Skip to content
Back to search results

Sliding Mode Control

Introduced by
Vadim Utkin
Year
1977
Type
algorithm
Family
Nonlinear Control
Font size

What does it do?

Sliding Mode Control (SMC) is a robust nonlinear control technique that forces a system to follow a predetermined surface (the sliding surface) in state space by using discontinuous (bang-bang or high-frequency switching) control inputs.

Developed by Utkin and further advanced by Slotine, SMC is remarkably insensitive to parameter variations and disturbances—once the system reaches the sliding surface, its behavior is determined solely by the surface geometry, not by uncertainty. This makes SMC powerful for nonlinear systems, manipulators, and uncertain systems where robustness is paramount.

The intuition behind the method

The key idea of SMC is to define an error surface s(x) = 0 in state space that represents the desired closed-loop behavior. The control law then uses a high-frequency switch to drive the system state toward this surface with maximum speed (reaching phase), and once on the surface, maintains the state there (sliding phase). During the sliding phase, the system's dynamics are governed purely by the surface equation, independent of model uncertainties and disturbances—a property called invariance. This invariance is the source of SMC's robustness: unmodeled dynamics and parametric perturbations do not affect the sliding mode behavior, only the reaching phase.

History

Utkin introduced variable structure systems and sliding modes in the 1970s, developing the theoretical foundation including invariance and reaching conditions. Slotine extended SMC to nonlinear systems in the 1980s and combined it with Lyapunov stability analysis, enabling systematic design for uncertain nonlinear systems. The main challenge addressed over the following decades was chattering reduction through continuous approximations (saturation, boundary layers) and higher-order sliding modes. SMC has seen significant industrial adoption in power electronics, robotics, and aerospace.

When to use it

Use SMC when the system is strongly nonlinear, when parameter uncertainty is large, or when disturbances are unpredictable. SMC is ideal for robotic manipulators (where payload and friction vary), aircraft with uncertain aerodynamics, and power electronics (inherently discontinuous). It excels when robustness to uncertainty trumps smooth control effort. Avoid SMC if your actuators cannot switch at high frequency, if chattering (high-frequency oscillations) is unacceptable, or if system order is very high (surface design becomes complex).

Limitations and considerations

  • Chattering: high-frequency oscillations near the sliding surface due to finite switching frequency; may excite unmodeled dynamics.
  • Requires accurate measurement or estimation of all states; sensor noise can trigger spurious switches.
  • Surface design is problem-dependent; no systematic procedure for choosing good surfaces for all systems.
  • Nonlinear dynamics on the sliding surface can still be unstable if the surface is chosen poorly.
  • Convergence to the sliding surface is not asymptotic; reaching time depends on switching magnitude and system bounds.

Source conditions do not replace independent scientific review or validation by an analysis engine.

Strengths

  • Robust to parametric uncertainty and unmodeled dynamics; behavior on the sliding surface is invariant to disturbances.
  • Handles strongly nonlinear systems without requiring linearization or model inversion.
  • Simple control law (switching logic); can be implemented with simple hardware (on-off switches, relays).
  • Guaranteed stability once on the sliding surface (invariance property); reaching phase can be made fast.
  • Natural handling of actuator saturation; the control law inherently respects hard bounds.

Applications

  • Robotics: Manipulator control with uncertain payload and joint friction, robust tracking of trajectories.
  • Power electronics: DC-DC converter control, inverter switching for AC power delivery, motor drives.
  • Aerospace: Aircraft pitch and roll control with uncertain aerodynamic parameters, missile guidance.
  • Automotive: Traction control, electronic stability program (ESP), anti-lock braking systems (ABS).
  • Process control: Hydraulic system control, uncertain chemical reactors, systems with varying operating points.

How it works

How it works

1

Define the sliding surface s(x) as a linear or nonlinear combination of state errors (e.g., s(x) = e + λ·e_integral for a proportional-integral surface).

2

Compute the discontinuous control law that ensures s(x) = 0 is attractive; typically u = u_eq + u_sw, where u_eq is the equivalent control (continuous) and u_sw is a switching term proportional to sign(s(x)).

3

Check that the reaching condition (d/dt[s^2] < 0) is satisfied to guarantee convergence to the sliding surface.

4

Implement the control law with a high-frequency switch (or use continuous approximation via saturation to avoid chattering).

Common pitfalls

  • Designing a surface with too high order or nonlinearity, leading to complex implementation and slow convergence.
  • Ignoring the chattering problem; assuming infinite-frequency switching is feasible in practice.
  • Setting switching gain too high to suppress chattering without verifying reaching-time guarantees.
  • Failing to account for measurement noise; high-gain switching with noisy sensors causes rapid oscillations.

Frequently asked questions

What is chattering and how can I reduce it?

Chattering is high-frequency oscillation near the sliding surface caused by finite switching frequency trying to keep the system on a surface of measure zero. Practical remedies include: (1) saturation function instead of sign, creating a boundary layer around the surface; (2) higher-order sliding modes where the switching acts on derivatives of the surface; (3) low-pass filtering of control inputs; (4) adaptive switching gain.

How do I design the sliding surface?

A common approach is the canonical form: s(x) = x_2 + c·x_1, where x_1 is the tracking error and c > 0. This ensures that once s = 0, the error x_1 decays exponentially. For higher-order systems, cascade multiple first-order surfaces. The key is ensuring that the zero-dynamics (when s = 0) are stable and yield the desired closed-loop response.

Can SMC be combined with other controllers?

Yes, SMC is often combined with LQR (on a sliding surface), adaptive control (to handle time-varying parameters), or H∞ control (to shape robustness). Combinations leverage the robustness of SMC with optimality guarantees or systematic design procedures of other methods.

What is the reaching phase and how long does it take?

The reaching phase is the time it takes for the system state to reach the sliding surface starting from an arbitrary initial condition. Once on the surface (s = 0), the system enters the sliding phase and remains there. Reaching time is proportional to the maximum error magnitude divided by the switching control magnitude. Faster reaching requires higher control authority but increases chattering risk.

Related methods

Sources

  1. Utkin, V. I. (1977). Variable structure systems with sliding modes. IEEE Transactions on Automatic Control, 22(2), 212-222.

    Open DOI reference
    Copied.
  2. Slotine, J. J. E. (1984). Sliding controller design for non-linear systems. International Journal of Control, 40(2), 421-434.

    Open DOI reference
    Copied.
  3. Utkin, V. I. (2009). Variable structure systems with sliding modes: A survey. IEEE Transactions on Automatic Control, 22(2), 212-222.

    Open DOI reference
    Copied.