Screw Theory-Based Decomposition and Compensation Control for Large Hydraulic Manipulators With Multi-Closed-Chain Structures

Abstract

High-precision control of Large Hydraulic Manipulators (LHMs) is challenging due to the complex kinematic constraints of multi-closed-chain structures and the nonlinear characteristics of Counterbalance valves (CBVs). To address these issues, a Screw Theory-Based Decomposition and Compensation Control (ST-DCC) method is proposed. The control framework addresses the mechanical and hydraulic subsystems explicitly. The complex multi-closed-chain structure is geometrically decomposed into reduced luffing and actuation subsystems via Virtual Cutting Points (VCPs). Screw Theory is utilized to establish a compact kinematic model and an exact internal force mapping, which simplifies the Jacobian derivation and ensures rigorous force allocation. A physics-informed feedforward controller is designed to compensate for the nonlinearities of the CBVs. Comparative experimental results on a 3-DOF heavy-duty LHM demonstrate the superiority of the proposed method. Compared with conventional VDC, the proposed strategy reduces the trajectory-tracking RMSE by over 34% and reduces the maximum jump in hydraulic cylinder force by up to 55.9%. The proposed method improves both trajectory tracking accuracy and dynamic stability, providing a practical control solution for LHMs with multi-closed-chain structures. Note to Practitioners—This paper is motivated by a major challenge in industry: automating large hydraulic manipulators for high-precision tasks. In practice, the complex linkages and hydraulic safety valves of these machines cause severe tracking errors and shocks. We propose a practical, software-based control framework to overcome these issues. This solution significantly improves operational quality (better trajectory precision) and equipment reliability (fewer mechanical vibrations). The key insight of the proposed approach is to use screw-theoretic modeling to resolve complex internal force distribution. A physics-informed hydraulic model is further used to compensate for CBV-induced dynamic lag and nonlinear flow characteristics. The proposed algorithm is directly implementable on standard industrial controllers and requires only standard industrial sensors, such as angular encoders and pressure sensors, which are commonly available in conventional LHM systems. Therefore, no additional specialized or high-cost sensing hardware is required. A current limitation is that our model assumes the massive steel arms are completely rigid. Future work will incorporate structural flexibility into the control loop to further enhance accuracy. We will also optimize the system for energy consumption. Beyond construction and forestry cranes, these results can be widely applied to automate other heavy equipment. Examples include foundry manipulators, subsea hydraulic robots, and automated mining systems.