summary

Problems solved by technology

First of all, the nature of the use of new energy vehicles in use determines that the annual inspection cannot be disassembled and tested for new energy vehicles like the other two tests, and the functions and structures of each module are individually tested; Based on the principles of non-destructiveness and repeatability, it is impossible to conduct acupuncture, wading, overcharging and overheating tests on batteries like factory inspections; finally, the annual inspection of new energy vehicles in use has a strong timeliness and needs to be Generate a test report on the spot in a short period of time, unlike the factory test that can be carried out for a long time

Images