Self-tuned Model-Based Predictive Control Using Evolving Fuzzy Model of a Non-linear Dynamic Process

This paper discusses the model-based predictive controller design of networked nonlinear systems with communication delay and data loss. Based on the analysis of the closed-loop networked predictive control systems, the model-based networked predictive control strategy can compensate for communication delay and data loss in an active way. The designed model-based predictive controller can also guarantee the stability of the closed-loop networked system. The simulation results demonstrate the feasibility and efficacy of the proposed model-based predictive controller design scheme.

Nonlinear model predictive control for the space heating system of a university building in Norway

Fuzzy predictive control integrates conventional model-based predictive control with techniques from fuzzy multicriteria decision making, and translates the goals and the constraints to predictive control in a transparent way. The information regarding the (fuzzy) goals and the (fuzzy) constraints of the control problem is combined by using a decision function from the fuzzy set theory. This paper investigates the use of fuzzy decision making in predictive control and compares the results to those obtained from conventional model-based predictive control. Experiments on a nonminimum phase, unstable linear system and air conditioning system with nonlinear dynamics are studied. It is shown that the performance of the model-based predictive controller can be improved by the use of fuzzy goals and criteria with fuzzy decision-making techniques.

The class of model-based predictive controllers (MBPCs) has been quite successful at overcoming many of the well-known difficulties inherent in nonminimum phase (NMP) systems, e.g., flexible structures, and among the various MBPCs, Unified Predictive Control (UPC) was selected for the control tasks in this paper. Two methods are used to model the tip force of a flexible manipulator. Modal expansion (summation) is a pole but not zero preserving approach, while Infinite Product Expansion preserves both poles and zeros of a system. It is shown that in a flexible structure framework the difference between the exact and modal expansion zeros is significant. The effect of transmission zero mismatch will be observed in a closed loop system, especially when a model based controller is used. Simulation results show that when a UPC is designed based on the modal expansion model, it destabilizes the closed loop system when an exact model is used as the plant. This suggests a severe problem in implementing the designed UPC experimentally. The use of an exact model in a model-based controller design procedure, such as UPC, is therefore suggested. (Author)

A type of recurrent neuro-fuzzy network is proposed in this paper to build long-term prediction models for nonlinear processes. The process operation is partitioned into several fuzzy operating regions. Within each region, a local linear model is used to model the process. The global model output is obtained through the centre of gravity defuzzification which is essentially the interpolation of local model outputs. This modeling strategy utilizes both process knowledge and process input output data. Process knowledge is used to initially divide the process operation into several fuzzy operating regions and to set up the initial fuzzification layer weights. Process input output data are used to train the network. Network weights are such trained so that the long-term prediction errors are minimized. Through training, membership functions of fuzzy operating regions are refined and local models are learn. Based on the recurrent neuro-fuzzy network model, a novel type of nonlinear model-based long range predictive controller can be developed and it consists of several local linear model-based predictive controllers. Local controllers are constructed based on the corresponding local linear models and their outputs are combined to form a global control action by using their membership functions. This control strategy has the advantage that control actions can be calculated analytically avoiding the time consuming nonlinear programming procedures required in conventional nonlinear model-based predictive control. The techniques have been successfully applied to the modeling and control of a neutralization process.

This paper is concerned with a computationally efficient (suboptimal) nonlinear model-based predictive control (MPC) algorithm and its application to a high-purity high-pressure ethylene-ethane distillation column. A neural model of the process is used on-line to determine the local linearisation and the nonlinear free response. In comparison with general nonlinear MPC technique, which hinges on non-convex optimisation, the presented structure is far more reliable and less computationally demanding because it results in a quadratic programming problem, whereas its closed-loop control performance is similar.

The present article discusses the possibility to reduce the computational effort of complex control algorithms by neuro-fuzzy systems. Thereby, great potentials can be released, especially in the automotive sector. A limiting factor for the design of control algorithms is the task of a real-time execution on cost-optimized control units [1]. The influence of this limitations can be reduced by neuro-fuzzy systems. This is shown exemplary for the model-based predictive control of the roll motion presented by Sieberg et al. [2]. The controller based on the adaptive neuro-fuzzy inference system is validated regarding the control quality and the computational effort. Thus it is compared to the origin model-based predictive control algorithm. The implementation and validation are based on a co-simulation of MATLAB/SIMULINK and IPG CarMaker.

Model Predictive Control in Milling based on Support Vector Machines

Energy-efficient fuzzy model-based multivariable predictive control of a HVAC system

This article studies the application of nonparametric modeling approach to model magnetorheological (MR) dampers. For comparison purposes, another typical parametric modeling method for electrorheological (ER) and MR dampers is reviewed. The existing parametric MR damper model includes a stiff Bouc-Wen model that is not friendly for simulation study and real time implementation of model-based advanced control algorithms. In order to avoid the difficulties by using the existing parametric model, the test data from a commercialized MR damper is employed to develop nonparametric models, which can consist of a series of numerically efficient mathematic functions. In addition, the selected functions are required to be continuous and differentiable for potential model-based control algorithms. The results of the nonparametric models show that such different models are comparable. Furthermore, one nonparametric model is selected to be compared with a parametric model and the test data to illustrate the accuracy of the model. The comparison shows that the proposed nonparametric models are able to accurately predict the damper force characteristics, damper bilinear behavior, hysteresis, and electromagnetic saturation. It is further shown that the nonparametric models can be numerically solved with an integration step size of the order of 10 2 s, much faster than the parametric models of the order of 10 5 s, which clearly shows that the proposed nonparametric models are feasible even for real time model-based control algorithms.

