Structural Optimization Of Switched Reluctance Motors By Using Improved Genetic Algorithms

The genetic algorithm is a intelligent evolutionary algorithm in the moderncomputer science and optimization theory with widespread concern. In practicalapplications, the basic genetic algorithm has many disadvantages. For the shortcomingof slow convergent speed and poor stability of the simple genetic algorithm, animproved algorithm which combined with the PL homotopy algorithm was proposed inthis paper. The PL homotopy algorithm is efficient algorithm for solving the fixed pointproblem with fast convergence and strong local search ability. While, the geneticalgorithm is lack of local search ability, the hybrid algorithm of both algorithms willreceive good effect. To use the improved genetic algorithm for function optimization,the optimal solution problem should be transformed into a fixed point problem firstly.The improved algorithm puts the function optimization problem of n-dimensional intoan n+1dimensional space, sets up tracking path by the changes of homotopy parameter,does shaft operation in accordance with the vertex information until the optimalsolution be found. Simultaneously, use the simplex subdivision with refining networkdiameter, so that the accuracy of the target solution to enhance under the premise of noloss of efficiency of the algorithm. The algorithm uses the vertex integer labelinformation of simplex subdivision as convergence criteria. Typical test function beused to verify the improved genetic algorithm is efficiency and high accuracy.Switched reluctance motors, which have lower cost manufacturing, higher systemreliability, larger speed range and more advantages in control flexibility than ordinarymotors, have important research value and application prospect. However, to somecertain extent, the radial electromagnetic force causes a large output torque ripple andacoustic noise, which become a key factor restricting its development. So, it is greatsignificant that a precise mathematical model be established and the relevant variablesbe optimized. With a2.2kw power for the prototype as an example, using the model tooptimization analysis, and compares with the original motor parameters. The resultsshow that the improved algorithm has better convergence and high efficient.