Identification of Multiple‐Mode Linear Models Based on Particle Swarm Optimizer with Cyclic Network Mechanism

Abstract

This paper studies the metaheuristic optimizer-based direct identification of a multiple-mode system consisting of a finite set of linear regression representations of subsystems. To this end, the concept of a multiple-mode linear regression model is first introduced, and its identification issues are established. A method for reducing the identification problem for multiple-mode models to an optimization problem is also described in detail. Then, to overcome the difficulties that arise because the formulated optimization problem is inherently ill-conditioned and nonconvex, the cyclic-network-topology-based constrained particle swarm optimizer (CNT-CPSO) is introduced, and a concrete procedure for the CNT-CPSO-based identification methodology is developed. This scheme requires no prior knowledge of the mode transitions between subsystems and, unlike some conventional methods, can handle a large amount of data without difficulty during the identification process. This is one of the distinguishing features of the proposed method. The paper also considers an extension of the CNT-CPSO-based identification scheme that makes it possible to simultaneously obtain both the optimal parameters of the multiple submodels and a certain decision parameter involved in the mode transition criteria. Finally, an experimental setup using a DC motor system is established to demonstrate the practical usability of the proposed metaheuristic optimizer-based identification scheme for developing a multiple-mode linear regression model.

Prediction of Compressive Strength of Concrete in Wet‐Dry Environment by BP Artificial Neural Networks

“…Relationship between multiple input variables (explanatory variables) and one output variable (a response variable) also can be expressed by the linear regression model [34][35][36]. Linear regression model is a statistical model determined as follows:…”

Piecewise affine identification of a hydraulic pumping system using evolutionary computation

“…In [28, 29], evolutionary approaches were used to estimate submodel parameters based on the prediction error. This is known to induce bias in a number of situations and model classes when proper techniques are not employed [30, 31].…”

“…This is known to induce bias in a number of situations and model classes when proper techniques are not employed [30, 31]. Besides, the optimisation problem formulated in [28, 29] required the implementation of non‐conventional EA and no partition classifier was attained. Since the proposed identification approach is decomposed into sub‐problems, the EAs are in charge of estimating only the Gaussian mixture model (GMM) parameters, hence rendering the algorithm less sensitive to scalability problems.…”

“…The PWA identification procedures can be classified into five categories [13]: optimisation, clustering, algebraic and recursive methods. Among them, the following approaches stand out: (i) procedure based on mixed integer programming [18]; (ii) methods based on clustering [12, 17, 34]; (iii) Bayesian approach [20]; (iv) bounded error procedure [19]; (v) evolutionary approaches [28, 29] and (vi) recursive or on‐line algorithms [13, 24, 35, 36]. A simple and efficient clustering method based on GMMs, trained by the expectation‐maximisation algorithm – presented in [21] – will be used as a framework in what follows.…”

Application of Simulated Annealing Particle Swarm Optimization Based on Correlation in Parameter Identification of Induction Motor

“…The correlation algorithm is used in this paper to extract the amplitude and phase of the stator current fundamental signal, and the harmonic components are filtered out to emphasize the fundamental signal given the characteristics of the stator voltage and current fundamental frequency equal to construct the same frequency with the stator voltage reference signal [10][11][12]. Then, to identify the parameters of the dynamic mathematical model of asynchronous motor, a novel optimization algorithm incorporating simulated annealing with the advantages of particle swarm optimization is used [13][14][15]. Some particles in the particle swarm get a new target position by flight in the identification iteration.…”