基于PSO-GA-ANN的铁矿采选品位与投资策略优化

Optimization of Iron Ore Grade Selection and Investment Strategies in Iron Mines Based on PSO-GA-ANN

在构建非线性多目标优化模型的基础上,集成粒子群(PSO)、遗传算法(GA)与人工神经网络(ANN)等智能优化算法,来确定实现铁矿资源减损和节能降耗的采选品位与投资策略。根据采选生产流程,建立以采选品位与投资策略为决策变量,计划精矿产量为约束条件,资源利用率、总电耗以及经济效益为目标的非线性多目标约束优化模型。将PSO、GA的高效搜索能力和ANN的建模功能相结合,构成PSOGA-ANN算法来寻找最优采选品位和投资策略。在PSO-GA-ANN算法中,进化个体采用二进制0-1编码和实数编码相结合,适应度函数为3个目标函数的加权和,各个目标的权重采用均匀设计的思想得到,采用基于可行性规则的约束处理技术引导搜索方向。最后,以D铁矿为例进行了研究,得到了其最优采选品位及投资策略。该方法为新时期铁矿应对资源利用和节能降耗难题提供了科学可行的思路。

Based on the nonlinear multi-objective optimization model,the particle swarm optimization（PSO）,the genetic algorithm（GA）,and the artificial neural network（ANN）were integrated to determine iron ore grade selection and investment strategies to enhance the resource utilization rate and reduce the energy consumption in iron mines.First,according to the production process,the nonlinear constrained multi-objective optimization model was established,with iron ore grade selection and investment strategies as decision variables,with expected concentrate output as constraint,and with resource utilization rate,total power consumption,and economic benefit as goals.Then,the PSO,the GA,and the ANN were combined to form a PSO-GA-ANN algorithm to find out the optimal grade combinations and investment strategies,In the PSO-GA-ANN algorithm,evolutionary individuals contain 0-1 binary coding and real coding,the fitness function was the weighted sum of three objective functions,and the weight value of each objective was obtained by using the uniform design method.The constraint processing technique of feasibility rules was used to lead the search process.Finally,taking D iron mine as an example,the optimal iron ore grade selection and investment strategies were obtained.The proposed method provides a scientific and feasible idea for efficient utilization of resources and energy conservation in current iron mines.