The fraction defective of semi-finished products is predicted to optimize the process of relay production lines, by which production quality and productivity are increased, and the costs are decreased. The process par...The fraction defective of semi-finished products is predicted to optimize the process of relay production lines, by which production quality and productivity are increased, and the costs are decreased. The process parameters of relay production lines are studied based on the long-and-short-term memory network. Then, the Keras deep learning framework is utilized to build up a short-term relay quality prediction algorithm for the semi-finished product. A simulation model is used to study prediction algorithm. The simulation results show that the average prediction absolute error of the fraction is less than 5%. This work displays great application potential in the relay production lines.展开更多
各行业规模化产量分析对产能建设和生产计划调度有着重要的指导意义。各行业生产产量数据为时间序列,针对现有的时间序列预测模型存在滞后性、模态混叠等缺点,提出一种基于EEMD-LSTM-BLS产量预测组合模型。模型首先利用集合经验模态分解...各行业规模化产量分析对产能建设和生产计划调度有着重要的指导意义。各行业生产产量数据为时间序列,针对现有的时间序列预测模型存在滞后性、模态混叠等缺点,提出一种基于EEMD-LSTM-BLS产量预测组合模型。模型首先利用集合经验模态分解(Ensemble Empirical Modal Decomposition,EEMD)将原始产量分解成更加平滑的子序列,可以减小噪声的影响提高预测准确性;再将分解后的子序列分别输入到长短时记忆-宽度学习系统(Long Short Term Memory-Broad Learning System,LSTM-BLS)中训练,利用BLS来解决LSTM预测中的滞后性。为了验证模型有效性,以某卷烟厂产量进行实例分析。通过与基线模型以及现有模型比较,验证提出的模型能更有效、准确的预测产量,为车间生产计划调度提供了便捷有效的方法。展开更多
为了推动大数据技术在制造车间的应用,针对复杂产品晶圆制造过程中海量制造数据时序性、强噪音影响加工周期预测精度的问题,提出考虑特征学习的改进粒子群优化长短期记忆网络(improved particle swarm optimization-long short term mem...为了推动大数据技术在制造车间的应用,针对复杂产品晶圆制造过程中海量制造数据时序性、强噪音影响加工周期预测精度的问题,提出考虑特征学习的改进粒子群优化长短期记忆网络(improved particle swarm optimization-long short term memory,IPSO-LSTM)的加工周期预测方法。采用降噪自编码器和稀疏自编码器联合构建深度自编码器,增强特征学习能力和抗噪能力;运用IPSO优化LSTM参数,克服时间依赖性,提升预测模型性能。实例验证了所提方法的预测精度优于传统机器学习方法,其平均绝对误差低于3%;并分析特征学习方法的有效性,将支持向量回归和多层感知器等传统方法加入特征学习方法,R^(2)分别提高了1.46%、1.05%,为晶圆加工周期的有效预测提供新的解决方法。展开更多
基金funded by Fujian Science and Technology Key Project(No.2016H6022,2018J01099,2017H0037)
文摘The fraction defective of semi-finished products is predicted to optimize the process of relay production lines, by which production quality and productivity are increased, and the costs are decreased. The process parameters of relay production lines are studied based on the long-and-short-term memory network. Then, the Keras deep learning framework is utilized to build up a short-term relay quality prediction algorithm for the semi-finished product. A simulation model is used to study prediction algorithm. The simulation results show that the average prediction absolute error of the fraction is less than 5%. This work displays great application potential in the relay production lines.
文摘各行业规模化产量分析对产能建设和生产计划调度有着重要的指导意义。各行业生产产量数据为时间序列,针对现有的时间序列预测模型存在滞后性、模态混叠等缺点,提出一种基于EEMD-LSTM-BLS产量预测组合模型。模型首先利用集合经验模态分解(Ensemble Empirical Modal Decomposition,EEMD)将原始产量分解成更加平滑的子序列,可以减小噪声的影响提高预测准确性;再将分解后的子序列分别输入到长短时记忆-宽度学习系统(Long Short Term Memory-Broad Learning System,LSTM-BLS)中训练,利用BLS来解决LSTM预测中的滞后性。为了验证模型有效性,以某卷烟厂产量进行实例分析。通过与基线模型以及现有模型比较,验证提出的模型能更有效、准确的预测产量,为车间生产计划调度提供了便捷有效的方法。
文摘为了推动大数据技术在制造车间的应用,针对复杂产品晶圆制造过程中海量制造数据时序性、强噪音影响加工周期预测精度的问题,提出考虑特征学习的改进粒子群优化长短期记忆网络(improved particle swarm optimization-long short term memory,IPSO-LSTM)的加工周期预测方法。采用降噪自编码器和稀疏自编码器联合构建深度自编码器,增强特征学习能力和抗噪能力;运用IPSO优化LSTM参数,克服时间依赖性,提升预测模型性能。实例验证了所提方法的预测精度优于传统机器学习方法,其平均绝对误差低于3%;并分析特征学习方法的有效性,将支持向量回归和多层感知器等传统方法加入特征学习方法,R^(2)分别提高了1.46%、1.05%,为晶圆加工周期的有效预测提供新的解决方法。
基金supported by the National Key Research and Development Program of China(2016YFC0600201)the Academic and Technical Leader Training Program of Jiangxi Province(20204BCJ23027)+1 种基金the Joint Innovation Fund of State Key Laboratory of Nuclear Resources and Environment(2022NRELH-18)the Funds for Guiding Local Scientifc and Technological Development by the Central Government(206Z1705G).
文摘产气量是评估天然气井生产能力和开发工艺效果的重要指标。准确的预测产气量是保证高效生产的关键。为了准确预测储层产气量,本文提出了一种基于鲸鱼优化算法(whale optimization algorithm,WOA)、注意机制(attention mechanism,AM)和双向长短时记忆(bi-directional long short-term memory,BiLSTM)相结合的日产气量预测模型。首先,以Volve油田的排采数据为研究对象,分析了产气量与这些排采参数之间的相关性,并利用LightGBM算法进行重要性排序;然后使用有较强非线性处理能力的基于注意力机制的双向长短时记忆神经网络(bi-directional long short-term memory based on attention mechanism,AM-BiLSTM)构建日产气量预测模型;最后通过鲸鱼算法对AM-BiLSTM模型中的相关参数进行优化,并将参数优化后的模型(WOA-AMBiLSTM)应用于Volve油田的A井。实验结果表明,WOA-AMBiLSTM模型的综合预测性能优于传统的反馈神经网络模型(the back-propagation neural network model,BP)和其他提出的深度学习模型(LSTM、BiLSTM和AM-BiLSTM)。WOA-AM-BiLSTM模型预测曲线与实测测井曲线更加接近,具有更好的预测表现,为储层产能预测提供了一种新思路。