Research On Adaptive Target Tracking Algorithm For Downhole WSN Multi-model Estimation

China's coal resources are abundant,and the coal industry has always been an important part of the national economy.In the information age,the introduction of information technology such as Internet + big data into the coal industry will greatly improve its production efficiency.Not only that,but also the effective implementation of monitoring and management under the mine,as well as the prevention of safety hazards and emergency applications in the event of an accident.save.Among them,ensuring the safety of underground personnel and equipment is the premise of the entire industrial chain.On the other hand,the underground environment of the coal mine is complex,and it is easy to have some safety hazards such as gas explosion,water permeable,roof collapse,etc.If the equipment and personnel can be tracked in real time,the rescue after the mine disaster and the usual safety production will be active.enhancement.However,the more mature tracking technology currently has radar and GPS tracking systems,both of which can be widely used on the ground,but the underground coal mine is tens of meters from the ground,there are many internal equipments,and the occlusion signal is serious.The above technology cannot be applied freely underground..Therefore,through the research and analysis of wireless sensor network,tracking algorithm research status and underground tracking system,the research direction of this paper is studied.The multi-model estimation target tracking algorithm of underground wireless sensor network is studied.Firstly,the establishment of the target motion model and the measurement model in the classic Kalman filter target tracking algorithm is studied.Based on the model's thought and the characteristics of the target in motion,the target uniform motion model,the CV model and the uniform acceleration of the target are studied respectively.The motion model,the CA model,the Singer model with a change in target acceleration,and the "current" statistical model.Aiming at the uncertainty of moving target motion mode,the simulation implementation of interactive multi-model target tracking algorithm is studied.Secondly,for the defects of the inter-model transition probability matrix and the invariance of the model set in the interactive multi-model algorithm,two kinds of correction algorithms are proposed respectively: the interactive multi-mode tracking algorithm and the model relationship under the hidden Markov model The KL-VSIMM target tracking algorithm assisted by the graph effectively improves the tracking accuracy and sensitivity and achieves adaptive tracking of the target.Finally,according to the narrow and long terrain characteristics of the coal mine underground roadway,the sensor nodes are evenly distributed.Based on the RSSI-distance curve fitted to the measured data of the underground pipe gallery,the target is positioned,and the local abnormality is combined with the placement position of the downhole sensor and the single sheet.The factor detection algorithm performs the wild-point elimination of the positioning results.Combined with the multi-model tracking algorithm,the tracking system of the distributed network is implemented by the distributed clustering method of the wireless sensor network.