摘要
针对单晶硅生产企业在车间生产调度方面的问题,提出了基于BP神经网络和改进遗传算法的单晶硅生产调度算法。该算法在对投料的准确度上进行优化,对初始种群进行子种群划分,采用自适应的变异率和交叉率以及混合交叉方法进行遗传编码的交换。对最小化最大完成时间、最小化交付期逾期时间和最小化总能耗多目标提出了综合适应度计算。最后通过实验对该算法与经典遗传算法进行比较,验证该算法具有较好的性能,并能解决实际问题。
Silicon production scheduling algorithm is proposed which is based on BP neural network and improved genetic algorithm.The accuracy of the feeding amount is optimized by this hybrid algorithm,and initial population are divided into sub-populations,the use of adaptive mutation rate,crossover rate and hybrid crossover genetic coding of the exchange.A comprehensive fitness calculation is presented containing three goals to minimize the maximum completion time,to minimize delivery overdue time period and to minimize the total energy consumption.
出处
《工业控制计算机》
2016年第1期119-120,123,共3页
Industrial Control Computer
关键词
遗传算法
BP神经网络
种群划分
自适应
综合适应度
genetic algorithm
BP neural network
population division
adaptive
comprehensive fitness
