Control-relevant Identification Research Articles

Echo state networks for online, multi-step MPC relevant identification

This paper presents an online nonlinear system identification algorithm for model predictive control relevant identification, based on an echo state network (ESN) trained using the recursive least squares method with a directional forgetting factor. The proposed online ESN architecture is formed by a reservoir, with fixed recurrent connections, and a reservoir readout mechanism, which is updated online to define the network output. The ESN is used to perform a multi-step prediction task, which can be used for model-predictive control purposes. The proposed online identification algorithm is evaluated in simulation using a conical tank level process and compared with the results of other approaches from the literature. The results show that the proposed algorithm is able to identify the nonlinear characteristics of the process without considering any prior information about it. In addition, the results for long-range predictions are better than the ones obtained with a model linear in its parameters and several baseline ESN models from literature, after the adaptation phase of the proposed algorithm.

Parametric system identification and robust controller design for liquid–liquid heat exchanger system

This study provides a step-by-step analysis of a parametric system identification procedure which provides a singleinput single-output, linear time invariant time-delay model of a liquid-liquid heat exchanger system. Different linear and blockoriented models are used to develop a control relevant identification model. The prediction error method is used to estimate the model parameters. From the simulation results, it can be found that the output error model (linear model) and Hammerstain- Wiener model (block-oriented model) provide the best model validation for a heat exchanger system at 72.23% and 81.5 % goodness-of-FIT, respectively. To achieve the control objectives of the system, the classical proportional-integral-derivative (PID) controller is used. For controller design, the linear model is considered. An industrial process is prone to encounter different uncertainties. Considering the time delay of the estimated linear model as the uncertainty, a ℋ ∞ criterion is used to find out the robust parameters of the PID controller. The developed controller satisfies the sensitivity constraints. Stability boundary locus of (K p ,K i ), (K p ,K d ) and (K d ,K i ) of the PID controller have been plotted to illustrate the robust parameters of the PID controller.

Control-Relevant Design of System Identification Experiments to Improve Behavioral Interventions

This paper presents the design of system identification experiments for physical activity behavioral interventions, utilizing a control-relevant identification test monitoring approach. The goal is to find, in real-time, the shortest possible experiment with sufficient information for identification meeting robust performance conditions. The procedure considers multisinusoidal periodic inputs, and allows modifications on the amplitude and harmonic related frequency content of input signals after each periodic evaluation cycle. The local polynomial method (LPM) is utilized for the computation of transfer function and estimation uncertainties. An improved stopping criterion is formulated relying on robust performance metrics. The procedure is tested through a simulation study relying on a hypothetical system based on a real behavioral intervention.

Bridging the gap between the linear and nonlinear predictive control: Adaptations for efficient building climate control

The linear model predictive control which is frequently used for building climate control benefits from the fact that the resulting optimization task is convex (thus easily and quickly solvable). On the other hand, the nonlinear model predictive control enables the use of a more detailed nonlinear model and it takes advantage of the fact that it addresses the optimization task more directly, however, it requires a more computationally complex algorithm for solving the non-convex optimization problem. In this paper, the gap between the linear and the nonlinear one is bridged by introducing a predictive controller with linear time-dependent model. Making use of linear time-dependent model of the building, the newly proposed controller obtains predictions which are closer to reality than those of linear time invariant model, however, the computational complexity is still kept low since the optimization task remains convex. The concept of linear time-dependent predictive controller is verified on a set of numerical experiments performed using a high fidelity model created in a building simulation environment and compared to the previously mentioned alternatives. Furthermore, the model for the nonlinear variant is identified using an adaptation of the existing model predictive control relevant identification method and the optimization algorithm for the nonlinear predictive controller is adapted such that it can handle also restrictions on discrete-valued nature of the manipulated variables. The presented comparisons show that the current adaptations lead to more efficient building climate control.

