Design and implementation of a multiple-processor-based real-time identification system

Abstract

The authors present the design and implementation of a real-time identification system. A parallelization scheme of the recursive least-squares estimation algorithm has been proposed. The speedup of the proposed parallel recursive algorithm has been analyzed. The parallel recursive estimation algorithm has been implemented on a multiple processor system with hierarchical architecture. Experimental results have shown that the parameters of the identified system can be estimated with acceptable accuracy and within acceptable time delay. The recursive least-squares estimation algorithm has fast convergence for the identification linear system. Using the designed identification system, it is possible to accomplish the update of the identified parameters of a linear second-order system in 2.5 ms. Experimental results have shown the feasibility of implementing a real-time identification system with low-cost microprocessors using parallel processing techniques. >