Vehicle suspension control 2: Modal Analysis+ Pole Placement

Diagonalize a matrix to decouple a system by writing A as P D P^{-1}, with D hosting eigenvalues and P its eigenvectors; the order matters for correct diagonalization.

Learn that eigenvectors must be nonzero to preserve direction, while eigenvalues are scalars that can be zero and map vectors to zero, scale them in diagonal or upper triangular cases.

Explore how eigenvectors depend on direction, not magnitude, since any nonzero scalar scales an eigenvector along the same line. Recognize that eigenvalues scale built eigenvectors, not vice versa.

Explore how eigenvectors point in fixed directions with infinite magnitudes, and how the null space of the tilde matrix collects all nontrivial solutions to the homogeneous system.

Identify when a matrix is invertible: a nonzero determinant guarantees an inverse and a unique solution. Zero determinant yields no inverse and no solution or infinitely many solutions.

Compute eigenvalues by solving det(A tilde)=0, where A tilde = A − λI, to obtain the characteristic polynomial of A, yielding nonzero eigenvectors and infinitely many eigenvector magnitudes.

Compute eigenvectors for modal analysis by solving the homogeneous system for lambda five, yielding v1y = -v1x, with nonzero v1x, then note linear independence concept.

Examine linear independence of A1 and A2, then show adding B makes the set dependent; relate independence to a unique solution, an invertible A with nonzero determinant.

Show how linearly independent vectors in R2 yield a unique solution by scaling a1 and a2 to reach b, illustrating that b lies in their span and defines a basis.

When a1 and a2 are linearly dependent, the span is one dimensional, yielding infinitely many solutions to reach b; if b is not in the span, no solution exists.

explain 3d case: a1, a2, a3 as columns of a 3x3 matrix, with a unique solution when the matrix is invertible and b lies in the span of basis vectors.

Examine how span and basis affect linear systems, distinguishing no, unique, and infinite solutions, and how determinant and invertibility of A determine reachability of b.

Identify the second eigenvector by solving A tilde v2 = 0, yielding eigen-directions for lambda=5 and lambda=3. Eigenspaces are the null spaces along these lines, with nonzero magnitudes preserving direction.

Diagonalize A using eigenvalues and eigenvectors to obtain a decoupled system in r, then transform from x to q to r with S to recover the original coordinates.

Explore how diagonalization simplifies raising a matrix to a power, using eigenvalues and eigenvectors to compute A^k quickly, and discuss invertibility and symmetric matrices for model analysis.

Learn how diagonalization decouples linear odes by converting A to a diagonal matrix of eigenvalues. Complex eigenvalues yield complex eigenvectors, increasing computation for large systems; noninvertible P makes diagonalization fail.

Show that a noninvertible P makes the full eigenvector set linearly dependent, while the smaller subset stays independent, so the equation V c = 0 has only the trivial solution.

Showcases that when all eigenvalues are distinct, the eigenvectors are linearly independent, enabling diagonalization via an invertible P and clarifying the role of triangular matrices.

Analyze diagonalizability by constructing independent eigenvectors for C and A1; show P inverse exists, C and A1 are diagonalizable, with eigenvectors v1, v2, v3 forming an invertible eigenvector matrix.

Analyze eigenvectors of a matrix, showing two equal eigenvalues yield dependent eigenvectors and a noninvertible P, then note symmetric matrices have real eigenvalues and invertible eigenvector matrices.

Discover how distinct eigenvalues in symmetric matrices yield orthogonal, linearly independent eigenvectors, making the eigenvector matrix invertible and the matrix diagonalizable.

Explore why orthonormal eigenvectors matter: normalize eigenvectors so P has orthonormal columns, enabling P^T as inverse, and note that distinct eigenvalues ensure orthogonality and symmetric matrices guarantee diagonalization.

Form an orthonormal basis from eigenvectors corresponding to the same eigenvalue of a symmetric matrix. Normalize these vectors to diagonalize the matrix using p transposed instead of p inverse.

Project vector b onto u1 and u2 to extract components parallel to each. Express projections in magnitude and cartesian form, normalize u1, and solve for coefficients via a linear system.

Discover how orthogonal bases simplify vector projections, with coefficients k1 and k2 becoming one when u1 and u2 are orthogonal, and extend to orthogonal vector sets.

Learn the Gram-Schmidt process to build an orthogonal, then orthonormal basis from non-orthogonal vectors, using projections to remove components and enable simple matrix diagonalization.

