基于自联想神经网络的测量数据自校正检验方法

SELF-VERIFYING DATA VALIDATION METHOD BASED ON THE AUTOASSOCIATIVE NEURAL NETWORK (AANN)

提出了基于AANN的测量数据自校正检验方法,采用参数预测模型分别对各个参数进行质疑,通过残差决策逻辑选择网络的输入数据,有效地避免了“残差污染”,提高了神经网络方法在线应用的准确率。

A novel data validation method based on autoassociative neural network (AANN) with self-verifying to each parameter is presented. Firstly the single parameter self-verifying model is used within data pre-estimation. Next, the input data of AANN can be selected by the residual decision-making logic. In this way, the accuracy of the online running AANN can be improved greatly.