Recognition And Prediction Of Catting Pick Abrasion Condition Status Based On Grey-Markov

As a key component of mining machinery,the pick has the direct direct impact on work efficiency.Due to the harsh environment and noisy noise,how to extract characteristic signals is a problem that needs to be solved today.The research on the identification and prediction of picking wear state is an important basis for enhancing the working efficiency of mining machinery,ensuring mining safety and promoting intelligent mining.The paper mainly studies the influence law of the characteristic signal and selected characteristic parameters generated and extracted in the picking process during the cutting process.The following work has been done for this purpose:This paper extracts and analyzes the vibration and acoustic emission signals generated during the cutting of the pick,and analyzes the extracted characteristic signals.The time domain,frequency domain and wavelet packet analysis are respectively carried out;the root mean square value of the vibration acceleration time domain signal;the root mean square value of the vibration frequency domain signal;the energy and value of the vibration signal wavelet analysis;the frequency domain signal of the acoustic emission signal Square root value;the energy and value of the acoustic analysis signal wavelet analysis to construct a feature sample library of the signal.Based on the BP neural network and the PNN neural network,the picking wear state recognition model is established.After the BP neural network recognition model and the PNN neural network recognition model,the highest overall recognition accuracy of the BP neural network recognition model is 99% and PNN.The highest recognition accuracy of the neural network recognition model is 100%.In this paper,the Gray gm(1,1)model and the Gray-Markov model are applied to the prediction of pick wear state.The Grey gm(1,1)model and the Gray-Markov model are used to predict the pick wear state.The results obtained from the study show that the average relative error of the Grey gm(1,1)prediction model is 2.3%,and the average relative error of the Gray-Markov model is 1.06%.By comparison,it is found that Gray-Markov is used,The Cove model is more able to fit the actual value curve and the accuracy is increased by 54%.For this reason,different error parameters are applied to the Gray-Markov model for error comparison.The results show that the energy and value of vibration acceleration signal are the best for Gray-Markov prediction,and the average relative error is the smallest.The average relative error is 1.45%.Finally,different signals are applied to grey-Markov model for error comparison.Theresults show that the acoustic emission signal is more suitable for grey Markov prediction model than the two kinds of signals,and the relative error is 1.1%.This provides a new method for the accurate prediction of picking wear state of fully mechanized mining face.There are 33 figures,23 tables,152 references in this paper.