Genetic Algorithm Applied Research In The Design Of Pig Feed Formula

The feed formula has significant meaning in poultry raising industry. Thetraditional manual computational method cannot satisfy the need of raw formulamaterial filter and optimized formula at low cost.Now, the feed formula, which results of optimized computation withcomputer, can satisfy the need of poultry nutrition and benefit to select thematerial, also can reduce the cost of feed formula. But the method can onlyresolve the problem which has less limits, in small scale. The method in geneticalgorithm can resolve the difficult problem which the traditional optimizedmethod can' t finish. And we can easily get the closed optimum solution ofthe feed formula by using genetic algorithm.Though the genetic algorithm has been applied usefully in most area, it' sa new subject yet. Its theory and method are immaturity. And the algorithmneeds to be improved and integrated constantly.Our work is about getting feed formula.by applying the genetic algorithmto optimized data. First, we presented the conception and the design methodof feed formula. Then we gave the research way based on genetic algorithm andthe model of getting feed formula. We applied the standard genetic algorithmto the manipulate operation and optimized the standard genetic algorithm. Theadvantage of the optimizing design is that we replaced the traditional roulettewheel model with random league matches in choosing strategy policy, and adoptedthe Elitism. So, we can advance the operation efficiency and ensure theevolution of the choiceness individual.In crossover operator, we used the random numbers to sub vector of parentvectors crossing. Which was enlarging the searching space of child individual,avoiding precocity happened, and also can improve global optimizationperformance.In mutation operator, we mended the average mutation of the genetic algorithm by using gauss mutation. The main area got better searchingperformance by searching close local area of some original individualemphatically. It also covered the shortage of standard genetic algorithm insearching local area.There are two parameters in modifying adaption, Pm and Pc. We used highcrossover probability and mutation probability to the individual with poorperformance, and proper crossover and mutation probabilities to the betterone. Which is benefits to the algorithm convergence while the number ofmultiply generation increasing, and the crossover and mutation probabilitiesare dropping. Avoiding manual estimating Pm and Pc of the genetic algorithm.So we can get the best close optimized result at first one calculation. Theefficiency and easytouse of the algorithm are both improved.We run the program which included the eugenic genetic algorithm, analyzed100 groups test data of feed formula, and compared the genetic algorithm withtraditional mathematics method. It can be concluded that the eugenic geneticalgorithm can help to get preferable result of algorithm performance,efficiency, benefit, precision in feed formula.