Information Fusion Prediction Estimation Theory&Transportation Navigation Application Research

As the controlled object scale and complexity increasing, it put higher requirement to the target state prediction accuracy and robustness. However, the traditional mono-source information is difficult to meet these requirements. Information fusion method can synthesize same or different types of information, complement each other, eliminate information redundancy and contradiction, and finally obtain satisfactory results. At present, most research are based on Kalman filter theory, but Kalman filter assumes the noise statistical characteristics is known, and the above premise may damage the filter output, finally, restricted the filters filter performance. In the real world, affected by the internal system and external disturbance complex environment influence, the filter system state equation and observation matrix will inevitably exit stochastic uncertainty and the noise statistical characteristics can not be priory known. Therefore, information fusion estimation theory and its transportation navigation application research have important theoretical significance and practical value.At the beginning, the thesis introduces the information fusion estimation theory and application related studies both at home and abroad, and then summarizes the still existing problems. To solve the information system existing parameters uncertainty and noise characteristic unknown problems, based on Hco filter theory and use LMI design method, we propose a distributed uncertain source information fusion estimates more than forecast filter design theorem, the simulation research validated the effectiveness of the proposed method and we also use real-time measurement data to analyze the method application in traffic navigation area. To meet the aviation target tracking high accuracy and robustness requirements, we propose a Hoo based aviation elevation tracking algorithm. To fulfill the bus arrival time and bus passenger flow prediction accuracy and timeliness requirements, based on Jinan city public transport ASA and IC data, apply information fusion theory, we propose a SVM and Hoo filter based dynamic bus arrival time prediction method and also a passenger flow forecast method incorporating AR,SARIMA,ARIMA and IMM models. By comparing with related studies validate our proposed method effectiveness.