1. Developed an MPC-based framework to enhance trajectory tracking for a tractor-trailer system, addressing the non-holonomic constraints, and optimizing control authority.

2. Achieved successful path tracking on both an 8-shaped and circular trajectory, demonstrating the efficacy of the control system through comprehensive simulation results.

3. Conducted robustness analysis revealing the controller’s resilience to parameter variations, with a focus on minimizing tracking errors and maintaining desired velocity.

The project was done in two stages:

1]         The first stage (accounts for at least 60% of the project) focused on the development/validation of a controllable and observable model of the open-loop plant (black box MIMO system) through “LINEAR SYSTEM IDENTIFICATION” techniques.

Ø  We used frequency response-based system ID methods for our model estimation. So, a major task was figuring out the proper excitation signal that could excite all of the dynamic modes in the bandwidth of interest and at the same time avoid clipping both the control and response signals.

Ø  After achieving acceptable signal-to-noise ratios associated with each path, we computed the H1 frequency response function estimates for all four paths.

Ø  We then obtained discrete-time transfer functions for each of the four paths based on the frequency response function estimates and converted them to corresponding discrete-time LTI state-space object.

Ø  Performed a model order reduction based on the Hankel Singular values to generate a balanced realization of the discrete SS model.

Ø  Lastly, we validated the estimated model with the actual plant in both frequency as well as time domains.

2]         The second stage dealt with the design of an output feedback control system that can regulate the responses of the dynamic plant.

Ø  First, an LQR-based state feedback control law with appropriate performance specifications was implemented. This was done to get an optimal full state feedback gain matrix “G” for the discrete-time estimated open-loop model.

Ø  Finally, a full state feedback control with Kalman state estimator (output feedback control) was implemented and simulated. The tuning parameter for the Kalman estimator (in our case QN-process noise covariance matrix) was tweaked to satisfy the required estimation convergence and then eventually control performance.

The aim of this simulation study was to develop a MIMO control system to regulate the flexible modes and the rigid body (i.e., pointing) modes of a satellite with flexible solar panels. The satellite has two bi-directional attitude control thrusters to help maintain the proper orientation of its solar panels and to actively damp out unwanted vibrations induced by attitude repositioning. For the sensing part, a total of four sensors were distributed along the length of the solar panels to measure the motions of the panels.

The final design objective was to develop an output feedback control system (i.e., full state-feedback with a state estimator) that can regulate (i.e., drive to zero) the four sensor measurements (Outputs) using the two attitude control thrusters (Inputs).

Major snapshots were:

1) Evaluation of the open-loop response (controllability, output stability, observability, etc.)

2) Design of a full state feedback control with the required performance.

3) Design of an output feedback control with state estimation based on the Luenberger observer.

4) Designing and validating an LQR optimal compensator.

1] Implemented an analytical model of a rigid Rotor bearing system with 4 DOF. This was followed by the study of variation of natural frequencies with RPM along with the identification of whirl nature (like Backward or Forward whirl) at each frequency. Secondly, animations of various mode shapes exhibited by the rotor at those natural frequencies were also generated.

2] Performed MATLAB-based numerical simulation of a Jeffcott rotor coasting up past its critical speed to its operating range. The simulation was then extended to include the effects of various rotor defects like transverse crack, stiffness asymmetry, Bent rotor, and rotor with rotor-stator rub.

3] Implemented a generic FEM model in MATLAB for the dynamic analysis of a rotor-bearing system with bearing characteristics. Further various analyses were performed like generation of critical speed map, unbalanced rotor vibration response, animation of operation deflection curves, etc.

A Fuzzy logic-based diagnostic model was implemented in MATLAB.

First, an analytical model of a three-phase induction motor along with its various commonly occurring faults have been implemented in MATLAB. Following that, the model's output (three-phase currents I_a, I_b, I_c) were observed based on which the input and output membership functions were arrived at. The Fuzzy logic rules were set to be the most commonly used rules given in the literature. Finally, the Fuzzy logic output (Stator condition - SC) were plotted and analyzed for different cases (both with and without faults).

CONCLUSION - A highly versatile technology for condition monitoring and fault analysis of electro-mechanical systems !!!

The objective was to identify the major sources of noise in a Vacuum cleaner through source characterization techniques followed by setting a target dB level after ranking them. Finally, appropriate noise reduction methods were implemented (in this case, mufflers for airborne noise and damping material mounts for reducing structure-borne noise) to meet the target level. During this project, we came up with an indigenously written MATLAB code for providing us with the 1/3rd octaves and overall sound level in dB from the sound clip measured using our smartphones. Although there will be some errors involved due to the non-uniform sound sensitivity characteristics of our smartphone microphones, the results were satisfactorily close to the ones given by a B&K professional sound analyzer.