EN-PH: a regularization-based model predictive control relevant identification method

Control-Relevant Identification (CRI) proposes to generate models that are suitable to the control problem at hand, creating synergy between the control and identification algorithms. For model predictive control (MPC), it is recommended to use a model obtained by the minimization of a multi-step ahead prediction error cost function. One way of minimizing the cost function is applying the Levemberg-Marquardt (LM) optimization algorithm, provided the identification data set is not ill-conditioned. A second and more interesting approach is the line search method known as PLS-PH. In this work, a regularization-based line search method denoted as EN-PH is proposed to perform the minimization of the identification cost function. Two examples are presented to compare the predictive performance of models obtained using Least Squares, LM, PLS-PH and EN-PH. The examples have shown that the models fitted with EN-PH outperform the other models.

Understanding closed-loop identification with ITCLI

Prior work by the authors (featured in software sessions held during SYSID 2009 and 2012) described ITSIE and ITCRI, a series of interactive tools designed to aid students and practitioners in better understanding the role of design variables associated with classical prediction-error and control-relevant identification, respectively. This paper presents ITCLI (Interactive Tool for Closed-Loop Identification), an interactive software tool for understanding design variables in closed-loop identification of LTI SISO systems using prediction-error techniques. The tool enables an interactive evaluation regarding how bias and variance effects are manifested when identification is performed under closed-loop circumstances. The role of external signal design, choice of model structure, controller tuning during identification testing, and signal injection points (at either the manipulated variable or the setpoint) while in the presence of autocorrelated disturbances can be evaluated in a comprehensive and easy-to-operate tool. The software is developed using Sysquake and is provided as a stand-alone executable version in the Windows and Mac OS operating environments.

Improving performance and stability of MPC relevant identification methods

Model Predictive Control (MPC) Relevant Identification (MRI) methods are a good option for identification, if there is model structure mismatch. Herein a new MRI method, named Enhanced Multistep Prediction Error Method (EMPEM), is proposed. EMPEM combines the best characteristics of others MRI methods in a single algorithm. It was developed to identify either closed-loop or open-loop systems; its convergence and stability make it perform better than the other presented methods. To show the advantages of EMPEM, a comparison is made against two other methods (one MRI and one PEM). The statistical analysis indicates that in the cases studied, the performance and the robustness of the new method is equal or better than the other ones.

A model predictive control approach with relevant identification in dynamic PLS framework

In this paper, a generalized predictive control (GPC) scheme under a dynamic partial least squares (PLS) framework is proposed. At the modeling stage, a model predictive control relevant identification (MRI) approach is used to improve the identification of the model. Within PLS framework, the MIMO system can be automatically decomposed into several SISO subsystems in the latent space. For each subsystem, MRI is implemented and GPC is designed independently. With the advantage of MRI and PLS framework, fewer parameters are needed to be estimated in the identification stage, nonsquare and ill-conditioned system can be handled naturally, control parameters tuning is easier and better control performance can be obtained. Furthermore, the computing time of control action which is very crucial for GPC on-line application decreases since each GPC is running in SISO subsystem in parallel. The results of two simulation examples and a laboratory experiment demonstrate the merit of the proposed method.

Perspectives on control-relevant identification through the use of interactive tools

System identification is an essential part of control design. Control engineers must devote substantial effort to identification issues in order to obtain suitable models for closed-loop control. Control-relevant identification seeks to both simplify the modelling task and improve the usefulness of the model by taking into account controller requirements during system identification. The advantages of this methodology can be better understood and appreciated through the interactive software tool described in this paper. The Interactive Tool for Control Relevant Identification (ITCRI) comprehensively captures the control-relevant identification process for the monovariable problem, from input design to closed-loop control, depicting these stages simultaneously and interactively in one screen. By simultaneously displaying both open- and closed-loop responses of the estimated models, ITCRI enables the user to readily assess how design variable choices during identification and control performance requirements impact model error and ultimately, closed-loop performance. Moreover, the work presents several examples which the aim to illustrate the tool and the considerations that arise when control requirements are taken into account during the identification stage.

