离散制造系统能耗动态建模与在线预测  

Dynamics Modeling and Online Prediction of Energy Consumption of Discrete Manufacturing System

作　　者：陈威 王艳[1] 纪志成[1] Chen Wei;Wang Yan;Ji Zhicheng(School of Internet of Things Engineering,Jiangnan University,Wuxi 214122,China)

机构地区：[1]江南大学物联网工程学院,江苏无锡214122

出　　处：《系统仿真学报》2023年第4期760-772,共13页Journal of System Simulation

基　　金：国家重点研发计划(2018YFB1701903)。

摘　　要：针对传统离散制造系统能耗建模方法难以适应工况复杂多变性的问题,提出一种基于实时数据的能耗在线动态建模方法。分析离散制造系统和加工设备运行机理确定能耗影响因素;提出一种可动态调节隐藏层节点数的在线贯序极限学习机算法来构建能耗模型,当有实时数据时可快速更新模型;引入伯恩斯坦不等式提高模型的数据筛选能力。通过仿真实验和对比,验证了该方法具有回归精度高、预测误差小且建模用时短的优点,可应用于离散制造系统能耗的动态建模与在线预测场景。Aiming at the traditional energy consumption modeling methods of discrete manufacturing system being difficult to adapt to the complexity and variability of working conditions,an online dynamic energy consumption modeling method based on real-time data is proposed.The energy consumption affecting factors are determined by analyzing the operation mechanism of the discrete manufacturing system and equipment.An online sequential extreme learning machine algorithm that can dynamically adjust the number of hidden layer nodes is proposed to construct the energy consumption model.The real-time data can update the model quickly.Bernstein's inequality is introduced to improve the model data screening ability.The simulation experiment and the comparison show that the method has better regression accuracy,smaller prediction error and shorter modeling time,and can be applied to the dynamic modeling and online prediction scenarios of energy consumption of discrete manufacturing systems.

关 键 词：离散制造 能耗预测 在线贯序极限学习机 伯恩斯坦不等式 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]