About estimation of the signal prediction

Abstract

This paper discusses a method for calculating a signal prediction estimate using an automatic realizable predictor. Such problems arise both in the theory of automatic control and in various applications where it is required to obtain a forecast for the observed implementation. The relevance of the study is determined by the fact that when finding a forecast of a certain signal for the lead time, the time of calculating the forecast can be significant. If the prediction calculation time is longer than the lead time, then the resulting prediction is meaningless. Therefore, when predicting, one should strive to ensure that the calculation time of the forecast is much less than the lead time. In this case, there will be a margin of time for decision making. The method proposed in the article under consideration makes it possible to rationally organize the computational prediction scheme and reduce the forecast calculation time. The article obtained an explicit calculation formula for calculating the estimate of the forecast for a given input signal. The formula is reduced to calculating the weighted integrals of the input signal at each time step. This article proposes a method of reducing the number of integration operations at each step. This allows you to reduce the calculation time of the forecast while maintaining accuracy. The proposed method can be used to both continuous (analog) and discrete signals.