With the increasing development of artificial intelligence and computer-aidedteaching system, algorithms of test paper generation attract more and more attention ofexperts and scholars. From the investigation of intelligent test paper system at home andabroad, most intelligent test paper algorithms are based on the random test paperalgorithm, which is clearly unable to meet the needs of users.In recent years, the genetic algorithm is applied to improve intelligent test paperproblems by scholars, including the convergence of genetic algorithm, crossovermechanism and control parameter determination. Based on a lot of reference literaturesand relevant achievements, in view of the group of roll speed and group roll quality, thisthesis researches the genetic algorithm in application of test paper generation. This workmainly includes:1. Based on the constraint condition analysis for test paper combination, theexamination paper matrix mathematics model is established, Then, the coding methodof storing attribute indexes and adaptive values is proposed. At the same time, to avoidthe decoding process and improve computing efficiency, test questions will betranslated into genetic algorithm genes..2. To avoid the conflict between key points, the piecewise multipoint mutationstrategy for mutation operator and piecewise multipoint cross strategy for cross operatorare proposed to improve the genetic algorithm.3. To maintain the diversity of population, based on the knowledge points, thisthesis produces initial individuals to meet the requirements of the answer time, questiondistribution and knowledge. This method uses the large ratio cross and mutation,improve genetic algorithm convergence speed, and improve group roll speed.4. Based on the above method, the design and implementation of an intelligentsystem is realized. Through the experimental verification, the results shows that thisproposed algorithm for test paper system can well meet the requirements. |