Research On Multi Sensor Cooperative Target Tracking Algorithm Under Uncertain Factors And Multi Constraints

With the development of sensor intelligence,multi-sensor network is widely used in target tracking,because of its complementary monitoring area and information sharing.However,the uncertain working environment of the sensor leads to the measurement deviation of the sensor,and indirectly reduces the tracking accuracy of the system,which is an important problem to be solved.On this basis,how to realize the cooperative consensus of multi-sensor networks further and efficiently has become a challenging problem.This paper focuses on the single sensor system target tracking,multi-sensor cooperative consensus estimation and multi-sensor dynamic optimization with constraints in the uncertain environment.The main contents of the paper are as follows:(1)To solve the target tracking problem of single sensor system with unknown input(UI)interference,a joint estimation algorithm of state deviation based on two-stage Kalman filter is proposed.By establishing a generalized unknown disturbance-driven sensor system deviation and measurement model,and proposed a system model which decoupled from the unknown input,then decoupled to the noise cross term in the new system model to avoid errors caused by related noise.Finally,a two-stage Kalman filter algorithm is designed to estimate and correct the target state and sensor deviation,so as to filter out the influence of unknown input and improve the accuracy of target estimation and deviation estimation.(2)To deal with the problem of collaborative consensus estimation in multi-sensor networks with unknown input disturbances,a distributed collaborative consensus fusion algorithm based on the communication traffic of sensor nodes is proposed.This method achieves the consistency estimation through two-stage information filtering in the local sensor and collaborative consensus in the network.Two stage information filtering is used to reduce the computational burden of the system,and the unknown parameters and state estimation are put into the network for consensus.According to the topological relationship of sensor networks,a data fusion weighting method is proposed,which gives higher weight to nodes with high traffic in the network.Compared with the average weighting and covariance weighting method,the convergence speed of consistency estimation is improved,and the target tracking accuracy is further improved.(3)An optimal sensor allocation decision is proposed to solve the problem of assigning optimal decision-making with multiple sensors under multiple constraints.Firstly,assigning the multi-sensor network by cluseters.Secondly,considering the constraints of communication bandwidth,energy consumption and tracking accuracy,and modeled the benefit function of sensor network is modeled.Finally,solving the objective function to obtain the optimal solution and identify the sensor group with the best system benefit,making the nodes in the group participate in the target tracking to obtain the measurement information,and then data fusion is carried out in the cluster head.Consensus estimation is carried out between the cluster heads and then transmitted to the nodes in the cluster,so that the sensor nodes can maintain the tracking accuracy and balance the cost consumption of each node.