论文部分内容阅读
本文给出了对一个线性动态系统的任何两个参数向量估计值进行微型机自适应检验,作出优劣判断的方法。对于未知参数向量的两个估计值,最初无法判断哪一个估计值比较接近真值,而只能认为未知参数向量等于这两个估计值的概率是相等的。但是随着观测数据的获得,在相同的观测数据的条件下,未知参数等于两个估计值之一的条件概率将不相等。随着观测数据的增加,未知参数等于比较接近真值的那个估计值的条件概率将越来越明显地大于另一个条件概率。这就是本文方法的要点。利用这种方法可以进行最优参数估计值的搜索,形成一个在线参数估计算法。这种算法计算量不大,不仅所得结果的准确性和方法的强壮性很突出,而且还可以把噪声的均值和方差也作为未知参数来估计。它的另一个优点是进行在线参数估计时同时获得的状态估计值可以直接用于最优控制算法。
In this paper, we give a method to judge the fitness of any two parameter vector of a linear dynamic system by the microcomputer adaptive test. For two estimated values of the unknown parameter vector, it is not possible to determine which one is closer to the true value at the outset, but can only consider the probability that the unknown parameter vector equals the two estimated values are equal. However, with the observation data obtained, under the same observation data, the conditional probability that the unknown parameter is equal to one of the two estimated values will not be equal. As the observed data increases, the conditional probability that an unknown parameter is equal to that estimate that is closer to the true value will be significantly greater than the other conditional probability. This is the point of the method of this article. This method can be used to search the optimal parameter estimation to form an online parameter estimation algorithm. This algorithm is not computationally intensive, not only the accuracy of the results obtained and the robustness of the method are prominent, but also the mean and variance of the noise can be estimated as unknown parameters. Another advantage of this is that the simultaneous estimation of the state values obtained during online parameter estimation can be used directly in the optimal control algorithm.