基于正交试验的粒子群优化算法对火焰原子吸收光谱法分析金元素参数的优化

Optimization of Parameters for Flame Atomic Absorption Spectrometry Analysis of Gold Based on Particle Swarm Optimization Algorithm Based on Orthogonal Experiment

国内新一轮战略找矿行动全面启动,金矿产资源以其独特的稀有性和战略性具有特殊意义,其分析检测技术直接影响金元素的准确测试。以矿石中金元素为研究对象,采用正交试验设计方案对实验要素中的王水浓度、振荡时间和硫脲浓度进行方法测试,测定结果相对误差为量化指标;按照层次分析法(AHP)中确定要素指标、建立矩阵、一致性判断步骤计算要素权重为(0.252,0.159,0.589),通过客观赋权(CRITIC)法计算正交试验数据的对比强度和冲突性,计算要素权重为(0.452,0.172,0.377),提出基于AHP-CRITIC混合加权算法对要素权重综合分析,其结果为(0.314,0.075,0.611);利用粒子群算法构建粒子多维空间,通过粒子的速度和方向属性迭代位置设计算法流程图,在迭代过程中结合混合加权算法结果通过线性递减的方式校正惯性权重,优化粒子在迭代初期和末期的学习因子,结合正交试验结果利用粒子群算法建立目标适应度函数,改进算法流程,应用MATLAB软件仿真模拟粒子群迭代过程,从全局各位置和方向逐渐向最优组合收敛,得到优化后的粒子群算法寻找原子吸收光谱法分析金元素的最佳条件参数为王水浓度10.62%、振荡时间32.8 min、硫脲浓度9.5 g·L^(-1)。粒子群优化算法验证结果表明,在分析条件最优化参数下对金标准分析物质GAu-15a、GAu-16b、GAu-17b、GAu-18b、GAu-19b、GAu-22a进行11次平行性试验测试,计算其平均值、相对误差和相对标准偏差指标,均满足《地质矿产实验室测试质量管理规范》,表明基于正交试验的粒子群优化算法对于原子吸收光谱法分析金元素参数的优化问题科学可行,验证了该优化算法的正确性和稳定性,对国内新一轮战略找矿事业提供新的研究思路。该方法提出混合加权算法结合进化计算技术对多目标参数寻求最优解,有望拓展于分析实验室其他领域的测试环境,更展望应用于寻求参数优化方向的科学研究中。

A new round of domestic strategic mining search operation is in full swing,gold mineral resources with their unique rarity and strategic with a special significance,its analysis and detection technology affects the accurate testing of gold elements directly.Taking gold element in ore as the research object,the orthogonal test design scheme is used to test the method of aqua regia concentration,oscillation time and thiourea concentration in the experimental elements,and the relative error of determination results is quantified;In accordance with the hierarchical analysis method AHP to determine the element indicators,establish the matrix,consistency judgment steps to calculate the element weights as(0.252,0.159,0.589),calculate the contrast strength and conflict of orthogonal test data through the objective weighting CRITIC method,the element weights are calculated as(0.452,0.172,0.377),and propose the combined analysis of element weights based on AHP-CRITIC hybrid weighting algorithm,the results are(0.314,0.075,0.611);Using particle swarm algorithm to construct particle multidimensional space,design algorithm flowchart by iterative position of particle velocity and direction attributes,combine the results of hybrid weighting algorithm to correct inertia weights by linear decreasing in the iterative process,optimize the learning factor of particles at the beginning and end of iteration,combine the results of orthogonal test to establish the target fitness function using particle swarm algorithm,improve the algorithm flow,applying MATLAB software simulate the iterative process of particle swarm,the optimized particle swarm algorithm is obtained converging to the optimal combination from each global position and direction by gradually,and the optimal condition parameters for finding the gold elements by atomic absorption spectrometry are 10.62%concentration of aqua regia,32.8 min oscillation time,and 9.5 g·L^(-1) concentration of thiourea.The validation results of the particle swarm optimization algorithm show that the gold standard analytes GAu-15a,GAu-16b,GAu-17b,GAu-18b,GAu-19b and GAu-22a have been tested in 11 parallel tests under the optimized parameters of the analytical conditions.The average value,relative error,and relative standard deviation indicatorsare calculated,and all of them satisfied the“Geology and Mineral Laboratory Test Quality Management Specification”.It is shown that the particle swarm optimization algorithm based on an orthogonal test is scientifically feasible for the optimization problem of gold elemental parameters analyzed by atomic absorption spectrometry,and the correctness and stability of the optimization algorithm are verified,which provides new research ideas for the new round of strategic mineral search business in domestic.The method proposes a hybrid weighting algorithm combined with evolutionary computational techniques to find optimal solutions for multi-objective parameters,which is expected to be extended to test environments in other fields of analytical laboratories and more prospective applications in scientific research seeking the direction of parameter optimization.