Use of partial least squares within the control relevant identification for buildings

Climate changes, diminishing world supplies of non-renewable fuels, as well as economic aspects are probably the most significant driving factors of the current effort to save energy. As buildings account for about 40% of global final energy use, efficient building climate control can significantly contribute to the saving effort. Predictive building automation can be used to operate buildings in an energy and cost effective manner with minimum retrofitting requirements. In such a predictive control approach, dynamic building models are of crucial importance for a good control performance. An algorithm which has not been used in building modeling yet, namely a combination of minimization of multi-step ahead prediction errors and partial least squares will be investigated. Subsequently, two case studies are presented: the first is an artificial model of a building constructed in Trnsys environment, while the second is a real-life case study. The proposed identification algorithm is then validated and tested.

Building modeling: Selection of the most appropriate model for predictive control

Model predictive control has become a widespread solution in many industrial applications and is gaining ground in the field of energy management and automation systems of buildings. A model with reasonable prediction properties is an ultimate condition for good performance of the predictive controller.This paper presents an approach in which a model of a building is selected by an iterative two stage procedure. In the first stage, a minimum set of disturbance inputs is formed so that the resulting model is the best with respect to a defined quality criterion; then the second stage comprises addition of the states to obtain the final minimum set of states maximizing the model quality. The procedure stops when it makes no sense to select more complex model as it brings no more quality improvements. Statistical tests such as the likelihood ratio test, the tests based on cumulative periodogram, the two-sample Kolmogorov–Smirnov test as well as others (fit factor and coefficient of determination) are used to evaluate the relationship between the addition of inputs/states and the model quality. Three identification approaches, namely model predictive control relevant identification, deterministic semi-physical and probabilistic semi-physical modeling are used for estimation of building parameters.Finally, a case study is provided where all the above mentioned approaches are investigated and tested.

ITCRI: An Interactive Software Tool for Evaluating Control-Relevant Identification

This paper describes an Interactive Tool for Control Relevant Identification (ITCRI). It evaluates all stages of a control-relevant identification process, from input design to closed-loop control, simultaneously and interactively in one screen. Control-relevance in ITCRI is accomplished primarily through prefiltering, which is implemented via single-pass and two-step algorithms. By simultaneously displaying both open- and closed-loop responses of the estimated models and important control-relevant validation criteria, ITCRI enables the user to readily assess how design variable choices and control performance requirements impact model error and ultimately, closed-loop performance. A case study involving a fluidized bed calciner is presented to illustrate the workings of the tool and the considerations that arise when control requirements are incorporated during identification.

ITCRI: An Interactive Software Tool for Control-Relevant Identification Education

AbstractThis paper describes the theoretical basis, features, and functionality of an Interactive Tool for Control Relevant Identification (ITCRI). ITCRI captures the control-relevant identification process, from input design to closed-loop control, depicting these stages simultaneously and interactively in one screen. Control-relevance in ITCRI is accomplished primarily through prefiltering, which is evaluated using single-pass and two-step algorithms. By displaying both open- and closed-loop responses of the estimated models and important control-relevant validation criteria, ITCRI enables the user to readily assess how design variable choices, control performance requirements and model error can impact the achievable closed-loop performance from a restricted complexity model estimated under noisy conditions.

Model predictive control relevant identification: multiple input multiple output against multiple input single output

Model predictive control (MPC) relevant identification (MRI) can be performed by minimising a cost function containing identification errors from multistep ahead predictions. The multiple MISO (multiple input single output) identification approach is often preferred to the MIMO (multiple input multiple output) identification approach in traditional one-step ahead identification. This study aims at comparing the MIMO and multiple MISO identification approaches in MRI. It is argued in this study that, unlike in the one-step ahead approach, MIMO identification is preferable in MRI. As an example, a non-linear MIMO proton exchange membrane fuel cell (PEMFC) is approximated in the neighbourhood of an operating point using the MIMO and MISO approaches.

