论文部分内容阅读
采用改进的交互式遗传算法,设计了一种烤漆门智能选型系统。在大规模进化种群前提下,该算法依据用户少数次对进化个体的评价,通过K均值聚类生成聚类中心,对个体适应值按距离聚类中心的距离估计,以机器评价代替人对个体适应值评价。对于该方法,本文给出了评价效果的评价指标,聚类估计进化个体适应值与人的评价之间的转换策略,并分析了算法的性能。该算法克服了交互式遗传算法种群规模小,全局搜索能力低的缺点。将该系统投入市场使用,结果表明能保证用户能找到满意的设计方案,大幅提高了选型效率,优化质量优于传统的交互式遗传算法。
An improved interactive genetic algorithm is used to design a kind of paint door intelligent selection system. On the premise of large-scale population evolution, this algorithm based on the user’s evaluation of evolutionary individuals a few times, generates clustering centers by K-means clustering, estimates individual fitness by distance from clustering centers, Evaluation of fitness. For this method, this paper gives the evaluation index of evaluation effect, the clustering method to estimate the conversion strategy between evolutionary individual fitness and human evaluation, and analyzes the performance of the algorithm. The algorithm overcomes the shortcomings of the small population of interactive genetic algorithm and low global search ability. The system put into market use, the results show that users can find a satisfactory design, greatly improve the selection efficiency, optimization of quality better than the traditional interactive genetic algorithm.