带未知参数系统的多传感器多新息卡尔曼滤波器

发布时间：2018-12-17 23:53

【摘要】：多新息辨识理论是将新息的修正技术从单新息扩展到了新息向量,从而改善参数估计精度的有用信息。当系统输入不可信数据时,该算法能跳过坏数据,避开坏数据和损失数据对系统参数估计的影响,得到更精确的参数估计。为了进一步增强系统的容错性和准确性,我们将多新息辨识理论和多传感器融合技术相结合。多传感器信息融合是将来自多个传感器的信息进行合成,形成一种对被测对象某一特征的表达式。信息经过融合后比单一传感器的信息更完善、准确,从而获得系统状态的最优融合估计。本文应用多新息辨识理论和多传感器信息融合技术,对未知参数系统的研究主要包括以下几方面:首先,对带有未知参数的状态空间模型,给出了估计未知参数的多新息最小二乘方法和多新息随机梯度方法等。运用基于状态空间模型与ARMA模型之间的转换,将状态分量估计问题转化为信号的估计问题,并利用多新息的辨识方法,得到系统的参数估值。其次,对带有未知参数系统,给出了自校正多新息Kalman状态估值器。首先运用多新息辨识理论,得到系统的未知模型参数估值,再将估出来的估值代入到Kalman滤波器中,从而得到带未知参数系统的自校正多新息Kalman滤波器。最后,对带未知参数的多传感器系统,给出了自校正多传感器多新息Kalman状态估值器。首先运用多新息辨识方法得到系统模型参数的估值,再基于集中式融合、矩阵加权融合和CI融合方法,从而得到带未知参数系统的自校正多传感器多新息Kalman滤波器。通过Matlab仿真,验证了估值器的有效性。
[Abstract]:The theory of multi-innovation identification is a useful information to improve the accuracy of parameter estimation by extending the modified technique of innovation from single innovation to innovation vector. When the system input untrusted data, the algorithm can skip the bad data, avoid the influence of bad data and lost data on the system parameter estimation, and get more accurate parameter estimation. In order to further enhance the fault tolerance and accuracy of the system, we combine multi-innovation identification theory with multi-sensor fusion technology. Multi-sensor information fusion is a kind of expression that combines the information from multiple sensors to form a feature of the object under test. The information fusion is more perfect and accurate than that of a single sensor, and the optimal fusion estimation of system state is obtained. In this paper, using the theory of multi-innovation identification and multi-sensor information fusion, the research of unknown parameter system mainly includes the following aspects: firstly, the state space model with unknown parameters is studied. The multi-innovation least square method and the multi-innovation random gradient method for estimating unknown parameters are given. Based on the transformation between state space model and ARMA model, the state component estimation problem is transformed into the signal estimation problem, and the parameter estimation of the system is obtained by using the multi-innovation identification method. Secondly, a self-tuning multi-innovation Kalman state estimator is given for the system with unknown parameters. First, using the theory of multi-innovation identification, the parameter estimates of the unknown model of the system are obtained, and then the estimated estimates are added to the Kalman filter to obtain the self-tuning multi-innovation Kalman filter for the system with unknown parameters. Finally, a self-tuning multi-sensor multi-innovation Kalman state estimator is presented for multi-sensor systems with unknown parameters. The estimation of system model parameters is obtained by using multi-innovation identification method, and then based on centralized fusion, matrix weighted fusion and CI fusion, a self-tuning multi-sensor multi-innovation Kalman filter with unknown parameters is obtained. The effectiveness of the estimator is verified by Matlab simulation.
【学位授予单位】：黑龙江大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TP212;TN713

【共引文献】