Application Of Information Fusion In Coal Mine Safety Monitoring

As we all know, our country is one of the countries with the most abundant coal resources in the world at present, accounting for a third of world consumption of coal, which shows the importance of our national coal production. At the same time, our country is the world’s coal mine disasters and serious country. Although a significant reduction in the data of China’s coal production accidents in recent years, but the problem is still very serious and the risks, China’s coal mine industry is still more serious problem. Thus, coal mine safety monitoring and reduce the accident rate still is the inevitable trend of coal production work.Coal mine placed a large number of coal mines to detect a wide range of sensor data, including CO sensors, temperature sensors, humidity sensors, wind speed sensors, dust sensors. However, changing the face of the mine and complex cases, there may be a single sensor failure, errors and other problems, any problems can cause a sensor to the whole environment of mine safety monitoring results. Therefore, this issue will introduce a two information fusion method for multiple sensors to monitor the coal mine data set for analysis, improving the accuracy of monitoring.According to the situation underground in a coal mine in Tangshan, the use of a temperature sensor, CO sensor, gas sensor research, through integration of monitoring data returned by the judgment of mine safety. Fusion is the use of a fusion algorithm jackknife and adaptive weighted combination of each of a plurality of monitoring data for each sensor processing, monitoring results mean for each sensor; two fusion using DS evidence theory, adopt a basic probability assignment confusion matrix based on a new method of constructing a data fusion output to three sensor fusion again. First it is known the actual state of training a large number of monitoring data, obtained for each sensor in the security, and the risk of an early warning three states corresponding to the BPA, and thus generate the evidence base. Again when monitoring data to predict, based on the preliminary criteria to judge the security status and find the evidence base to obtain the corresponding BPA, use DEMPSTER rules to obtain the final integration of the monitoring state as a result.Respectively for single sensor and the method, and the analysis, the results show that this method of downhole environment security situation has significantly improved prediction, and thus the validity of the proposed research program. Also in terms of the cost of time, the process in time than a single sensor used downhole environment forecast increase in the time 0.026 s, but this delay will not have timely warning of underground dramatic effect and delay, therefore sum up this method is a research significance.