Research On Real-time Localization Algorithm For Underground Wireless Sensor Network In Coal Mine

Coal is the driving force to support the sustainable development of China's economy.Coal mine safety accidents are also a major obstacle to the healthy development of the coal industry.Therefore,coal mine safety issues cannot be ignored.Most of China's coal mines are well coal mines hundreds of meters deep underground,which increases the risk factor for operators,and the perfect underground safety monitoring system is the guarantee for the smooth operation of underground work.As the key technology of underground safety monitoring system,coal mine target positioning has become the research focus of today's scholars.Aiming at the shortcomings of low positioning accuracy and poor real-time performance of radio frequency identification and positioning system adopted in most coal mines in China,this paper studies the underground coal mine communication environment and wireless sensor network localization algorithm,and determines the research topic--Research on real-time RSSI positioning algorithm of coal mine underground wireless sensor network.This paper introduces the basic principles,advantages and disadvantages of several classical ranging positioning algorithms,non-ranging positioning algorithms and position calculation methods.By analyzing the underground coal mine environment,the ranging error problem may be brought to the ranging positioning algorithm,and the fingerprint matching non-ranging positioning algorithm is selected as the positioning algorithm of this paper to reduce the distance error caused by the signal strength conversion.Since the space environment under the coal mine is changing in real time,it is easy to cause the fingerprint map to be invalid at the online stage.Therefore,this paper solves the problem of real-time positioning by correcting the target node fingerprint and updating the fingerprint map.In terms of real-time,the dynamic compensation algorithm is a way to solve the realtime matching and positioning by correcting the fingerprint of the target node.In this paper,the beacon node is upgraded to a calibration node.The variation of the RSSI between the calibration nodes reflects the time-varying characteristics in the roadway,and thefingerprint of the target node is dynamically corrected to reduce the matching error between the online real-time fingerprint data and the offline fingerprint map.Segmenting a Neighbor Relationship Model Updating an offline fingerprint map is another way to solve the real-time location problem.In order to solve the problem of uneven environmental change in different sections of narrow and long roadways,this paper classifies the roadway according to the nearest neighbor principle,and assigns the nearest neighbor reference point to each calibration node.The BP neural network establishes the neighbor relationship model of each calibration node,and updates the fingerprint map by calibrating the realtime fingerprint information of the node,which enhances the real-time positioning.In terms of matching and positioning,BP neural network is a commonly used matching localization algorithm.It is slow to converge on BP neural network training process and easy to fall into local minimum.This paper uses PSO algorithm to optimize BP neural network weight and expand weight search.The range accelerates the convergence of the BP neural network.In the same experimental background,the applicability of PSO-BP neural network was verified by comparing the positioning accuracy of PSO-BP neural network with commonly used matching localization algorithm.At the end of this paper,we choose the underground passage that approximates the roadway environment as the experimental scene,and verify the feasibility of the algorithm.The experimental results show that the positioning result of the target node fingerprint dynamic compensation is compared with the positioning without dynamic compensation.As a result,the positioning accuracy was improved by about 15.47%.The positioning accuracy of the second method segmentation update fingerprint map location algorithm is improved by about 20.88% compared with the positioning result of not updating the offline fingerprint map location mode.