Combine GP With PSO Algorithm To Realize The Evolutionary Design Of Active Filter

Genetic Programming (GP) is a branch of evolutionary algorithms. It is based on theprinciple of Darwin's survival of the fittest. It starts from a population of the computer programswhich describe the solutions of a problem, and the solutions are evolved by simulating thenatural evolutionary process until the best solution is finded. Because the operate targets of GPare hierarchical structure computer programs whose size and shape can change dynamically,make it in hardware circuits'representation, means in circuits'encode mode, become a goodmethod relatively.Particle Swarm Optimization (PSO) is a random evolutionary computation method basedon a group of particles'intelligent movement. It is derived from the behavior study of the birdpopulations'prey on. Similar to GP, PSO algorithm is an optimization tool which based oniteration; the system initializes to a group of random solution and searches the optimal valuethrough iteration. But PSO algorithm hasn't the obvious crossover and mutation used by GP; itis the search procedure that the particles follow the optimal particle in solution space. Preciselybecause this point, make it has more advantages in parameters optimization.With the development of modern electronic circuit technology, analog filter has become abasic module in high-precision large-scale integrated circuits. The paper combined GP and PSOalgorithm to design the active filter by analysis the idea and realization way of them: first, usesthe GP to evolve the circuit organization of the active filter; second, uses the PSO algorithm tooptimize the parameters of the filter circuit designed by GP, and obtain the final active filterelectric circuit. The paper studies the improvedH measureH of GP stressly, proposes a newindividual representation approach which based on HmoduleH, and establishes correspondingcrossover and mutation etc. genetic operation rules and the fitness evaluation methods too. Thismethod is effective in overcoming the large amount of redundancy in the calculation andreusability is small etc. shortcomings in individual representation approach of existing GP at theprocess of electric circuit evolution. This method also improves the algorithm optimizationcapabilities and convergence rate. We take the evolutionary design of various kinds of two-orderactive filter and the four-order state variable filter as instances, give the method and the step ofevolutionary design on combine GP with PSO algorithm to realize the evolutionary design ofactive filter. The evolved active flter were simulated with Multisim 8 software. The resultsshowed that the design of the active flter was satisfied with specification. Programs in GP andPSO algorithm takes C and C ++ language respectively, they all were executed with VC ++6.0.