Research On Typical Working Condition Identification Of Excavator Based On Main Pump Pressure

Accurate identification of typical working conditions of excavators is an important guarantee for research related to construction site supervision,excavator energy conservation,and intelligent control.It has important theoretical and practical significance for improving production efficiency and reducing energy consumption.However,existing research is mostly limited to preliminary pattern recognition,and the recognition accuracy of models is not high.Therefore,based on the pressure and differential pressure signals of the two main pumps of the excavator,this article conducts a certain depth of research on the problems of difficult to obtain,insufficient representativeness,low recognition accuracy,and low efficiency of typical working conditions of the excavator.This article conducted a cyclic operation test on excavators.Determine the testing equipment and plan.Collect the pressure signals of the two main pumps from different operators operating in different environments through a data collection system.Based on experimental data,analyze the four typical working conditions that need to be identified in this article: excavation,lifting rotation,unloading,and empty bucket return.Research on filtering and feature extraction of main pump pressure.Based on wavelet threshold denoising algorithm,denoise and filter the main pump pressure data collected from the experiment.Based on the time window and update rate size,data interception is performed on the denoised pressure data for fast Fourier transform(FFT)and time-domain feature extraction.Obtain a feature dataset based on principal component analysis and data normalization.Identification of typical working conditions of excavators based on Honeybadger algorithm(HBA)optimization.Compared with other swarm intelligence optimization algorithms,it verifies the optimization of HBA to the hyperparameter of the classification algorithm.Using HBA to optimize the classification algorithm and obtain a condition recognition model,identify datasets with different time windows and update rates,and verify its impact on the accuracy of condition recognition.Based on the accuracy of each model in identifying different datasets,the best performance working condition recognition model is obtained.End-to-end recognition of typical working conditions of excavators based on Long Short-Term Neural Memory Network(LSTM).Based on algorithms and data features,construct a working condition recognition model to achieve end-to-end recognition of typical working conditions.Using different datasets,verify the impact of time steps and other indicators on the accuracy of working condition recognition.Develop an intelligent software that integrates data preprocessing and condition recognition based on the LSTM algorithm model,achieving real-time visualization of data preprocessing and condition recognition results.