Research On The Method Of Coal-rock Recognition Based On Technology Of Multi-sensor Information Fusion

In the mining process of coal mine，coal and rock are mutually staggered andunsystematic，so it brings difficulties in mining operations.In order to improve theproduction efficiency and reduce unnecessary work，we need some technologies to docoal-rock recognition which can direct the operation of mining coal.The coal-rock interface recognition is the key technology of unmanned mining inoperation of mining coal.According to coal-rock recognition，we can adjust the heightof roller of coal winning machine.It can not only improve efficiency，but also canreduce cutting rock，cut down content waste rock in coal.At the same time，it canavoid cutting rock which may attrit gearwheelon.Above all，it can avoid gasexplosion caused by high temperature of cutting rock.With the widespread application of sensor technology，we consider using sensorto implement coal-rock recognition.Now we often use single sensor.However，usingonly one sensor has many shortage.For instance，its signal is single and erratic.So weconsider using different types of sensors at the the time，in order to collect responsesignals of coal winning machine， extract signals and carry on informationfusion.Finally，we can achieve intent of coal-rock recognition.In this paper，we collect signals by wavelet packet method.we use the waveletpacket technique to determine the sensitive signal frequency band of sensor，extracting signal features，eventually it transforms between pattern space and featurespace.So that we can get precise feature data for data fusion in the next step.Information fusion (data fusion) technology is an synthesized informationprocessing technology，so using it for coal-rock recognition is very feasible.Beacausewallflower of BP network，in this paper we improve BP network by genetic neural network.So we establish model of the neural network，fix specific structure andalgorithm.The results showed that， experimental data can train and simulate neuralnetwork.We combine the genetic neural network and data fusion，and it can achieverelational recognition.Compared with other methods,this method has higherrecognition accuracy.Our mothd of coal-rock recognition in this paper can cometrue.