摘要
利用三层误差反向传播(back propagation,BP)神经网络建立磨削能耗预测模型,以砂轮线速度、进给速度和磨削深度为影响因素设计125组全因子试验,并取其中的75组试验数据作为该预测模型的训练样本与测试样本。采用动态惯性权重改进粒子群算法(adaption particle swarm optimization,APSO),以BP神经网络的预测作为适应度函数,以最小能耗为目标进行迭代寻优获取最优工艺参数。结果表明:模型预测结果较为准确,采用优化后的工艺参数能够有效降低磨削能耗。
A three-layer back propagation(BP)neural network was used to establish a grinding energy consumption prediction model.125 single-factor experiments were designed with the grinding wheel linear velocity,feed rate and grinding depth of cut as the influencing factors.75 sets of experimental data were obtained as the training samples and test samples of the prediction model.Particle swarm optimization algorithm was improved by using adaptive dynamic inertia weight(adaption particle swarm optimization,APSO),and the prediction of BP neural network was used as fitness function.The optimal process parameters were obtained by iterative optimization aiming at minimum energy consumption.The results show that the prediction model is accurate and the optimized process parameters can effectively reduce the grinding energy consumption.
作者
张昆
田业冰
丛建臣
刘俨后
闫宁
鲁涛
ZHANG Kun;TIAN Yebing;CONG Jianchen;LIU Yanhou;YAN Ning;LU Tao(School of Mechanical Engineering, Shandong University of Technology, Zibo 255049, Shandong, China;Tianrun Industry Technology Co. , Ltd. , Weihai 264400, Shandong, China;Zhengzhou Research Institute for Abrasives & Grinding Co. , Ltd. , Zhengzhou 450001, China)
出处
《金刚石与磨料磨具工程》
CAS
北大核心
2021年第1期71-75,共5页
Diamond & Abrasives Engineering
基金
国家自然科学基金资助项目(51875329)
山东省泰山学者工程专项(tsqn201812064)
山东省自然科学基金资助项目(ZR2017MEE050)
山东省重点研发计划资助项目(2018GGX103008,2019GGX104073)
山东省高等学校青创科技项目(J17KA037)
淄博市重点研发计划项目(2019ZBXC070)。
关键词
改进粒子群算法
BP神经网络
磨削能耗
参数优化
adaption particle swarm optimization
BP neural network
energy consumption of grinding
parameter optimization