Remaining Life Prediction System Of Key Parts Of Coal Shearer Rocker Arm Based On Deep Learning

Coal mines are one of the important energy sources in our country.As a vital equipment for coal energy mining,shearers are responsible for coal cutting and coal loading tasks,but they are under severe working conditions,high-intensity,high-load working conditions and delayed maintenance.Measures and other factors lead to frequent failures,reduce equipment efficiency,endanger the production interests of enterprises,society and the country,and threaten the lives of relevant personnel.Therefore,efficient and reliable health management of shearers is imperative.The shearer rocker arm is the key actuator of the shearer,and its remaining life is closely related to the health of the whole shearer.Therefore,it is extremely important t o predict the remaining life of the key components of the shearer rocker.At present,research on remaining life analysis and prediction based on deep learning theory is in full swing,but related research on shearer equipment is still not comprehensive enough,and most of the research is mainly based on expert experience or static simulation of software and mathematical models,lacking real-time and dynamic performance.It also lacks practical exploration to transform theoretical analysis into practical application.With the emergence and maturity of new theories and technologies such as machine learning,neural networks,big data,and Internet+,it is a general trend to build an intelligent,economical,and efficient and accurate remaining life prediction platform for equipment.This paper adopts the method of combining deep learning theory and system development technology,takes the key parts of the shearer rocker arm as the main monitoring and research object,and takes the dynamic prediction of the remaining life of the key parts as the main purpose,and builds the residual life based on deep learning.Life prediction neural network model,development of the remaining life prediction system for the key parts of the shearer rocker arm based on deep learning.It has realized the transformation of deep learning theoretical analysis to practical application,and enriched the health management methods such as predictive maintenance of the shearer to meet the needs of users,and conform to the trend of social progress and the development requirements of the new "four modernizations".Analyze the failure phenomenon and reasons of the key components of the shearer rocker arm,clarify the components of the remaining life prediction system.Analyze the data types of the key parts of the shearer rocker arm,and build a deep convolutional neural network(DCNN)model for the remaining life cycle analysis and prediction for the parts with full life cycle data.Data components,construct a deep neural network model combining auto-encoder(AE)and bidirectional gated recurrent unit(bi-GRU)to analyze and predict the remaining life,and give full play to the prediction characteristics of each model And forecasting ability,improve the difference and pertinence of component analysis,and improve the effectiveness and accuracy of forecast results.Explore the means of embedding deep neural network models into the system environment,encapsulate various types of deep neural network models covering data extraction,data preprocessing,feature extraction,remaining life analysis and prediction,etc.,and use the process class in C# language to solve the Python language Compatibility between the remaining life prediction model in the environment and the Web system in the C# language environment.Based on mature ASP.NET system development framework and B/S system architecture,using Microsoft Visual Studio 2010 system development platform and Microsoft SQL Server 2008 R2 database software as system development means,combined with Echarts image generation tools,Python programming language and other collaborative technologies,Design and develop a system for predicting the remaining life of key parts of the shearer rocker arm based on deep learning,and realize the functions of component introduction,monitoring dynamics,life prediction and history records for each key part.After testing,the constructed system has a friendly interface,rich functions and strong applicability,good compatibility,high security and stability.