Research On Optimized Neural Network Indoor Positioning Algorithm Based On Zigbee Technology

The arrival of the information age has promoted the rapid development of Internet of Things technology and high-frequency wireless technology in the close combination.People's demand for location services is increasing day by day,especially in the complex indoor environment,mobile terminal location information is often needed.Therefore,how to achieve low-cost,high-precision indoor positioning,has very important practical significance.As one of the wireless sensor network(WSN)technologies,ZigBee technology has high application value in indoor positioning because of its low cost,low power consumption and high reliability.This paper expounds the research background and current situation of indoor positioning technology,introduces ZigBee technology,and deeply studies the advantages and disadvantages of typical WSN positioning algorithm.By comparing and analyzing,the positioning algorithm based on received signal strength value(RSSI)is selected.However,the traditional location algorithm based on RSSI is to obtain the distance between the receiving and sending nodes according to the wireless signal propagation model,and then uses the position distance algorithm to estimate the position.However,the parameters in the model are easily affected by the environment,and their values are usually obtained by fitting or directly based on experience,which inevitably lead to inaccurate positioning.In order to improve the positioning accuracy,a generalized regression neural network(GRNN)with strong non-linear fitting ability and fault-tolerant ability is introduced to construct the positioning model.The RSSI value between the unknown nodes and the reference nodes is used as the input of the network,and the position coordinates of the unknown nodes are used as the output to fit the network model.At the same time,elimination processing and Kalman filter processing are used to preprocess the collected RSSI values to weaken the disturbance of environmental factors.In order to avoid the randomness of GRNN parameter selection and the interference of human factors,the swarm intelligence algorithm-chaotic quantum particle swarm optimization(CQPSO)is used to optimize the smooth parameters of the network,and the fitness function is constructed by choosing the root mean square error between the predicted coordinates and the actual coordinates of the positioning nodes.The global optimization ability is enhanced by the quantum characteristics of the particles,and combines the chaotic characteristics to avoid the population falling into local optimum and increasing the diversity of the population,to perform a finite number of iterative optimization,and finally the smooth parameters corresponding to the minimumfitness function value are searched to establish the optimal network positioning model to realize the prediction of unknown node coordinates.On the platform of MATLAB,the positioning algorithm based on CQPSO-GRNN model is compared with the unoptimized GRNN positioning algorithm.The results show that the improved positioning algorithm has higher positioning accuracy,and improves the prediction effect of GRNN regression and the generalization ability of the model.Finally,the CC2430 wireless positioning chip developed by TI company of the United States is selected to design the software and hardware of the positioning system.In order to display the positioning results more vividly,a simple PC monitoring interface is developed by using QT programming software.The workshop of the teaching building is selected as the experimental environment to test the improved location algorithm.The results show that the positioning error can be controlled within 1 m,which basically meets the requirements of complex indoor environment for positioning.