Boris Rohal'-Ilkiv

Model predictive control algorithms for active vibration control: a study on timing, performance and implementation properties

The dynamic properties of vibration control systems pose unique requirements and challenges on the implementation of model predictive control (MPC) algorithms with stability and feasibility guarantees. This article presents a comprehensive experimental comparison of computation timing and damping performance for various MPC methods; analyzing their offline and online properties in active vibration control and their impact on practical implementability. Optimal and sub-optimal MPC algorithms providing guaranteed stability and constraint feasibility have been applied to the real-time active vibration attenuation of a lightly damped mechanical test structure. Based on the experiments presented in this paper, the standard and sequential quadratic programming-based, optimal and sub-optimal minimum time multi-parametric programming-based and the sub-optimal Newton–Raphson’s algorithm-based MPC methods demonstrate closely comparable vibration attenuation performance. The offline and online timing analysis indicates that the underlying difference between the investigated MPC algorithms lies mainly in practical implementability difficulties caused by inherent algorithm efficiency, rendering certain variants of MPC more suitable for vibration control than others.

Basic laboratory experiments with an educational robotic arm

This work addresses design and construction issues of a laboratory robotic arm for educational purposes. First of all, the robotic arm performance analysis has been accomplished using Matlab / Simulink / SimMechanics. The obtained knowledge has been utilized to develop the suitable algorithms for analyzing the robotic arm kinematics. Once the SimMechanics model is successfully determined, a real-time xPC target system is used in order to connect the real laboratory robotic arm with the corresponding Matlab / Simulink block diagram. It is important to remark that the developed robotic arm is a convenient tool for learning robotics at any favorable technical university laboratory. On the other hand, the manipulator has six degrees of freedom. Three degrees of freedom correspond to the robotic arm and the rest belongs to the gripper. Moreover, the necessary electronic modules have been developed in order to allow a successful standard communication with the available laboratory devices.

Moving Horizon Observer for vibration dynamics with plant uncertainties in nanopositioning system estimation

This paper considers the estimation of states and parameters of a Single-Degree-of-Freedom (SDOF) vibration model in nanopositioning system based on a nonlinear Moving Horizon Observer (MHO). The MHO is experimentally tested and verified on measured data. The information about the displacement and speed together with the system parameters and unmodeled force disturbance is estimated through the Sequential Quadratic Programming (SQP) optimization procedure. The MHO provided superior performance in comparison with the benchmark method Extended Kalman Filter (EKF) in terms of faster convergence.

Damped One-Mode Vibration Model State and Parameter Estimation via Pre-Filtered Moving Horizon Observer

Abstract The estimation of parameters and states is one of the core data-processing algorithms used for the monitoring and control of continuum or structured mechatronical systems (e.g., flexible robotic arms and cantilevers). The measurements taken from the sensors combined with an appropriate model can filter the states and extract information about vibration dynamics parameters such as damping and spring constants. This paper presents a method based on the application of a Moving Horizon Observer (MHO) for state and parameter estimation of a lightly-damped vibrating system. A numerical experiment on a single-degree-of-freedom (SDOF) system is performed to demonstrate the method on noisy measurements and to compare the MHO with the benchmark Extended Kalman Filter (EKF). Test scenarios show that the MHO appears to be more robust with respect to modeling errors.

Parallel numerical optimization for fast adaptive nonlinear moving horizon estimation

This paper proposes a novel strategy using parallel optimization computations for nonlinear moving horizon state estimation, and parameter identification problems of dynamic systems. The parallelization is based on the multi-point derivative-based Gauss-Newton search, as one of the most efficient algorithms for the nonlinear least-square problems. A numerical experiment is performed to demonstrate the parallel computations with the comparison to sequential computations.

Nonlinear air-fuel ratio predictive control of spark ignited engines

The aim of this paper is to present a nonlinear model-based predictive control (NMPC) for air-fuel ratio (AFR) of spark ignited (SI) engines. Complexity of NMPC implementation causes that the computational requirements on storage space and online computation time have to be considered in the controller design. Especially a limitation on computation time makes often the optimal control law too complex to be implemented and some suboptimal solution has to be provided to meet the requirements. The point of the paper is to present the controller which utilizes the simplified control problem in a warm-start technique and allows to decrease the online computational effort caused by the nonlinear predictive controller.

Control systems in omni-directional robotic vehicle with mecanum wheels

Mobile robotic systems are mechatronic devices that are currently becoming more and more complex. To design such a system, a combination of expertise from the fields of mechanical, electrical and computer engineering is required. This paper describes our custom omni-directional robotic platform designed for both indoor and outdoor use that contains a lot of prototypic hardware. All the designed components are interesting as they permit control of the robot as a complex mechatronic system. The main objective here was to use a larger amount of smaller control subsystems, rather than a central one. This is more advantageous from the control point of view. The electronics of the robot consists of microcontroller controlled distributed subsystems that are able to communicate with the master system. The following sections offer an insight into the control structures of the mobile robot.

MULTIPLE ARX MODEL-BASED AIR-FUEL RATIO PREDICTIVE CONTROL FOR SI ENGINES

Abstract In this article, predictive control is suggested to control the injection fuel pulse width in such a manner that the air-fuel ratio deviates as little as possible from the stoichiometric ratio during the transients of an SI engine. The applied control strategy is based on the knowledge of an internal model of the air-path, predicting the change of the air flow through cylinders, and consequently, setting the prediction profile of the desired values of the objective function. The second modeled subsystem, the fuel-path, is an explicit component of the objective function where the amount of the fuel is a function of the control action. It was demonstrated by simulation that multiple model predictive control has the potential to be appropriate mixture control strategy. Thus with further research in the predictive control of air-fuel ratio, cleaner exhausts may be expected.

Efficient Embedded Model Predictive Vibration Control via Convex Lifting

This paper presents an efficient real-time implementation of embedded model predictive control, adopted in the context of active vibration control with the objective of minimizing the tip deflection of lightly damped cantilever beams. In particular, we focus on memory and time-efficient explicit solutions of the associated constrained optimal control problems that are easily implementable on low-end embedded hardware. To this end, we exploit the concept of convex lifting and show how it can be used to devise low-complexity, regionless piecewise affine controllers without any loss of optimality and performance. The efficiency of this constructive procedure is quantified via an extensive complexity analysis, evidenced by a successful practical deployment and optimal vibration control performance using a family of 32-bit ARM Cortex-M-based microcontroller platforms.

Efficiency and performance of embedded model predictive control for active vibration attenuation

The development of efficient solution approaches and technological advances facilitate the use of predictive control on embedded systems, even for fast systems or on computationally limited hardware platforms. The practical implementation of predictive control is, however, still often very time consuming and demands insight into the formulation and solution strategies for predictive control. Tools for automatic code generation tailored for deployment on embedded systems promise to overcome these problems and thus enable the fast and reliable implementation of predictive control. This paper exploits the efficiency and performance of automatic code generation for linear model predictive control for the embedded active vibration control of a flexible mechanical structure. μAO-MPC is used to automatically generate C code, based on a high level problem description. The resulting code is used for real-time implementation on an embedded platform. The embedded model predictive controller efficiently computes the voltage input for a piezoceramic actuator based on state estimates from a Kalman filter. The performance, computational efficiency, memory requirements, task execution timing and other properties of practical interest are examined via experiments on the embedded controller for active vibration control.