基于GA_BP算法的化工设备设计人工时预测

摘要

针对目前人工估算人工时预测精度低的问题,提出基于遗传算法优化反向传播神经网络的算法建立人工时预测模型,对化工设备设计人工时进行定量预测.首先对化工设备设计项目的管理流程和设计特点进行分析；然后对用户数据库中的历史项目数据进行统计,并对统计数据进行参数的贡献度和相关性分析,同时结合参数在预测时获取的难易程度,选择出适当的人工时预测模型输入参数；再建立基于反向传播算法的预测模型,并针对反向传播算法的缺陷选择遗传算法优化反向传播神经网络和支持向量机算法进行建模,预测结果表明遗传算法优化反向传播神经网络算法更适合化工设备设计人工时预测.采用基于遗传算法优化反向传播神经网络算法的模型进行了实例预测.%Aiming at the low prediction accuracy of man-hour estimation,a man-hour quantitative forecast model based on Genetic Algorithm_Back Propagation (GA_BP) for chemical equipment design was put forward.The management workflow and design features of chemical equipment design project were analyzed,and the historical data in user database were counted.The parameters" contribution and correlation analysis for statistical data were did to select proper input parameters for the model based on the acquirable difficulty levels of parameters during the forecast phase.A man-hour forecast model based on BP neural network was implemented,and the modeling of GA_BP and Support Vector Machine (SVM) algorithm were selected for defect of BP algorithm.The result showed that GA_BP algorithm was more suitable for the man-hour forecast of chemical equipment design.An application case modeled by GA_BP was presented.