Control Relevant Blind Identification of Disturbances With Application to a Multivariable Active Vibration Isolation Platform

This paper deals with the identification of root causes of disturbances in multivariable systems. It is shown that this boils down to a blind identification problem that can be solved, under additional assumptions, within certain indeterminacies. Using the results from identification, it is illustrated how sources can be physically interpreted and the location of the sources can be recovered. Also, a design tool is developed to select multivariable feedback controller candidates. The practical feasibility is demonstrated on an industrial multivariable controlled active vibration isolation platform.

PLS-based model predictive control relevant identification: PLS-PH algorithm

Abstract Control-relevant identification produces a model by minimizing a cost function that is commensurate with the control cost function. This paper focuses on model predictive control (MPC); thus, a multi-step ahead prediction error cost function is minimized. Numerical optimization algorithms such as Levenberg-Marquardt can be used to minimize the non-linear identification cost function provided the identification data set is not ill-conditioned. A PLS-based line search numerical optimization approach denoted PLS-PH is proposed to tackle the minimization of the identification cost function in case the identification data set is ill-conditioned. PLS-PH fits a MIMO linear model to an identification data set that may be ill-conditioned. Two chemical processes are identified to compare predictive performance of models obtained using Least Squares, Levenberg-Marquardt, and PLS-PH. The two examples show that the models fitted with PLS-PH outperform the other models if the identification data set is ill-conditioned.

Integrating virtual reference feedback tuning into a unified closed-loop identification framework

This paper compares particular instances of control-relevant identification, closed-loop identification, and controller identification (either in an actual loop or in a virtual reference feedback tuning approach, VRFT) problems. Significant similarities appear between them which allow, in particular, alternative formulations of the VRFT methodology in a differentiable setting.

Simple control-relevant identification test methods for a class of ill-conditioned processes

The identification of a class of ill-conditioned processes is considered. The goal of identification is model predictive control (MPC). For this class of processes, it is essential to have good estimation of the very small difference between certain transfer functions, or, low gain direction. Two simple and practical test methods will be proposed that can enhance the model quality of low gain direction. The main idea is to use the high amplitude and high correlation test signals that excite the low gain direction more and that will not disturb process operation. The test signals can be used in both open-loop and closed-loop tests. A high purity distillation column model is used to show the effectiveness of the methods.

CONTROL-RELEVANT DEMAND MODELING FOR SUPPLY CHAIN MANAGEMENT

The development of control-oriented decision policies for inventory management in supply chains has received considerable interest in recent years, and demand modeling to supply forecasts for these policies is an important component of an effective solution to this problem. Drawing from the problem of control-relevant identification, we present an approach for demand modeling based on data that relies on a control-relevant prefilter to tailor the emphasis of the fit to the intended purpose of the model, which is to provide forecast signals to a tactical inventory management policy based on Model Predictive Control. Integrating the demand modeling and inventory control problems offers the opportunity to obtain reduced-order models that exhibit superior performance, with potentially lower user effort relative to traditional “open-loop„ methods. A systematic approach to generating these prefilters is presented and the benefits resulting from their use are demonstrated on a representative production/inventory system case study.

MISO-Structure-Based Control-Relevant Identification of MIMO Systems with Steady-State Gain Matching

The prefilter-based control-relevant identification (CRI) scheme proposed by Rivera et al. [IEEE Trans. Autom. Control 1992, 37 (7), 964−974] for single-input/single-output (SISO) systems estimates a model of the system that matches the true system dynamics in a range of frequencies that are important and relevant for closed-loop performance. In this paper, an extension of this approach is proposed for the CRI of multi-input/multi-output (MIMO) systems by partitioning the overall system into individual multi-input/single-output (MISO) structures. Furthermore, CRI methodologies proposed hitherto have not examined the issue of accurate identification of the steady-state gain. In this paper, the importance of achieving good steady-state gain match is emphasized, and a method is proposed to identify a control-relevant model that has a good match, at steady state (zero frequency), as well as in the control-relevant frequency band, with the true plant. The effect of the proposed control-relevant prefiltering on the identification of the high- and low-gain directions is also analyzed in this paper. The proposed methodology is validated using simulations on representative problems from the literature and on a paper machine problem.