This paper describes an algorithm for solving the optimization problem which occurs in a model-based predictive control (MPC) algorithm for discrete time bilinear models. This optimization problem is nonlinear in general, because the model acts as a nonlinear equality constraint. Common approaches of performing such a nonlinear optimization problem boil down to (successively) approximating the nonlinear objective function, followed by performing a line search. In this paper it is demonstrated that the structural properties of the bilinear state space model enable to formulate the nonlinear optimization problem as a sequence of quadratic programming problems which exactly represent the original objective function, implying that no additional line search is needed. The proposed optimization algorithm is compared to one that is based on linearization around an input trajectory. To benefit from the advantages of both algorithms, a hybrid algorithm is proposed, which outperforms the other two in most cases.

The following paper presents an improved self-tuned Model-based Predictive Controller (MPC) with constraints through an Extended Kalman Filter (EKF). The algorithm is tested in a Continuous Stirred Tank Reactor (CSTR) process for temperature and molar concentration regulations. The proposed method was developed from a previous work [1], where the MPC was tuned from error and control weights of equal value. A posterior analysis demonstrated that the MPC tuning was not robust. Sometimes it generated negative weight values due to the EKF convergence and the algorithm cannot always assure a global minimum value at performance index. Therefore, this work considers this problem and always generates positive weights which assures a minimum global value in the cost function of the MPC.

This paper presents an optimal control scheme based on model-based predictive control (MPC) for a wheeled mobile robot (WMR) with nonholonomic constraints. It is shown that, by using MPC, some advantages can be obtained, such as the ability to handle constraints due to state or input limitations and performance improvement. To solve some problems with other MPCs for WMRs, this paper proposes to formulate a cost function in polar coordinates. Considerations regarding the computational effort of the MPC are developed with the purpose of analysing the viability of the proposed technique in real-time.

The growth of distributed solar photovoltaic (PV) generation in low-voltage grids in recent years has raised concerns about power quality disruptions introduced by solar variability. In response, some electricity utilities have introduced restrictions to new PV installations, such as imposing limits to power ramp rates and exports to the grid. These restrictions effectively require PV installations to locally compensate rapid variations in power output, such as those arising from passing clouds, by using batteries, for example. The direct control of responsive loads to follow variations in solar capacity is a cost-effective alternative to batteries to mitigate impacts to the grid. This paper explores the application of model-based predictive control (MPC) of a population of air conditioners (ACs) to achieve optimal tracking of PV generation to compensate solar variability at a substation or distribution transformer aggregation level. The proposed approach exploits an existing model for aggregate AC demand to design an MPC scheme to modulate the demand of clusters of ACs to track fluctuations in PV capacity. The effectivity of the proposed MPC scheme to achieve AC load following and regulation at different levels of PV penetration is analysed in a numerical simulation study, which shows that solar variability can be reduced by 80 to 95% at PV capacities between 25 and 10% of the peak AC demand with minimal impact to end-user comfort levels.

Model-based predictive signal control is a popular method to pro-actively control traffic and to reduce the effects of congestion in urban networks. In combination with structure-free controllers, which adapt signal settings in arbitrary order and combination (no imposed cycles), these predictive control methods have a high potential to increase system performance by adapting to individual vehicle patterns, which are increasingly available due to new technology. However, most of these control methods assume perfect predictions, while in practice there are prediction errors due to various reasons. In this paper, the sensitivity of the system performance to these prediction errors is analyzed, for an urban corridor with spillback.In a microscopic simulator, first the ideal world is created for the structure-free model-based predictive signal controller, in which perfect predictions are made and the controller can reach its optimal performance. Then prediction errors are introduced in this perfect world, distinguished in aggregation errors that arise using a macroscopic prediction model and biases that represent structural errors in the prediction model or in its demand and state input. The effects of these prediction errors on the system performance are analyzed, as a function of the prediction horizon and update frequency of the control system.The results show that, even under errors, longer prediction horizons lead to better performance, up to a certain optimal prediction horizon length. A high update frequency dampens the influence of prediction errors, enabling the structure-free controller to correct mistakes faster. However, there remains a significant performance loss due to aggregation errors and biases in the prediction model, indicating a promising performance gain of more reliable predictions and the incorporation of information on individual vehicles in future control applications. Moreover, for all model quantities one direction of the bias has more impact on the system performance than the other direction, indicating guidelines towards a more robust control system that suffers less from erroneous predictions.