Explore the gram-schmidt process to build an orthogonal basis from non-orthogonal vectors, scale for convenience, and verify orthogonality while preparing for eigenvalue analysis with an orthonormal eigenvector matrix.

Discover how symmetric matrices diagonalize into an orthonormal eigenvector matrix, with eigenvalues seven (multiplicity two) and minus two, and use Gram-Schmidt to obtain an orthogonal basis.

Apply Gram-Schmidt to form orthogonal eigenvectors, normalize to create an orthonormal P, and diagonalize a symmetric matrix A as A = P D P^T, illustrating eigenspaces, multiplicities, and orthogonality.

Apply modal analysis to a double mass-spring system by transforming to mass-normalized coordinates, diagonalizing the stiffness matrix with an orthonormal eigenvector basis, and deriving decoupled modal equations.

Relate the physical x coordinates to the modal frame R with the mode shapes matrix S, and note S inverse does not equal Stranspose.

Decouple a double mass-spring system with modal analysis, transform to the r coordinate, solve the decoupled ODEs, then return to the x frame via the S matrix.

Apply modal analysis to decouple a two-mass spring system, construct the k tilde and S matrices, transform to modal coordinates, and solve the decoupled ODEs with given initial conditions.

Apply modal analysis to decouple a double mass-spring system and obtain its physical coordinates from modal coordinates using the S matrix, yielding two undamped natural frequencies.

Apply modal analysis to the linearized truck model to derive the analytical response to a disturbance impulse, using a Dirac delta moment to model a sudden engine shutdown.

Apply modal analysis to decouple a truck suspension, transforming to modal coordinates and exploring decoupling of damping by expressing the damping matrix as a linear combination of mass and stiffness.

Decouple the truck with modal analysis by making the damping matrix a linear combination of mass and stiffness, mass-normalize to modal coordinates, and solve impulse responses in that frame.

Decouple the truck dynamics with modal analysis and transform to modal coordinates to show how L1, L2 and inputs distribute across modes, affecting x and theta.

Decouple the system with modal analysis, solve two ODEs in Laplace, map R1 and R2 to x and theta, revealing four poles and damped natural frequencies in a passive response.

Derive the truck’s state-space model, define the states, and form the A and B matrices, distinguishing real and equivalent inputs and a disturbance term for pole placement.

Analyze the eigenvalues of the non-symmetric matrix A to identify open-loop poles and stability, noting possible complex values, disturbance damping, and the need for P inverse for diagonalization.

Compare analytical and numerical solutions for the absolute x and theta, showing how linearization and small angle assumptions affect accuracy, and guide nonlinear state-space formation for Euler simulation.

Rewrite nonlinear odes in state space form, replace Dirac delta with initial angular impulse, and solve with Euler's method to obtain x, theta, and velocities.

Explore solving a truck’s vertical and angular dynamics using Euler’s numerical method, starting from zero initial conditions and no control inputs, then compare analytical and numerical results in python.

Explore pole placement with a full state feedback controller for lti systems, defining u as cr r minus k x and placing closed loop poles at chosen locations.

Learn how to solve an inverse eigenvalue problem by selecting desired closed-loop eigenvalues and computing the gain matrix K to place poles in the left half-plane for stability.

Compute a pole placement controller for a two-state open-loop system by choosing k1 and k2 to place closed-loop poles at minus five and minus six, with k1=14 and k2=57.

Apply a nonzero reference input to steer linearized vehicle suspension dynamics, compute the reference gain CR to eliminate steady-state error, and discuss linearization around equilibrium and pole placement limitations.

Apply pole placement to a suspension system with state feedback u = -k x. Move one pole from 1 to -3 with k1 = 4, revealing uncontrollability and modal analysis.

Assess system controllability via modal analysis and diagonalization, revealing how only one mode responds to input and identifying uncontrollability when P inverse b contains a zero row.

Apply the controllability matrix to determine controllability and stabilizability; if rank MC equals the number of states, all poles are controllable.

Evaluate controllability with the controllability matrix and rank; one system is controllable, the other not. Show how A, B and state feedback allow moving the poles with k1 and k2.

Apply modal analysis to test controllability, showing how influencing each mode lets you span the state space; without all modes being controllable, full control isn’t possible.

Explore pole placement for multi-input systems, deriving closed-loop eigenvalues from open-loop A and B with four unknowns, and using MATLAB's place function to minimize sensitivity of poles.

Use pole placement on the truck to stabilize at x = -1 m and theta = 0, via linearization and a gain matrix K ensuring negative real closed-loop poles.

