Grey-box Model Identification Research Articles

Distributed agent-based building grey-box model identification

The paper deals with an identification (calibration) of a building thermal model as a crucial part of modern control algorithms. It describes a modelling issue and parameter identification approach. There is presented an identification approach using a so called dual decomposition method which decomposes a large optimization problem into smaller local ones which are then solved by local agents. The local models are found using a grey-box calibration and by a coordination of agents’ mutual shared parameters a global consistency is obtained.The proposed method is tested on two examples; one is basically trivial, in the latter one, more complex, a model of a building simulated in the EnergyPlus program is obtained and offers promising results for (predictive) control-oriented purposes.

Grey-box model identification of temperature dynamics in a photobioreactor

This article presents a general strategy for grey-box model identification and deals with some issues that might be present in real life applications. An Unscented Kalman Filter (UKF) is used to train a grey-box temperature model with experimental data from an internally illuminated photobioreactor. The model structure is derived by means of heat balance analysis with the aid of a heat flow diagram. Then, the model is discretized and given an alternative state space representation in such a way that parameters can be readily estimated with an UKF. In order to avoid performance degradation and to improve the stability of the UKF algorithm, the prediction error covariance matrix is estimated and the state covariance matrix square root is calculated with a method based on Schur spectral decomposition to ensure positive semi-definiteness.

Impact of the Heat Emission System on the Identification of Grey-box Models for Residential Buildings

Grey-box modelling is a system identification technique that combines parameter estimation methods to quantify the model parameters and physical knowledge to define the model structure. Consequently, the potential of grey-box modelling lies, in addition to the application of reduced-order models in e.g. control strategies or district simulations, in the characterization of the thermal properties of buildings. Nevertheless, the quality of the obtained model is governed by the dynamic information that is available in the training data. Thereby, the required accuracy of the models, and thus the requirements of the training data, differ from application to application.This paper analyses how the significant difference in dynamic behavior of a slow floor heating system compared to a highly responsive radiator heating system results in a variation of the time constants of the building that are excited by the system and therefore identifiable in the system identification process. The performance of grey-box models for prediction and simulations are contrasted for cases with radiators and floor heating and the physical interpretability of the model parameters is demonstrated.

A grey-box model identification of an advanced oxidation process for wastewater treatment

This paper deals with a grey-box model identification of a catalytic ozonation pilot process used in wastewater treatment. The grey-box model is presented. For the parametric estimation, a non convex optimization problem must be solved. Then, we propose to solve this problem, in practice, by using a Levenberg-Marquardt algorithm applied to a composite criterion based on a priori knowledge which imposes a physical meaning to the solution. The method is applied to experimental data.

A Practical Approach to Grey-box Model Identification

AbstractToday's computing power can provide a way how to fit physically well founded models to a given experimental data. The combination of user's subjective information and objective data (grey-box modeling) is especially profitable when obtaining sufficiently excited data is impracticable or costly. This is typical in the industrial practice. However a simple prediction error fitting scheme can results in poor estimate of parameters. In this article, a new method how to improve parameters estimate to be relevant in commissioning model predictive control, is proposed. The method is derived on the base of maximum likelihood estimation and demonstrated using a practical application.

Grey-box model identification – control relevant approach

Grey-box modeling is an advantageous tool for system identification when obtained input/output experimental data are insufficiently excited. The lack of information in the data can be often replaced with some additional knowledge about the modeled system, which constricts the class of models under consideration. The real system is usually more complex and do not fit the model class, thus a bias error occurs. The main goal of this paper is to show an effective way how to identify grey-box models, which would be relevant in commissioning predictive control.

EFFICIENT PARAMETERIZATION FOR GREY-BOX MODEL IDENTIFICATION OF COMPLEX PHYSICAL SYSTEMS

Grey box model identification preserves known physical structures in a model but with limits to the possible excitation, all parameters are rarely identifiable, and different parametrizations give significantly different model quality. Convenient methods to show which parameterizations are the better would be very useful. This paper shows how we can assess the parameter interdependence and model quality. Hessian matrix decomposition is employed to show linear dependencies between variables and to put a quality tag on different parameterizations. The method determines parameter relations that need be constrained to achieve satisfactory convergence. Identification of nonlinear models for a ship illustrate the concept.

Grey-box model identification via evolutionary computing

Abstract This paper presents an evolutionary grey-box model identification methodology that makes the best use of a priori knowledge on a clear-box model with a global structural representation of the physical system under study, whilst incorporating accurate black-box models for immeasurable and local nonlinearities of a practical system. The evolutionary technique is applied to building dominant structural identification with local parametric tuning without the need of a differentiable performance index in the presence of noisy data. It is shown that the evolutionary technique provides an excellent fitting performance and is capable of accommodating multiple objectives such as to examine the relationships between model complexity and fitting accuracy during the model building process. Validation results show that the proposed method offers robust, uncluttered and accurate models for two practical systems. It is expected that this type of grey-box models will accommodate many practical engineering systems for a better modelling accuracy.

Model-based fault detection and isolation for a gas–liquid separation unit

A complete fault detection and isolation system is designed for a gas–liquid separation unit. It involves the determination and identification of grey box models, the design of a model-based residual generator, and finally the evaluation of the residuals via a set of statistical tests. The latter are cumulative sum (CUSUM) tests which are combined in such a way that both fault detection and fault isolation can be achieved. The performance of the resulting diagnosis system, such as missed alarm rate, wrong isolation rate and mean detection delay, are studied via simulations.

Experiment design for grey box identification

AbstractGrey box models are characterized by their physical significance e.g. in parametrization and by the partial prior information that is available about e.g. the parameter values. These aspects of the grey box model affect the design of optimal excitations for identification and we study the extension of classical theory for experiment design to input design for identification of grey box models. Partial prior information is expressed as a probability distribution and is employed in the design of optimal excitations through optimization of Bayesian criteria.