最优和自校正广义系统信息融合状态估计算法

项目名称： 最优和自校正广义系统信息融合状态估计算法

项目编号： No.61203121

项目类型： 青年科学基金项目

立项/批准年度： 2013

项目学科： 自动化学科

项目作者： 冉陈键

作者单位： 黑龙江大学

项目金额： 24万元

中文摘要： 对于多传感器线性离散定常广义系统，应用加权观测融合算法和协方差交叉融合算法得到相应的广义系统信息融合状态估值器。基于加权观测融合算法的广义系统估值器能得到全局最优的状态估值器，而基于协方差交叉融合的广义系统状态估值器虽然只能得到全局次优的状态估值器，但该方法避免了计算各个局部估值器之间的互协方差，能显著的减少计算负担。对于带未知模型参数和噪声统计的多传感器线性离散随机广义系统，提出了未知模型参数和噪声统计的多段辨识方法。基于这些未知参数估值和最优的多传感器广义系统状态估值器，提出了自校正多传感器信息融合状态估值器。所提出的理论和方法能广泛应用到电力系统、Leontief动态投入产出模型、Hop-field神经网络模型等广义系统的仿真应用研究中。本项目解决了多传感器广义系统最优的和自校正状态估值问题，具有重要理论意义和应用意义。

中文关键词： 广义系统；多传感器信息融合；Kalman滤波；；

英文摘要： For the multisensor linear stochastic descriptor system, applying the weighted measurement fuiosn method and the corvariance intersection fusion method, the information fusion state estiamtors of descriptor system are presented. The estimator based on the weighted measurement fusion algorithm is globally optimal, while the estimator based on the corvariance intersection fusion algorithm is globally sub-optimal. They avoid to compute the the cross-variance of the local estimators in order to reduce the computational burden. For the multisensor linear stochastic descriptor system with unknown model parameters and unknown noise variances, the multi-stage information fusion system identification method is presented. Based on these estimates and the optimal information fusion state estimator of the descriptor system, the self-tuning information fusion state estimator is presented. These presented theory and method can be applied to simulation application and research of descriptor system, such as power sytem, Leontief dynamic input-output model,Hop-field neural network model. This project solves the optimal and self-tuning estimation problem of the multisensor descriptor system, and has important theory and application mean.

英文关键词： descriptor system；multisensor information fusion；Kalman filtering；；