摘要
教学质量评价是教学研究中的重点之一,但已有的数学评价模型不适合解决非线性问题,神经网络模型收敛速度慢、准确率不高。针对以上问题,文中提出一种基于改进PSO(Particle Swarm Optimization)-BP(Back Propagation)神经网络的教学质量评价模型。通过引入动量和自适应学习率优化BP神经网络,采用惯性权重线性递减、学习因子异步变化,并引入速度收缩因子和自适应变异策略来优化PSO算法;再使用PSO粒子群优化算法计算BP神经网络的初始连接权重和阈值,从而提升模型的全局寻优能力和收敛速度、精度。为验证模型效果,使用评价体系指标层的10个指标数据作为模型的输入,评价结果作为输出,进行模型对比实验。实验结果表明,所提模型的准确率达到96.33%,比一般BP神经网络模型提高4.68%,比自适应BP神经网络模型提高4.07%,比PSO-BP神经网络模型提高1.2%,且收敛曲线平稳,整体性能优于其他模型,说明运用该模型能够有效地对教学质量进行评价。
Teaching quality evaluation is one of the key points in teaching research,but existing mathematical evaluation models are not suitable for solving nonlinear problems,and neural network models have slow convergence speed and low accuracy.In response to the above issues,a teaching quality evaluation model based on improved PSO(particle swarm optimization)⁃BP(back propagation)neural network is proposed.By introducing momentum and adaptive learning rate to optimize the BP neural network,the inertia weight decreases linearly,the learning factor changes asynchronously,and the velocity contraction factor and adaptive mutation strategy are introduced to optimize the PSO algorithm.PSO particle swarm optimization is used to calculate the initial connection weight and threshold of BP neural network,so as to improve the global optimization ability,convergence speed and accuracy of the model.In order to verify the effectiveness of the model,10 indicator data from the evaluation system indicator layer were used as inputs to the model,and the evaluation results were used as outputs for model comparison experiments.The experimental results show that the accuracy of the proposed model is 96.33%,which is 4.68%higher than the standard BP neural network model,4.07%higher than the adaptive BP neural network model,and 1.2%higher than the PSO⁃BP neural network model.The convergence curve is smooth,and the overall performance is better than other models,indicating that the use of this model can effectively evaluate teaching quality.
作者
郭欣
殷子龙
陈瑛
吴玉佳
GUO Xin;YIN Zilong;CHEN Ying;WU Yujia(School of Information Science and Technology,Sanda University,Shanghai 201209,China;Department of Electronic and Electrical Engineering,Shanghai University of Engineering Science,Shanghai 201620,China)
出处
《现代电子技术》
2023年第12期146-152,共7页
Modern Electronics Technique
基金
上海市自然科学基金资助项目:基于字符信息抽取的文本分类方法研究(22ZR1445000)
上海市高等教育学会2021年规划研究重点课题:基于大数据技术与模糊综合评价理论的高校教学质量评价体系与方法研究(Z1-07)
上海杉达学院校基金项目(2021BSZX06
2021YB14)
上海杉达学院校级课程建设项目:大数据技术(A020201.21.314)。
关键词
粒子群优化算法
BP神经网络
教学质量评价
自适应变异策略
连接权重
性能对比
PSO algorithm
BP neural network
teaching quality evaluation
adaptive mutation strategy
connection weight
performance comparison
