摘要
桥式起重机作为工业领域核心装备,准确预测其在复杂多维吊装-运输作业中的全时域服役能耗,是实现设备节能运行与生命周期能源优化的关键技术问题,而当前单一预测模型难以兼顾时序动态性与复杂运动适应性,导致能耗评价精度不足。为突破上述技术难题,开展基于起重机-货物空间运移特性的服役能耗时序预测研究。以起重机起升、小车运行、大车运行三大机构驱动电机为研究对象,设计不同运动参量下的电机能耗实验并获取基础数据,基于电机耗能原理与能耗累计原则构建基于神经网络的能耗预测模型,进一步融合回归型支持向量机(SVR)提升模型对时序数据的处理能力与预测精度,探索时序状态下驱动电机能耗的瞬态量化原则。通过实验详细分析起重机吊装过程中各电机功率因数的分布规律和动态特性,以深入了解起重机实际运行过程中的能耗动态特性。对比预测结果与实际测量数据可知:文中提出的基于神经网络和SVR的能耗预测方法能够为后续起重机能耗评价提供参考。
As critical equipment in industrial settings,accurately predicting the full-time-domain service energy consumption of bridge cranes during complex multidimensional lifting and transportation operations is a prerequisite for achieving energy-efficient operation and life-cycle energy optimization of the equipment.However,current single-model predictive methods often fail to simultaneously account for temporal dynamics and adapt to complex motion patterns,resulting in inadequate energy consumption evaluation accuracy.To address this challenge,a method for temporal prediction of service energy consumption was proposed,leveraging the spatial movement characteristics of the crane and its lifted load.Taking the crane lifting mechanism,trolley and crane operation motors as research objects,experiments were carried out under diverse motion parameters to gather energy consumption data.A neural network-based energy consumption predictive model was developed,integrating the principles of motor energy consumption and cumulative energy calculation methods.Furthermore,a support vector regression(SVR)was further integrated into the model to enhance its ability to process time-series data and improve prediction accuracy,thereby enabling the transient quantification of motor energy consumption under dynamic operational conditions.A detailed analysis of the distribution and dynamic characteristics of motor power factors during hoisting processes of cranes was conducted through experiments,in order to gain a deeper understanding of the energy consumption dynamics of cranes during actual operation.By comparing the predicted results with actual measurement data,it can be seen that the proposed energy consumption prediction method based on the neural networks and SVR can provide a reference for subsequent evaluation crane energy consumption.
作者
戚其松
郭志恒
刘文明
赵忠祥
QI Qisong;GUO Zhiheng;LIU Wenming;ZHAO Zhongxiang(School of Mechanical Engineering,Taiyuan University of Science and Technology,Taiyuan Shanxi 030027,China;Henan Dongqi Machinery Co.,Ltd.,Xinxiang Henan 453400,China)
出处
《机床与液压》
北大核心
2025年第18期212-222,共11页
Machine Tool & Hydraulics
基金
山西省基础研究计划项目(202203021221156)
山西省重点研发计划项目(202402150101009)。
关键词
起重机
服役能耗
机器学习
神经网络
SVR
全时域
crane
in-service energy consumption
machine learning
neural network
SVR
full-time domain
