摘要
针对深海起重机升沉运动时延的问题,文中提出了一种基于支持向量机时滞补偿的滑模预测控制新方法。该方法采用支持向量机对深海起重机的船体升沉运动进行极短期预报,得到深海起重机升沉补偿系统最终的负载位移;将其与麻雀优化算法SSA(Sparrow Search Algorithm)-Elman神经网络的滑模预测控制方法结合;采用SSA对Elman神经网络的权值和阈值进行寻优,通过深海起重机负载位移误差建立滑模面并设计其参考轨迹,利用天牛须算法对控制律进行滚动优化。支持向量机极短期预报方法对深海起重机的船体升沉位移实现精准预测,使控制器可根据升沉运动预测结果提前控制负载位移,补偿时延控制误差,提高了深海起重机系统负载位移的控制精度。
To address the time delay associated with the heave motion of deep-sea cranes,a novel sliding mode predictive control method incorporating time delay compensation based on support vector machines(SVM)is proposed.Utilizing this method,the SVM is employed to rapidly predict the heave motion of deep-sea crane within an extremely short time frame,thereby obtaining the final load displacement of the crane’s heave compensation system.This approach is further integrated with the Sparrow Search Algorithm(SSA)-Elman neural network sliding mode predictive control method.Specifically,the weights and thresholds of the Elman neural network were optimized using the SSA.The sliding surface was established based on the load displacement error of the deep-sea crane and its reference trajectory was designed accordingly.Finally,the control law was optimized using the longicorn algorithm.The very short-term prediction method based on SVM can accurately forecast the heave displacement of the deep-sea crane’s hull,which enables the controller to preemptively adjust the load displacement in accordance with the predicted heave motion.By doing so,it effectively compensates for time delay control errors and significantly enhances the control accuracy of the load displacement in deep-sea crane system.
作者
卢莹斌
周亮亮
孙来庆
朱东科
秦霄
Lu Yingbin;Zhou Liangliang;Sun Laiqing;Zhu Dongke;Qin Xiao
出处
《起重运输机械》
2025年第12期30-37,共8页
Hoisting and Conveying Machinery
基金
2024年山西省重点研发计划项目(202102020101013)。
关键词
深海起重机
支持向量机
升沉运动
时滞补偿
滑模预测控制
SSA-Elman神经网络
天牛须算法
deep-sea crane
support vector machine
heave movement
time delay compensation
sliding mode predictive control
SSA-Elman neural network
beetle antennae search algorithm
作者简介
秦霄,电子邮箱:17836229638@139.com。
