An on-line technique for system identification

In most practical applications, dynamic systems can be modeled by differential equations with prescribed forms but unknown parameters. An analog method is developed for identifying these unknown parameters. The method is based on the learning model concept. Therefore, it is capable of performing on-line identification in the presence of noise. It has shown that single-variable and multivariable linear systems can be easily identified. Also the same algorithm is applicable to zero memory nonlinear systems. Hence the proposed method is regarded as very useful Various examples simulated on a digital computer are presented and the results are good.