Multirate sliding mode disturbance compensation for model predictive control

Davide M. Raimondo, Matteo Rubagotti, Colin N. Jones, Lalo Magni, Antonella Ferrara, Manfred MorariInternational Journal of Robust and Nonlinear Control 2014 · DOI

TL;DR

This paper introduces a novel hierarchical control strategy that combines a robust nonlinear Model Predictive Controller (NMPC) with a faster-acting Multirate Sliding Mode Disturbance Compensator (MSMDC). This approach aims to improve the robustness and performance of NMPC for nonlinear systems by actively compensating for disturbances and model uncertainties, leading to better constraint satisfaction and stability.

What Problem Does This Paper Address?

Controlling complex nonlinear systems, especially those with uncertainties and external disturbances, is a significant challenge. Model Predictive Control (MPC) is a powerful technique for handling constraints and optimizing performance. However, standard MPC can become conservative or even infeasible when faced with significant uncertainties. Robust MPC methods exist, but they often come with a high computational burden or can be overly conservative. Sliding Mode (SM) control is known for its robustness to matched disturbances but can suffer from chattering and doesn't inherently handle constraints or optimality as well as MPC. This paper addresses the need for a control strategy that leverages the strengths of both MPC (optimality, constraint handling) and SM control (robustness to disturbances, lower computational cost) in a practical, multirate framework for nonlinear discrete-time systems.

Methods

The core of the proposed solution is a hierarchical control architecture. The system is controlled by two components operating at different rates:

  • High-Level Controller (HLC): This is a robust Nonlinear Model Predictive Controller (NMPC). It operates at a slower rate, making decisions based on a longer prediction horizon and handling state and input constraints. The paper specifically uses a "tightened constraints" NMPC approach for computational efficiency.
  • Multirate Sliding Mode Disturbance Compensator (MSMDC): This component operates at a much faster rate than the NMPC. Its primary role is to actively compensate for matched disturbances and model uncertainties. It does this by estimating the disturbance and generating a control signal that steers the system towards a "sliding manifold" defined to account for the difference between predicted and actual system behavior.

A key innovation is the dynamic input allocation between the NMPC and MSMDC. Instead of a fixed split of control effort, the allocation adapts based on the system's state. When the system is far from the desired trajectory, more control effort is allocated to the MSMDC for disturbance rejection. As the system approaches the desired trajectory, more control effort can be shifted to the NMPC to ensure optimality and constraint satisfaction. The paper also introduces a "M-sampled" system representation to facilitate the integration of the faster MSMDC with the slower NMPC.

Datasets

Not discussed in the paper. The evaluation is performed using a simulated two-dimensional nonlinear system.

Key Insights and Ideas

  • Synergy of MPC and SM Control: Combining NMPC with a faster-acting MSMDC offers a powerful way to achieve both robust disturbance rejection and optimal, constrained control.
  • Multirate Operation is Key: The significant difference in sampling rates between the NMPC and MSMDC is crucial. The faster MSMDC can react to disturbances much more quickly than the NMPC, effectively "hiding" much of the disturbance from the NMPC.
  • Dynamic Input Allocation: A fixed allocation of control effort between the two controllers can be suboptimal. Dynamically adjusting this allocation based on the system state allows for a better balance between disturbance rejection and optimal control, potentially improving the domain of attraction and overall performance.
  • Reduced Conservativeness: By actively compensating for disturbances, the MSMDC reduces the "equivalent disturbance" seen by the NMPC. This allows the NMPC to operate with less conservative constraints, potentially leading to better performance and a larger feasible region.
  • Practical Implementation: The paper focuses on a discrete-time implementation of SM control, avoiding the chattering issues associated with directly discretizing continuous-time SM controllers.

Key Results

The paper presents theoretical conditions for input-to-state stability (ISS) and constraint satisfaction for the combined system. A numerical example demonstrates the effectiveness of the proposed strategy:

  • The multirate SM disturbance compensation effectively counteracts rapidly varying disturbances.
  • The dynamic input allocation strategy adapts the control effort, allowing the MSMDC to take over disturbance rejection when needed and freeing up control amplitude for the NMPC when the system is closer to the desired state.
  • The proposed approach leads to a significant reduction in the chattering effect compared to a directly discretized SM controller.
  • The combined strategy allows for ISS to be guaranteed under conditions that would not hold for the NMPC alone.

Limitations and Future Work

  • Conservativeness of SM Control Design: The paper acknowledges that the functions used in the SM control design (e.g., $\eta$) can be conservative, which might affect the performance.
  • Computational Burden: While the MSMDC is computationally cheaper than NMPC, the overall scheme still requires solving optimization problems, and the multirate nature adds some complexity.
  • Future Work: The paper suggests exploring other robust NMPC strategies that could be combined with the MSMDC. It also hints at further research into dynamically adjusting the MSMDC control amplitude to preserve performance and a large domain of attraction. The authors also mention that the advantage of MSMDC becomes more apparent as the sampling time ($T_s$) increases.

Summary updated

Abstract

Summary In this paper, a novel hierarchical multirate control scheme for nonlinear discrete‐time systems is presented, consisting of a robust nonlinear model predictive controller (NMPC) and a multirate sliding mode disturbance compensator (MSMDC). The proposed MSMDC acts at a faster rate than the NMPC in order to keep the system as close as possible to the nominal trajectory predicted by NMPC despite model uncertainties and external disturbances. The a priori disturbance compensation turns out to be very useful in order to improve the robustness of the NMPC controller. A dynamic input allocation between MSMDC and NMPC allows to maximize the benefits of the proposed scheme that unites the advantages of sliding mode control (strong reduction of matched disturbances, low computational burden) to those of NMPC (optimality, constraints handling). Sufficient conditions required to guarantee input‐to‐state stability and constraints satisfaction by the overall scheme are also provided. Copyright © 2014 John Wiley & Sons, Ltd.
BibTeX
@article{raimondo2014multirate,
  author = {Davide M. Raimondo and Matteo Rubagotti and Colin N. Jones and Lalo Magni and Antonella Ferrara and Manfred Morari},
  title = {{Multirate sliding mode disturbance compensation for model predictive control}},
  journal = {International Journal of Robust and Nonlinear Control},
  year = {2014},
  doi = {10.1002/rnc.3244},
}
Figure