基于QPSO-BP神经网络的数学学科质量评价模型  

A Mathematics Subject Quality Evaluation Model Based on QPSO-BP Neural Network

作　　者：李刚[1] LI Gang(Department Of Public Teaching,Chuzhou City Vocational College,Chuzhou 239000,China)

机构地区：[1]滁州城市职业学院公共教学部,安徽滁州239000

出　　处：《西安航空学院学报》2024年第3期77-82,88,共7页Journal of Xi’an Aeronautical Institute

基　　金：滁州城市职业学院科研重点项目(2023skzd03);滁州城市职业学院质量工程重点项目(2022zdyxm14);安徽省职业与成人学会一般项目(AZCJ2023182)。

摘　　要：为降低BP神经网络初始权值和阈值随机选取导致的评价误差,在BP神经网络中融合QPSO算法构建数学学科质量评价模型。以19个学科质量评价二级指标为范围,基于主成分分析法提取关键指标成分,并计算二级指标贡献率,数据降维后选出累计贡献率不低于85%的指标,输入BP神经网络模型;采用QPSO算法优化BP神经网络初始权值和阈值,更新了粒子位置,考虑了当前粒子局部最优位置与全局最优位置,引入“粒子平均最优位置”,强化了粒子之间的相互作用,同时利用权重系数平衡了粒子收敛能力;由此构建QPSO-BP数学学科质量评价模型,可将数学学科质量评价的效果划分为优秀、良好、中等、较差4个等级。实验结果显示,融合QPSO算法的数学学科质量评价模型可将累计贡献率达到85%的指标保留下来,且评价误差均低于预设误差0.01。该模型收敛性能较好,得出的数学学科质量评价结果符合实际情况,避免人为主观随意性,为数学学科建设提供了有效的质量反馈。In order to reduce the evaluation error caused by the random selection of initial weights and thresholds in the BP neural network,a mathematics discipline quality evaluation model is constructed by integrating the QPSO algorithm in the BP neural network.Based on the principal component analysis,key indicator components are extracted from 19 secondary indicators for quality evaluation,and the contribution rate of secondary indicators is calculated.After data dimensionality reduction,indicators with a cumulative contribution rate of no less than 85%are selected and input into the BP neural network model.The QPSO algorithm is used to optimize the initial weights and thresholds of the BP neural network,update the particle position,the local optimal position and global optimal position of the current particle are considered,and the“particle average optimal position”is introduced to strengthen the interaction between particles,and the weight coefficient was used to balance the particle convergence ability.Therefore,the QPSO-BP mathematics subject quality evaluation model is constructed,and the effectiveness of mathematical discipline quality evaluation can be divided into four levels:excellent,good,moderate,and poor.The experimental results show that the mathematical subject quality evaluation model integrating QPSO algorithm can retain indicators with a cumulative contribution rate of 85%,and the evaluation errors are all lower than the preset error of 0.01.The model has good convergence performance and the quality evaluation results of the mathematical discipline are in line with the actual situation,avoiding subjective randomness and providing effective quality feedback for the construction of the mathematical discipline.

关 键 词：QPSO算法 BP神经网络 学科质量 权值 阈值 

分 类 号：O13[理学—数学] G420[理学—基础数学]