Abstract:

A testing methodology for the system maximum-error specification based on the timing series model is presented. It can provide a theoretic basis for the testings of maximum-error specification on long working and strong correlation systems in some special fields, such as aviation, astronautics and national defenses. Firstly, a proper timing series model is selected, which can model the error series of system outputs and trace the practical system outputs on the integral trend. After taking both the model residuals and predicting outputs into account, the system maximum-error specification is tested and calculated by using the classic statistical theories. To introduce the methodology more directly, a practical sample which is based on an autoregressive integrated moving average timing series modeling methodology is given. The test results show that this methodology is feasible.