Self-learning control of load changes in motordriven load simulator using CMAC

How to retain the high load precision of a motor-driven load simulator in the case of great change in load gradient is one of its key problems. In the past, the compound PID control method was used to improve its load precision. However, because of the influence of its time-varying character and non-linearity, the method does not produce ideal load speed or precision. Taking the characteristics of the load simulator into account, the paper applies the CMAC neural-network control structure to the load simulator and presents its control structure and algorithm. The analysis of the experimental results, given in Figs. 5 and 6 and Table 2, indicates preliminarily that our method overcomes the shortcomings of the sole use of PID control method and satisfies the requirements for high-precision in the case of great changes in load gradient.