Study On Fault Diagnosis Method Of Mine Ventilation Network Based On Machine Learning

Due to the numerous underground branches and complex topological relationship,ventilation network fault diagnosis is always a difficulty.Therefore,exploring an efficient ventilation network fault diagnosis method is of great significance for ventilation safety.This dissertation combines theoretical analysis,experimental research and software simulation to establish a machine learning algorithm model for fault diagnosis.The main research results are as follows:(1)The fault diagnosis index of mine ventilation network is determined,and the steps of using machine learning algorithm for fault diagnosis of ventilation network are proposed.Three machine learning algorithms,support vector machine,random forest and neural network,are used to establish the fault diagnosis model with air volume,air pressure and their composite characteristics as input characteristics respectively.(2)The experimental platform for fault diagnosis of mine ventilation network is built,the experimental research on fault diagnosis of ventilation network is carried out,the parameters of the diagnosis model are optimized,and the neural network model based on the combined characteristics of air volume and air pressure is determined as the best fault diagnosis model.(3)The neural network fault diagnosis model based on the air volume and air pressure composite characteristics is applied to the ventilation networks with three different ventilation modes: Central parallel,zoning,central separation and zoning hybrid.The results show that the fault diagnosis accuracy of the neural network model on the three ventilation networks is more than 0.9,which verifies that the neural network model with the monitoring air volume and air pressure composite characteristics is suitable for mine ventilation networks with different ventilation modes.