Research On Optimization Algorithm Of Evolutionary Design Of Digital Circuit

The research on Evolvable Hardware that inspired by the essence of the biologicalevolution shows a new circuit design method, which can imitate the biological evolution toconfigurate the interior structure by evolutionary computing techniques to obtain the specificfunction. As an important branch of the Evolvable Hardware research, evolutionary design ofcircuits has incomparable advantages over conventional design, which can use the ability ofcombinatorial optimization and global search of evolutionary algorithms to complete theautomatic design, self-adaptive and fault-tolerant capability of complex circuit. With the scaleof electronic circuit system continuously increasing, the convergence speed and scalebottleneck of evolutionary design of circuits appeared. To solve the scale and speed ofevolutionary design of circuits effectively is open problems to be solved in current domesticand international.The optimization algorithms of evolutionary design of digital circuits are studied in thispaper. The traditional genetic algorithm was optimized through improving and innovating thegenetic algorithm, and a new kind of modular evolutionary method based on the ideaof genetic algorithm was proposed. The main research contents of this paper are as follows:(1) The paper explored new fitness evaluation method and genetic strategies on thebasis of the original genetic algorithm. The improved method includes some optimizationmethods of fitness sharing, exponential weighting，double selection population,"Queen bee"mating and module crossover. The experiment showed that the new evolutionary algorithmaccelerated evolution convergence greatly, improved the success rate effectively andoptimized the evolutionary design of circuits obviously.(2)Modular Evolutionary Algorithm was proposed in order to solve the problems thattraditional genetic algorithm present in convergence difficulties and a low success rate inlarge scale and complicated digital circuit problems. The Modular Evolutionary Algorithmcan protect good gene rfagment through packaging the local solution and subcircuits formodule, which made the complex circuit be simplified.(3)Considering the particularity of circuit form，the genotype coding uses real numberand makes the nodes as basic unit in chart form. The genetic operation and evolution process of Modular Evolutionary Algorithm was studied and the experiment was designed with adderand multiplier as an example. Compared with the traditional genetic algorithm, ModularEvolutionary Algorithm has an obvious advantage in complex circuit evolutionary design,which completed the evolutionary design of complex circuit successfully.