Robust Control Basics

Research Depth 143 in the knowledge graph I know this Set as goal
Unlocks 2 downstream topics
robust-control uncertainty gain-margin phase-margin H-infinity multiplicative-uncertainty robust-stability

Core Idea

Robust control designs controllers that maintain stability and acceptable performance despite uncertainty in the plant model — acknowledging that every model is an approximation and the true plant dynamics are never exactly known. Uncertainty is typically modeled as multiplicative uncertainty G_true(s) = G_nom(s)(1 + Δ(s)W(s)), where G_nom is the nominal model, W(s) is a known frequency-dependent weighting function bounding the uncertainty magnitude, and Δ(s) is an unknown stable transfer function with ||Δ||_∞ ≤ 1. The robust stability condition requires |T(jω)W(jω)| < 1 for all frequencies, meaning the complementary sensitivity function T(s) must be small wherever model uncertainty is large — typically at high frequencies where unmodeled dynamics, resonances, and parasitic effects dominate. Classical gain and phase margins are scalar robustness measures: they quantify how much the loop gain or phase can change before instability, but they capture only specific perturbation directions and can miss structured uncertainty. The H∞ framework generalizes this by formulating the controller design as an optimization: minimize ||T_zw||_∞ (the peak gain from disturbance inputs to performance outputs across all frequencies), which directly shapes the sensitivity and complementary sensitivity functions to meet weighted performance and robustness specifications simultaneously. The small gain theorem provides the foundational result: interconnection of two stable systems with loop gain less than one is stable, and this generalizes to the robust stability condition for multiplicative uncertainty.

How It's Best Learned

Start by computing gain and phase margins for a feedback system and then introducing plant perturbations that violate one margin but not the other, demonstrating that scalar margins can be misleading. Next, model the perturbation as multiplicative uncertainty with a weighting function W(s) and verify the robust stability condition |T(jω)W(jω)| < 1 graphically. Finally, use MATLAB's hinfsyn or Python's control library to design an H∞ controller for a simple plant and compare its sensitivity/complementary sensitivity tradeoff with a classically tuned PID, observing how the H∞ controller explicitly shapes these functions to meet specifications.

Common Misconceptions

Explainer

From your study of sensitivity and disturbance rejection, you know that the sensitivity function S(s) = 1/[1 + L(s)] and the complementary sensitivity function T(s) = L(s)/[1 + L(s)] characterize how a closed-loop system responds to disturbances and reference inputs respectively (S + T = 1). From the Nyquist stability criterion, you know that closed-loop stability depends on how the loop transfer function L(jω) encircles the critical point −1 in the complex plane. Robust control begins by asking: if the true plant differs from your model, how does the Nyquist plot shift, and can it encircle −1 when the nominal plot did not?

The standard way to model this uncertainty is multiplicative uncertainty: the true plant is written as G_true(s) = G_nom(s)[1 + Δ(s)W(s)], where G_nom is your nominal model, W(s) is a known weighting function that describes *how large* the uncertainty can be as a function of frequency, and Δ(s) is an unknown stable transfer function with |Δ(jω)| ≤ 1 for all ω. At low frequencies, your model is usually accurate — physical parameters are well-measured and low-frequency dynamics are well-understood. At high frequencies, unmodeled resonances, computational delays, and parasitic effects can make the true plant deviate substantially from the model. W(s) is typically small at low frequencies and large (possibly greater than 1) at high frequencies, encoding this frequency-dependent uncertainty profile.

The robust stability condition follows from the small gain theorem: two stable systems in a feedback loop are stable if the product of their gains is less than one at every frequency. Applied to multiplicative uncertainty, the loop is robustly stable for all perturbations |Δ| ≤ 1 if and only if |T(jω)·W(jω)| < 1 for all ω. Rearranged: |T(jω)| < 1/|W(jω)|. Where uncertainty is large (high frequencies, |W| large), the complementary sensitivity must be small. This is precisely the Bode "waterbed" tradeoff you saw in sensitivity analysis: pushing T down at high frequencies requires accepting a larger S (reduced disturbance rejection) at low frequencies, and vice versa. Robust control makes this tradeoff explicit and quantitative rather than handled informally.

Classical gain and phase margins are special cases of this framework, but they only measure robustness along specific directions: how much can loop gain increase (gain margin) or phase rotate (phase margin) before crossing the −1 point. A system can have large gain and phase margins yet be brittle to *simultaneous* gain and phase perturbations — the classical margins miss this. The H∞ framework generalizes robustness to arbitrary perturbation types by posing the controller design as a minimax optimization: find the controller C(s) that minimizes the peak gain ||T_zw||_∞ from a vector of exogenous inputs (disturbances, noise, reference signals, uncertainty inputs) to a vector of performance outputs. The ∞-norm picks out the worst-case frequency — the frequency at which the gain from disturbance to error is largest — and the controller is designed to make even the worst case acceptable. Choosing the weighting functions W_S(s) on sensitivity, W_T(s) on complementary sensitivity, and W_U(s) on control effort is the designer's art: it encodes domain knowledge about where the system must reject disturbances, what uncertainty profile the plant has, and how large a control signal is acceptable. The resulting H∞ controller automatically trades off all these objectives simultaneously, something a classically tuned PID cannot do systematically.

