A Fault Prediction Method for Electronic Steering Systems Based on Time Series Data Analysis

Abstract

This invention relates to the field of intelligent monitoring technology, specifically to a fault prediction method for electronic steering systems based on time-series data analysis. The method includes: mapping the steering system as virtual computing nodes for perception, decision-making, and execution to construct a digital twin model; extracting time lag correlations between nodes through multivariate Granger causality operations to construct a virtual causal matrix; injecting fault parameters such as current, angle, and bias into the model for numerical simulation, and generating a response sequence using time-domain iteration; calculating the morphological difference between the response sequence and a reference sequence using a dynamic time warping algorithm to generate time-series features; constructing a dynamic digital twin graph and inputting it into a dynamic graph attention network to extract spatiotemporal evolution features and generate fault prediction results. This invention achieves accurate simulation of the entire lifecycle fault evolution process of the steering system through digital twins.

Images