Nonlinear algorithms for fast and robust control of electrical drives
Several new nonlinear algorithms for speed control of electrical drives are developed. They are compared with the algorithms for integral-proportional (I-P) control, sliding mode control (SLM) and adaptive control which uses the torque and parameter observer. To achieve fast and robust response, all algorithms use very large gains.
In a new, variable limit PI (VLPI) control algorithm, integrator windup is completely prevented by using a high gain, "variable dead zone" nonlinearity as a local feedback over the integrator.
Recently proposed soft variable structure (SVS) control, derived by using the Liapunov direct method, is modified so that the algorithm can be implemented with only the output measurements. Proper operation is achieved for any value of the output variable. The new control is very robust, but exhibits a steady state error.
Two versions of the adaptive PI (API) control algorithm are developed that have fast and robust transient response with zero steady state error. The SVS API version operates similarly as the modified SVS control, but does not have its drawbacks. The SLM API version operates like the SLM control during large transients, and like VLPI control when close to the steady state. The local stability of the control is proved using the "small gain theorem". Its global behavior is analyzed by describing functions.
Very good operation of the SVS API speed control within the proportional position loop is demonstrated. Faster transient response is achieved by implementing the SLM adaptive proportional control in the position loop. The operation is the same as the operation of the SLM API control in the speed loop.
Similarity between modified SVS control, and classical adaptive algorithms is shown. API control, All the algorithms are simulated and compared for twofold and tenfold changes in plant parameters. The experimental verification of the results for I-P control, SLM control, and modified SVS control, are presented.
Theory of the new algorithms is general, such that the results are applicable to any SISO plant that can be stabilized.