Improvement And Application Of Genetic Programming

Evolutionary algorithms follow the natural law of "survival of the fittest", is a kind of the global optimization techniques. The character of the evolutionary algorithms is that the optimal solutions come from a set of random candidate solutions, the random candidate solutions is operated by the operations such as replication, crossover and mutation, approximating the optimal solution step by step. In the 20th century, some major evolutionary algorithms such as genetic algorithms, genetic programming, have been proposed. Genetic algorithms use fixed-length strings to express the problem. Because of the complexity of the nature, some problems can not be expressed by the fixed-length strings. With the improvement of the expression of genetic algorithms, genetic programming algorithm comes into being. Genetic programming uses a flexible hierarchy structure (tree) to describe the problem. The structure and size of the tree is adjusted dynamically, so it is more suitable to express the complex problems, and more widely applied.The basic theory of evolutionary algorithms, including the basic idea, the introduction and present status of the branches, is introduced in the first chapter. The second chapter of this article describes the basic idea and shortage of genetic programming algorithm, also gives a deeply analysis of the related theories of genetic programming, including the methods of individual production, fitness measures and genetic operators. Given the lack of genetic programming algorithm, the third chapter expresses the method how to improve the efficiency of the traditional genetic programming algorithm. On the one hand, the speed of optimization is improved by the improved algorithm significantly. On the other hand, the complexity of the best individual has been effectively controlled. Based on self-improvement of genetic programming algorithm, the niching method which is used in genetic algorithm, is used for programming algorithm ,proposing the niche genetic programming algorithm , and verify the effectiveness of the algorithm in the symbolic regression. NGP algorithm searches for the better solution from the first optimization, and gradually approximates to the best optimization until it meets the precision requirement of the problem . Then the improved genetic programming algorithm is applied to the credit evaluation of the scientific research institutions in order to establish a scientific evaluation model. The usefulness and correctness of the model is proved by the experiments later.