Online Monitoring System Of Coal Moisture Content Under Dynamic Application Scenario

Coal moisture content is an important indicator for coal quality evaluation and port environmental management,which is closely related to coal quality and dust suppression strategy.If the coal moisture content is too high,the coal quality will be degraded;if it is too low,a large amount of coal dust will be generated,which will lead to a more severe environment.The complexity of port conditions,the large number of coal types,and the large differences in the moisture content of different coal types make it a challenge to accurately monitor the moisture content of a large number of different coals.In this study,we analyze the current research status of coal moisture content monitoring,and design an online coal moisture content monitoring system under dynamic application scenarios for a variety of coal transfer processes in ports.The system collects microwave loss energy and coal seam height,and adopts a machine learning based water content prediction algorithm combined with coal inspection history data to achieve the purpose of real-time monitoring of water content of multiple coal,and the contributions of this study are as follows.Firstly,in order to solve the problem of abnormal sensing data under dynamic application scenarios,this paper proposes an algorithm based on support vector machine(SVM)algorithm and cubic spline interpolation for abnormal data processing;to enhance the prediction accuracy of coal water content,this paper analyzes the correlation between sensor data,time,coal type,coal inspection attributes and coal water content,conducts feature selection,and selects input attributes with high correlation.Secondly,to solve the problem of large differences in water content of different coals,prediction models are established separately for each coal;to solve the problem that it is difficult to unify prediction models due to different data volumes of different coal types,linear prediction models of coal water content suitable for coal types with low transfer frequency and coal water content prediction models based on extreme gradient boosting(XGBoost)algorithm suitable for coal types with frequent transfers are proposed respectively based on coal transfer frequency.The system can monitor 54 coal types in the error range.Finally,in order to verify the superiority of the water content prediction model,the sensor mounting bracket,the system hardware circuit and the system software were designed to establish the online coal water content monitoring system,conduct comparison experiments and monitor the port data in real time.In the comparison experiments,the XGBoost model works best with 14 coal types,water content range from 8%to 27%,mean absolute error(MAE)of 0.53,model fit(R~2)of 0.9735,and mean square error(MSE)of0.87.The reliability of the system and the superiority of the model are confirmed by one and a half years of monitoring experiments.