论文部分内容阅读
为降低目标性能对工业决策参数因不确定性而变化的敏感性,实现即稳健又靠近理想性能的目标,提出了一种工业过程决策参数的稳健优化方法。首先用神经网络建立复杂工业过程模型,确定决策参数与目标性能之间的映射关系;再用模型的目标性能均方差构建稳健准则,通过稳健准则量化目标的稳健性;然后将稳健准则与目标性能作为两个目标,构造出多目标稳健优化模型,利用具有多体交叉和大搜索范围的改进的强度Pareto进化算法(Improve Strength Pareto Evolutionary Algorithm,ISPEA)对多目标稳健优化模型进行搜索,搜索稳健性和目标性能最好的解,据此对实际生产进行指导。通过对氢氰酸(HCN)生产工艺的仿真验证,显示了该方法的有效性。
In order to reduce the sensitivity of objective decision-making to the uncertainty of industrial decision-making parameters and achieve the objective of being robust and close to the ideal performance, a robust optimization method for industrial process decision-making parameters is proposed. Firstly, a complex industrial process model is established by neural network to determine the mapping relationship between the decision parameters and the target performance. Then the robust criterion is constructed by the mean square error of the target performance of the model, and the robustness of the target is quantified by the robust criterion. Then, the robust criterion is compared with the target performance As the two objectives, a multi-objective robust optimization model is constructed, and the improved robust Pareto Evolutionary Algorithm (ISPEA) with multi-body crossings and large search range is used to search for robust multi-objective optimization models. And the best performance of the target solution, based on which the actual production guidance. Through the simulation of hydrocyanic acid (HCN) production process, it shows the effectiveness of the method.