Dynamic Modeling And Sliding Mode Control For A Novel3-DOF Parallel Mechanism With Actuation Redundancy

Technology of agricultural robot is an important symbol of agricultural modernization. However, with the expanding of research and applied field, the increasing of agricultural operating requirements and the raising of the complexity for control, the control requirements are improving. Compared with serial robot mechanism, parallel mechanism has the distinct advantages such as low inertia, high stiffness, high accuracy, fast response and so on. The agricultural robots based on parallel mechanisms are expected to further improve the operating performances in some agricultural engineering applications. This paper focuses on the research of parallel mechanism and lays the foundation for further practical application of agricultural parallel robot. Parallel mechanism is a complex multi-chains system, with characteristics such as time-varying, highly nonlinear, and strong coupling. Affected by parameter uncertainty, un-modeled dynamics, load disturbance, joint and servo friction and other outside interference, the traditional control strategies have been difficult to achieve desired control performances. The controllers based on dynamics can meet the control requirements. Compared with serial robot mechanism, the dynanics model for parallel mechanism is more complicated. Therefore, it is necessary to research on dynamics modeling and control methods.The applied research of agricultural robot based on parallel mechanism is at starting stage, due to diversity of parallel mechanism, there are no mature and effective control methods by now. The sliding mode control is independent on mathematical model and insensitive to external disturbance and parameter variation. This paper focuses on various control methods and searches a kind of control scheme with superior comprehensive property to improve the performances of the studied parallel mechanism. Considering a novel3-DOF parallel mechanism with actuation redundancy as research subject, some related issues are deeply researched. The main contents are as follows:(1) In order to research the dynamics and control methods of the novel parallel mechanism with actuation redundancy, the mechanical characteristic and constrain between three pose parameters are analysed, and the inverse kinematic solution is established. Finally, the workspace and singularity for the studied mechanism are studied.(2) Without considering the nonlinear dynamics and the coupling effect among each join, the precision for control method based on kinematicsis seriously affected, even the stability cannot be guaranteed in the high speed and high-acceleration cases. So the control methods based on dynamical model are introduced to parallel mechanism system. However, due to the closed-chain structure and the coupling effect between each chain, it is difficult to establish precise dynamic model. Meanwhile, the control method based on dynamics model is closely related to the precision of the dynamical model, so it is necessary to establish an accurate and efficient dynamic model from the control subject. The inverse dynamic model for the parallel mechanism is formulated using Lagrangian formalism and the driving force is optimized utilizing the minimal2-norm method. By investigating the contribution of each term in the dynamic model, a simplified strategy of the dynamic model for real-time control application is presented, and the subsequent model error is compensated by adopting the RBF neural network.(3) Considering the large calculation load and the relative independence of movement and control between the redundant chain and the non-redundant chains, a control scheme is proposed to improve the rapidity and real-time performances, employing dynamic control in the redundant chain and kinematic control in the non-redundant chains. For the non-redundant chains, a backstepping sliding mode control strategy is proposed to improve the tracking performances and an adaptive law is introduced to reduce the effect of parameters’ variation and external disturbances.(4) Without considering the coupling effect between non-redundant chains, the backstepping adaptive sliding mode control method can improve the rapidity and real-time. According to the simplified dynamic model and the characteristics of sliding mode control, a dynamic sliding mode control based on synchronisation error is proposed to enhance the control accuracy.(5) The dynamic sliding mode control based on synchronisation error can further improve the motion control precision, but it decreases the rapidity and real-time. In order to obtain a control method with better comprehensive performances, this paper addressed a nonsingular sliding mode control method. The principle of this approach is that decouple the whole system into three subsystems by extracting the coupling force and gravityand design controllerfor each subsystem independently. Meanwhile, a RBF neural network is used to compensate the cross-coupling force and the gravity to enhance the control precision. In order to obtain better comprehensive performances, the nonsingular sliding mode controllers are designed to drive the system states converge to equilibrium point in finite time. Comparing all the simulation results of the backstepping adaptive sliding mode control, the dynamic sliding mode control based on synchronisation error and the decoupled nonsingular sliding mode control, the last controller is of the better comprehensive performances.(6) According to mechanical features and control requires of parallel mechanism with actuation redundancy, the software is developed based on Visual C++. Finally, the relevant experiments on the parallel mechanism are given, and the experiment results indicate that the decoupled nonsingular sliding mode controller is of the better comprehensive performances.The research lays the foundation for the control theory research of agricultural parallel robot and the practical application of parallel mechanism in agricultural engineering.