| 引用本文: | 廖霜霜,褚菲,傅逸灵,王军,王福利.基于ILSTM网络的工业过程运行状态评价[J].控制理论与应用,2024,41(11):2112~2120.[点击复制] |
| LIAO Shuang-shuang,CHU Fei,FU Yi-ling,WANG Jun,WANG Fu-li.Operating performance assessment of industrial process based on ILSTM network[J].Control Theory and Technology,2024,41(11):2112~2120.[点击复制] |
|
| 基于ILSTM网络的工业过程运行状态评价 |
| Operating performance assessment of industrial process based on ILSTM network |
| 摘要点击 132 全文点击 34 投稿时间：2022-10-05 修订日期：2024-07-10 |
| 查看全文 查看/发表评论 下载PDF阅读器 |
| DOI编号 10.7641/CTA.2023.20868 |
| 2024,41(11):2112-2120 |
| 中文关键词 动态特性 综合经济指标 LSTM网络 运行状态评价 非优因素识别 |
| 英文关键词 dynamic characteristics comprehensive economic indexes LSTM network operating performance assessment non-optimal factor identification |
| 基金项目 国家自然科学基金项目(61973304, 61873049, 62073060), 江苏省第十六届“六大人才高峰”高层次人才选拔培养项目(DZXX–045), 中央高校基 础研究经费项目(2022ZZCX01K01)资助. |
|
| 中文摘要 |
| 实时掌握复杂生产过程的运行状态对保证企业综合经济效益最大化具有重要意义. 针对工业过程非线性、动态特性显著问题, 本文提出了一种基于综合经济指标驱动的长短期记忆(ILSTM)网络, 用来对复杂工业过程的运行状态进行评价. 该方法利用综合经济指标信息和重构约束, 迫使LSTM网络在学习过程中关注与综合经济指标相关的动态特征. 进一步级联状态识别模型, 构建完整的运行状态评价方法框架. 针对过程的非优运行状态, 提出一种基于重构的贡献图方法, 通过对比各过程变量对非优状态的贡献率识别导致过程运行状态非优的主要原因变量. 最后, 通过重介质选煤过程验证了所提方法的有效性. |
| 英文摘要 |
| It is of great significance to master the operating performance of complex production process in time to ensure the maximization of comprehensive economic benefits of enterprises. For the problem of nonlinear, dynamic characteristics of industrial processes, this paper proposes a comprehensive economic index driven long short-term memory (ILSTM) network for evaluating the operating performance of complex industrial processes. This method utilizes comprehensive economic indexes information and reconstruction constraints to force the LSTM network to focus on the dynamic features related to comprehensive economic indexes in the learning. Further, cascade the performance assessment model to construct a complete operating performance assessment framework. For the non-optimal operating performance of process, a reconstruction-based contribution plot method is proposed to identify the main variables by comparing the contribution rates of each process variable to the non-optimal performance. Finally, the effectiveness of the proposed method is demonstrated on the dense medium coal preparation process. |
|
|
|
|
|