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。
Production Prediction Model Based on Deep-Broad Fusion Time-Frequency Decomposition
Scale production analysis of various industries plays an important role in productivity construction,pro-duction planning and scheduling.The production output data of various industries are time series.Aiming at the short-comings of the existing time series forecasting models such as lag and mode mixing,the article proposes a new time series forecasting combined model based on EEMD-LSTM-BLS.The model first performs Ensemble Empirical Modal Decomposition(EEMD)on the time series data.EEMD decomposes the original time series into smoother sub-se-quences,which can improve the prediction accuracy;then input the decomposed sub-sequences into training in the Long Short Term Memory-Broad Learning System(LSTM-BLS),and use BLS to solve the lag in LSTM prediction.The article takes the cigarette production of a cigarette factory as an example to verify that the EEMD-LSTM-BLS model can more effectively and accurately predict the cigarette factory's cigarette production output compared with baseline models as well as existing deep models.Providing a convenient and effective method for cigarette production planning and scheduling in workshops.
Time seriesProduction forecastEnsemble empirical mode decompositionLong and short-term mem-ory networkBroad learning
万方数据
维普
