No abstract

The self-tuning regulator (STR) is a popular adaptive control algorithm. A high-performance computer is required for its implementation due to the heavy online computational burden. To extend STR for more realtime applications, a parallel hardware implementation on a low-cost reconfigurable computer is presented. The hardware was incorporated with multistage matrix multiplication (MMM) and trace technique to enhance the processing speed. This design was deeply pipelined to achieve high throughput. The algorithm was prototyped on a Xilinx field-programmable gate array (FPGA) device with a maximum operating frequency of 210.436 MHz. Application-specific integrated circuit (ASIC) implementation of STR was reported. ACM Reference Format:Ananthan, T. and Vaidyan, M. V. 2013. A reconfigurable parallel hardware implementation of the self-tuning regulator.

Abstract-A Computer aided implementation of an estimator based inverse dynamics controller (EBIDC) algorithm for the efficient control of the states, especially the non-measurable states of an autonomous hybrid system (AHS) is proposed in this paper. Case of a cylindrical three-tank hybrid system is demonstrated. The studies combining behaviors of both discrete event and continuous systems are of utmost importance nowadays and thus study of dynamics of hybrid systems and control strategies for performance improvement is very important in the present scenario. This algorithm makes use of a state estimator for estimating all the states of the system which uses a generic nonlinear model of the actual system. The efficacy of the developed EBIDC is demonstrated by conducting simulation studies using Matlab under different operating modes of the system. Both set point tracking and disturbance rejection capabilities of the proposed controller are also confirmed in this work. The robustness of EBIDC in controlling the non-measurable states is also analyzed under different uncertainties.Index Terms-Autonomous hybrid systems, derivative-free state estimator, estimator based inverse dynamics controller, unscented kalman filter (UKF) in state estimation. I. INTRODUCTIONHybrid system is a very dynamic branch of engineering which deals with the systems of continuous, discrete time and discrete event behavior. In some of the earlier works reported, theories behind hybrid automata and the discrete control of continuous plant were considered [1], [2]. Later many approaches were reported in the literature in which the systems were considered to contain both continuous and discrete behavior and control of such systems were accomplished [3]- [16]. Subsequently studies like Lyapunov technique, applied for continuous nonlinear systems were extended to hybrid systems in [4]. Linear approximations of nonlinear systems were carried out in most of such studies as in [4] and [5]. In [6], with the help of constraint logic programming with interval arithmetic, a rigorous modeling, simulation and analysis approach for hybrid systems were proposed. A collection of modeling and classification of different benchmark hybrid systems are presented in [7]. However, direct use of nonlinear models for analysis and control of hybrid systems were reported in very few works and still it is a challenging area of research.The potential of dynamic models of hybrid systems developed in state estimation along with the modeling was investigated in [10]-[14] and [16]. Recently, using Bayesian approach, a nonlinear model of hybrid system was proposed by combining multiple local linear models in [11].Using such model a novel state identification method and predictive control of the system were proposed for a class of hybrid systems.[12], [13] had recommended an ensemble Kalman filter (EnKF) based nonlinear model predictive control (NMPC) and UKF based fault tolerant control schemes for controlling the output variables of the AHS. In addition to UKF and EnKF bas...

No abstract

No abstract