This introductory lecture for a control theory course establishes the fundamental concepts of dynamic systems. It defines what constitutes a system, its autonomous evolution over time, and introduces the core areas of automation engineering: analysis, synthesis, and identification. The lecture focuses on defining dynamic systems formally, outlining their components (input, output, state, disturbances) and presenting the state-space mathematical model. Emphasis is placed on Linear Time-Invariant (LTI) systems, their properties (free/forced evolution), and matrix representation (A, B, C, D). A significant portion is dedicated to stability analysis, defining equilibrium points, stability (Lyapunov), and asymptotic stability. The connection between system stability and the eigenvalues of the system matrix A is explained, detailing conditions for asymptotic, marginal, and unstable behavior based on eigenvalue location in the complex plane. The lecture then addresses solving LTI system equations, highlighting the complexity of time-domain solutions involving matrix exponentials and convolution. The Laplace transform is introduced as a powerful tool to simplify analysis by converting differential equations into algebraic ones in the frequency domain, leading to the concept of the transfer function relating system input to output. The relationship between transfer function poles and system eigenvalues (controllability/observability) is briefly touched upon. Finally, the concepts are illustrated with a practical example of bicycle dynamics, comparing a simplified unstable model with a more realistic one that exhibits velocity-dependent asymptotic stability, aligning with real-world experience.
Basics of Dynamic Systems Theory
0:00:15 Motivation: Introduces the concept that many things around us, from simple heaters to complex robots, can be viewed as systems.
0:00:40 System Definition: A system is formally defined as an interconnected aggregation of parts, whose autonomous evolution over time is of interest in automation engineering.
0:01:46 Pillars of Automation: Automation involves Analysis (predicting output from input/system), Synthesis (finding input for desired output/system), and Identification (determining the system from input/output). This course focuses on Analysis and Synthesis.
0:03:31 Dynamic System Defined: An oriented system (with cause-effect relationships) that evolves over time, characterized by inputs (u), outputs (y), internal state (x), and potential disturbances (d).
0:05:48 State-Space Representation: Introduced as a general mathematical model for dynamic systems: x_dot = f(x, u, t), y = g(x, u, t).
0:07:10 LTI Systems: The course focuses on Linear Time-Invariant (LTI) systems, described by constant matrices A, B, C, D: x_dot = Ax + Bu, y = Cx + Du.
0:07:52 Free vs. Forced Evolution: Defines system behavior with no input (free evolution, u=0) versus behavior driven by input from a zero state (forced evolution, x(0)=0).
0:12:37 Stability Concept: Introduces stability as a crucial system property, analyzing how the system's state evolves after small perturbations from an initial state, typically studied in free evolution (u=0).
0:14:23 Equilibrium Points: Defines equilibrium states (x_e) as points where the system's evolution stops (x_dot = 0, or Ax_e = 0 for LTI systems). Equilibrium does not imply stability.
0:15:55 Stability (Lyapunov): An equilibrium is stable if starting arbitrarily close to it ensures the system's trajectory stays within a bounded region around it.
0:17:43 Asymptotic Stability: A stronger condition where the system is stable, AND trajectories starting close enough to the equilibrium converge back to it over time. This is a key desirable property.
0:19:01 Stability Analysis via Eigenvalues: For LTI systems, stability is determined by the eigenvalues (λ) of the system matrix A.
Asymptotically Stable: All eigenvalues have negative real parts (Re(λ) < 0).
Marginally Stable: All eigenvalues have non-positive real parts (Re(λ) ≤ 0), AND eigenvalues with Re(λ) = 0 have equal algebraic and geometric multiplicities.
Unstable: At least one eigenvalue has a positive real part (Re(λ) > 0) OR an eigenvalue with Re(λ) = 0 has differing multiplicities.
0:25:25 Pendulum Example: Illustrates equilibria and stability concepts using a simple pendulum (downward position is asymptotically stable, upward balanced position is unstable).
0:27:50 Time-Domain Solution: Solving x_dot = Ax + Bu explicitly involves complex calculations with matrix exponentials and convolution integrals.
0:30:05 Laplace Transform: Introduced as a mathematical tool to simplify analysis by transforming differential equations in the time domain (t) into algebraic equations in the frequency domain (s).
0:30:38 Key Laplace Properties: Linearity, derivative property (d/dt becomes multiplication by 's'), and convolution property (convolution in time becomes multiplication in 's').
0:32:54 Transfer Function G(s): Derived in the Laplace domain, G(s) = C(sI - A)^-1 B + D algebraically relates the input U(s) to the output Y(s) for zero initial state: Y(s) = G(s)U(s).
0:33:30 Poles and Eigenvalues: The poles of the transfer function (roots of its denominator) are a subset of the system's eigenvalues. Eigenvalues appear as poles if they are both controllable and observable.
0:36:01 Bicycle Dynamics Example: A practical application of the concepts.
0:38:44 Bicycle Model 1 (Neutral Fork): A simplified model assuming a 90° front fork angle results in an inherently unstable system (one positive pole/eigenvalue), requiring active control (steering) to stay upright. This doesn't match reality for riding "no hands".
0:43:24 Bicycle Model 2 (Inclined Fork): A more realistic model with an inclined fork (lambda < 90°) shows that the bicycle becomes asymptotically stable (self-stable) above a certain critical velocity (V_crit), matching the real-world ability to ride without hands at sufficient speed. All eigenvalues move to the left-half plane above V_crit.
0:45:48 Rider Influence: Briefly mentions that adding a rider model introduces additional control possibilities (body lean), potentially making stabilization easier (teaser for future lectures).
0:47:36 Recap: Summarizes the lecture's key topics: system definitions, LTI models, stability analysis via eigenvalues, Laplace transforms/transfer functions, and the bicycle example.