Research On Kalman Fusion Method In Multi-sensor System

In the twenty-one century, with the high development of communication technology, sensor technology and computer technology, promote the development of wireless sensor network technology. Benefitting from the advantage of less wiring, strong self-organizing capacity and low cost.etc, Wireless sensor networks are widely used in agriculture, industry, medicine, military.etc, has broad application prospects and promising potential.Inevitably, with the expansion of the application scope of wireless sensor networks, some shortcomings are gradually revealed, such as limited communication capabilities, inadequate energy supply, limited bandwidth, and so on. These will simply cause the phenomenon of delay, out-of-sequence and loss when the information transmitted in wireless sensor networks, leading to the traditional information process methods are no longer adapt to deal with. Therefore, research on new information process technologies to solve the phenomenon of delay, out-of-sequence and loss will become a hot issue. Under the hot research issue, this paper mainly focus on the part of the phenomenon of delay in wireless sensor network start the work, the main contents are as follows:(1) It briefly introduces the state estimation theory, the derivation of Kalman filtering algorithms and several representative information fusion algorithms, among them; it mainly introduces the centralized fusion algorithm, sequential fusion algorithm, and matrix fusion algorithm.(2) It proposed a matrix weighted fusion algorithm based on local pseudo-measurement model library. First, assume the local sensor node can process the data, and the phenomenon of delay may occur at the time of data collection; then contrary to the above situation, setting a certain number of delay steps, to establish a local pseudo-measurement model library, processing the filtering based on pseudo-measurement model local library to get the local state estimates; at last, when the local state estimates reach the fusion center on time and orderly, using the matrix fusion method to achieve the state estimates, the sstimation accuracy is higher than the single-sensor system.(3) It proposes centralized and sequential delay filter algorithms, first contrary to the situation of delay when the original data transmit to the fusion center, setting a certain number of delay steps, to establish a global pseudo-measurement model library; then it derives the centralized filter algorithm and sequential filter algorithm based on the pseudo-measurement model, in order to obtain the global state estimate;finally, simulations verify the optimality of the two methods.