Apply a pole placement controller to the truck, using place to obtain a real, negative closed-loop pole set and a gain matrix k, with Matlab or Python code.

Explore step-by-step code for a vehicle suspension control system, detailing open-loop and closed-loop matrices, pole placement gains, equilibrium inputs, state increments, and plant input conversion.

Explains applying Euler's algorithm to a truck with a controller, updating state time derivatives and nonzero inputs through f_u and m_u, and compares controller versus no controller for tracking.

Explore simulation results of a nonlinear vehicle suspension model by linearizing at the natural equilibrium, applying a gain matrix, and observing controller-driven damping and reduced x and theta oscillations.

Explain how repeated closed-loop poles are limited by the rank of the input matrix, illustrate with two independent inputs, and discuss implications for diagonalization and pole placement.

Linearize the nonlinear truck model about x = -1 m and theta = 0, and show disturbance rejection with a two-input pole-placement controller stabilizing at -1 m and 0 rad.

demonstrate stabilizing theta at 0.1 radians by tuning the reference input delta r and CR gain while linearizing at zero radians, with x = -1 and theta = 0.

Linearize the system at x = 0.2 m and theta = 0.3 rad to achieve precise stabilization, using a zero delta r reference input and an adjusted cr matrix.

Explore how the final value theorem yields steady-state values from Laplace and time-domain limits. Note its applicability to stable LTI systems only, with examples of unstable or marginal cases.

Apply the final value theorem to derive a universal CR matrix that, with a fixed delta r vector, drives delta x and delta theta to specified reference values.

Apply the final value theorem to derive a universal reference input gain matrix for a two-input two-output system, so the closed-loop outputs converge to constant reference values under pole placement.

Compute the universal CR matrix by forcing the closed-loop transfer function to the identity, ensuring the final state equals the constant reference; the method requires a square transfer matrix.

The lecture shows tracking a moving theta by repeatedly linearizing the truck model at new equilibria. It updates inputs and gain matrices, noting coupling prevents reaching 0.8 radians.

Explore how the generalized A matrix uses cosine theta terms to drive linearization at different angles, and how the gain matrix adapts after each linearization at 0.8 radians.

Modal analysis with pole placement becomes unwieldy for the linearized truck system due to C star and K star symmetry constraints; prefers using Euler's numerical method.

Review modal analysis and pole placement for linear time invariant systems, contrasting analytical and numerical odes. Explain how controllable and disturbance inputs shape the system response and equilibrium values.

Explore solving an LTI state-space system with the state transition matrix, using Laplace transforms to obtain x(t) from x0, with inputs or disturbances, and verify via modal analysis.

Explore solving an LTI state-space system with the state transition matrix, separating the response from initial conditions and external inputs. Learn how outputs are formed with c and d matrices.

Express the state transition matrix as an infinite power series. Differentiate it term by term to obtain x dot and the A factorial pattern; this form supports control engineering proofs.

Apply modal analysis to a non-symmetric system by using P inverse to decouple into modal coordinates, transform initial conditions, and solve decoupled linear time invariant equations via superposition.

Solve an LTI state-space system via modal analysis, transforming initial-condition and input solutions to modal coordinates with p matrix; symmetry in K, C, M yields real eigenvalues and simpler analysis.

Learn to obtain the transfer function matrix for mimo systems from state-space equations in the Laplace domain, with zero initial conditions and the identity C and zero D case.

Derive transfer function matrix from B, C, and D matrices and output vector. Obtain h(s) as a transfer function vector for a single input, showing its effect on all outputs.

Explore why the controllability matrix confirms system controllability in linear time-invariant state-space models, deriving xf as a function of time through a variable change and integral manipulation.

Apply the Cayley Hamilton theorem to rewrite any power of the A matrix as a mix of A and the identity for state transition in LTI systems. Connects characteristic equation and eigenvalues to validate powers like A^2 and A^3.

Apply the Cayley-Hamilton theorem to rewrite A^k in terms of lower powers and the identity for any n by n matrix, yielding a finite n-term state-transition series with time-dependent alphas.

Derive the state transition form and express xf via a convolution with constant A and B, and show the controllability matrix must have rank equal to the four states.

Explore linear algebra, modal analysis, pole placement, and differential equations for MIMO systems, and learn how LQR automatically selects pole locations by prioritizing error minimization and input usage.

Install Python on Windows 11, verify with Hello world, and install NumPy, Matplotlib, SciPy, CVXOPT, SymPy, control, and movie Pi for MPC, LQR, and pole placement.