摘要
针对目前不同生产过程开发不同的建模和优化软件,但软件之间在一定程度上存在重复的现象,为了减少重复开发,设计了集成多种建模方法和智能优化方法的通用软件包的原型机OptSoftWare,并实现了对用户输入表达式的自解析和对约束条件的处理。结合SoftWare从建模、约束条件的处理、表达式自解析和优化这一整个流程做了详细介绍。该软件已成功应用于某冶炼厂锌电解过程分时供电决策系统,模型的预测作用精度达到98%,优化调度后使其电解过程的电耗降低为原来的94%。
The problem of repetition in developing different software of modeling and optimization for different industrial process is discussed. The prototype of universal software package is designed named OptSoftWare, which intergrated many kinds of methods in modeling and optimization, and reahzed self-analysis and dealing with restriction by itself. According to OptSoftWare, the flow from modeling to optimization is presented. The software is applied to the decision-making system for optimizing time-distributed power supply during electrolysis of zinc. The precision of prediction is up to 98 %, and the electricity of electrolysis is reduced to 94 % after optimization.
出处
《控制工程》
CSCD
2005年第5期401-404,共4页
Control Engineering of China
基金
国家973重点基础研究资助项目(2002CB312200)
关键词
智能优化
建模
神经网络
自解析
intelligent optimization
modeling
neural network
serf-analysis
