Shearer Cutting Load Based On Multi-information Fusion Research On Recognition Technology

As one of the key equipment of coal mining,intelligent control of coal mining machine is an important part of building "intelligent mine".Accurate identification of coal rock cutting load state of coal mining machine is a prerequisite for automatic drum height adjustment and one of the key technologies to realize intelligent control of coal mining machine.This paper takes a mine in northern Shaanxi Province as the test site,and takes MG600/1590-WD coal miner as the research object,collects the rocker vibration signal,acoustic emission signal and cut-off current signal under the normal working condition of coal miner,and uses them as the target database to build the data layer fusion and decision layer fusion load recognition models,and based on the recognition results of these two models,builds a multi-level information fusion based Based on the recognition results of these two models,a multi-level information fusion based truncated load recognition model is built.The main work and research results of the paper are as follows:(1)The mechanical characteristics of coal rock load and coal breaking mechanism of drum coal miner are analyzed,and the rocker vibration signal,cut-off audio signal and cut-off current signal are used as the database of coal rock cut-off load recognition features of coal miner,and the framework and technical route of coal rock cut-off load recognition system of coal miner are constructed by combining the process characteristics of well mining.(2)Through rocker loading test,the signal characteristics of vibration signals at different measurement points and downward in different axes are analyzed to provide theoretical guidance for industrial test sensor installation.The vibration signals of y and z axes are selected to characterize different cut-off loads,and the data acquisition scheme of the industrial test is designed by combining the cut-off audio signals and cutoff current signals for the downhole environment.For the obtained data,the accuracy of the data is initially verified by using the short-time energy and short-time over-zero rate,and the signal denoising is completed by the wavelet packet denoising method based on the optimal wavelet basis.(3)The truncated load recognition based on data layer fusion is realized.Based on BP neural network,a data layer fusion load recognition model is built for unprocessed raw data,and truncated load recognition is completed with an accuracy of 78.5%.(4)The truncated load recognition based on feature layer fusion is realized.The time-frequency domain features of the data are analyzed,and the correlation fusion and dimensionality reduction of the features are completed for the selected feature vectors to obtain the feature matrix that can characterize the intercepted load state.Based on the improved sparrow search algorithm,the selection of RBF neural network centers is optimized,and based on this,a feature layer fusion-based truncated load recognition model is built,with 84% recognition accuracy.(5)The decision layer fusion truncated load recognition based on data layer fusion and feature layer fusion recognition results is implemented.Based on the recognition results of data layer fusion and feature layer fusion,the basic probability assignment of each evidence body is assigned,and the decision layer fusion is performed by using DS evidence theory,and the final coal rock cut-off load recognition results are obtained,and the recognition accuracy is 87.3% under the specific coal mining machine and specific working face conditions.