Feature Extraction of Loader Operation Based on Kernel Principal Component Analysis

The response characteristics of the vehicle are complex under different working conditions. To facilitate the calculation and analysis of a variety of attributes, based on the kernel principal component analysis method, loaders are used as the research object of this article, data signals of 11 attributes such as throttle signal, speed, pressure, etc. are collected. After denoising and reconstruction of the original signal, the variance contribution rates of principal components 1-4 are 50.99 %, 29.90 %, 14.50 % and 4.27 % by using the kernel principal component analysis. The variance contribution rate of the original 11 attributes is 99.66 %, which could accurately reflect the working condition information. The research methods of multiattribute dimension reduction in this paper could be applied to road construction machinery, mining machinery, petroleum machinery and other engineering fields.