Model reference adaptive control-implementation considerations on an integer-based DSP

Model reference adaptive control (MRAC) techniques may be used in high-performance applications of induction motor drives to minimize adverse effects from changes in the load conditions and/or system parameters. Although the MRAC technique accounts for uncertainties and/or inaccuracies of the motor and load parameters in the design stage, its implementation on an integer-based DSP has several difficulties associated with the large dynamic range of the covariance matrix and the finite length of the DSP word. This paper addresses the solution to these difficulties using double-word fixed-point algorithms which increase the dynamic range of the data and concurrently reduces rounding and/or truncation errors. The unavoidable disadvantage is a longer execution time. First, the general MRAC technique is briefly discussed for reference. Double-word fixed-point addition, subtraction, multiplication and division are then described in detail. Finally, experimental results are presented to illustrate the proposed ideas when controlling the speed of a 2 hp induction motor drive using the proposed double-word fixed-point MRAC technique for the speed controller.