Practice Questions 5 questions

Prerequisite Chain

Counting to 10Counting to 20Understanding ZeroThe Number ZeroCounting to FiveOne-to-One CorrespondenceCombining Small Groups Within 5Addition Within 10Addition Within 20Two-Digit Addition Without RegroupingTwo-Digit Addition with RegroupingAddition Within 100Repeated Addition as MultiplicationMultiplication Facts Within 100Division as Equal SharingDivision as Grouping (Measurement Division)Division: Grouping (Repeated Subtraction) ModelDivision: Fair Sharing ModelDivision as Equal SharingDivision as GroupingBasic Division FactsDivision Facts Within 100Two-Digit by One-Digit DivisionDivision with RemaindersRemainders and Quotients in DivisionDivision Word ProblemsIntroduction to Long DivisionFactors and MultiplesPrime and Composite NumbersEquivalent FractionsRelating Fractions and DecimalsDecimal Place ValueReading and Writing DecimalsComparing and Ordering DecimalsAdding and Subtracting DecimalsMultiplying DecimalsDividing DecimalsDividing FractionsMixed Number ArithmeticOrder of OperationsInteger Order of OperationsVariable ExpressionsCombining Like TermsOne-Step EquationsTwo-Step EquationsSolving Multi-Step EquationsEquations with Variables on Both SidesAngle Pairs: Complementary, Supplementary, and VerticalParallel Lines and TransversalsCorresponding AnglesAlternate Interior AnglesTriangle Angle Sum TheoremExterior Angle TheoremTriangle Inequality TheoremSimilar Triangles: AA SimilaritySimilar Triangles: SSS and SAS SimilarityProportions in Similar TrianglesRight Triangle Trigonometry IntroductionTrigonometric Ratios ReviewRadian MeasureConverting Between Degrees and RadiansThe Unit CircleGraphing Sine and CosineGraphing Tangent and Reciprocal Trigonometric FunctionsDerivatives of Trigonometric FunctionsAntiderivativesIterated Integrals and Fubini's TheoremDouble Integrals in Cartesian CoordinatesDouble Integrals over Rectangular RegionsDouble Integrals in Polar CoordinatesDouble Integrals: Definition and SetupIterated Integrals and Fubini's TheoremDouble Integrals over Rectangular RegionsDouble Integrals over General RegionsApplications of Double Integrals: Area, Mass, and MomentsTriple Integrals in Cartesian CoordinatesTriple Integrals in Cylindrical and Spherical CoordinatesChange of Variables and the Jacobian DeterminantApplications of Triple Integrals: Volume and MassVector Fields and Their RepresentationsLine Integrals of Vector FieldsGreen's TheoremSurface Integrals and Flux of Vector FieldsSurface Integrals and Flux of Vector FieldsDivergence Theorem: Flux and OutflowDivergence TheoremElectric FluxGauss's LawConductors in Electrostatic EquilibriumCapacitance and CapacitorsDielectricsDielectric Constant and Relative PermittivityElectric Field Inside Dielectric MaterialsDielectric Materials and PolarizationDielectric Susceptibility and PermittivityEnergy Density in Electric FieldsElectric Current and Current DensityElectrical Resistance and ResistivityOhm's Law and Circuit ElementsElectromotive Force (EMF) and BatteriesKirchhoff's Circuit Laws: Voltage and CurrentDC Circuit Network Analysis MethodsTransient Response in RC CircuitsRC CircuitsLC and RLC CircuitsAC Circuits: FundamentalsImpedance and ReactanceAC Power and ResonanceElectromagnetic WavesFrequency-Dependent Permittivity and DispersionElectromagnetic Waves in Anisotropic MediaBirefringence and DichroismWave Plates: Quarter-Wave and Half-Wave PlatesCircular and Elliptical Polarization ProductionPolarization States: Linear, Circular, and EllipticalLinear Superposition of WavesSuperposition Principle in ElectrostaticsElectric Field Lines and VisualizationElectric Potential and Potential EnergyElectric Potential and VoltageIdeal Voltage and Current SourcesSeries, Parallel, and Combined Resistor NetworksVoltage Divider Principle and ApplicationsKirchhoff's Voltage and Current LawsNodal Analysis MethodLinearity, Superposition, and ScalingAC Steady-State Circuit AnalysisAC Circuit Analysis Using PhasorsAC Power AnalysisResonance in RLC CircuitsFrequency Response and Bode PlotsBode Plot Stability AnalysisNyquist Stability CriterionGain and Phase MarginsPID ControllersLead and Lag CompensatorsLead Compensator DesignCompensator Realization: Active and Passive NetworksLead-Lag Compensation Design and ImplementationCompensation Design: Cascade vs. Feedback Control TradeoffsCascade and Feedforward ControlDisturbance Rejection and Feedforward ControlSensitivity and Disturbance RejectionRobust Control Basics

Longest path: 144 steps · 836 total prerequisite topics

Prerequisites (2)

Leads To (1)