On-line Identification Control Research Articles

Hybrid thermal modeling of additive manufacturing processes using physics-informed neural networks for temperature prediction and parameter identification

Understanding the thermal behavior of additive manufacturing (AM) processes is crucial for enhancing the quality control and enabling customized process design. Most purely physics-based computational models suffer from intensive computational costs and the need of calibrating unknown parameters, thus not suitable for online control and iterative design application. Data-driven models taking advantage of the latest developed computational tools can serve as a more efficient surrogate, but they are usually trained over a large amount of simulation data and often fail to effectively use small but high-quality experimental data. In this work, we developed a hybrid physics-based data-driven thermal modeling approach of AM processes using physics-informed neural networks. Specifically, partially observed temperature data measured from an infrared camera is combined with the physics laws to predict full-field temperature history and to discover unknown material and process parameters. In the numerical and experimental examples, the effectiveness of adding auxiliary training data and using the pretrained model on training efficiency and prediction accuracy, as well as the ability to identify unknown parameters with partially observed data, are demonstrated. The results show that the hybrid thermal model can effectively identify unknown parameters and capture the full-field temperature accurately, and thus it has the potential to be used in iterative process design and real-time process control of AM.

Reinforcement Learning-Based Composite Optimal Operational Control of Industrial Systems With Multiple Unit Devices

This article investigates the optimal operational control (OOC) problem for a class of industrial systems consisting of multiple unit devices with fast dynamics and an unknown operational process with slow dynamics. First, the OOC problem is formulated as a noncascade optimal control problem of two-time-scale systems with a novel performance function. Second, using singular perturbation theory, a decentralized composite control scheme is proposed by decomposing the original optimal problem into reduced-order fast and slow subsystem problems. Then, in the framework of reinforcement learning, an online controller design method for the slow subsystem is proposed by using the online measurement, and an offline controller design for the fast subsystem is proposed by using the unit device models. The obtained decentralized composite optimal controller achieves both the desired operational index tracking and disturbance rejection without requiring the dynamics of the operational process. Different from the existing cascade design methods, the proposed approach regulates the unit devices and operational process simultaneously, as well as overcomes the potential high dimensionality and ill-conditioned numerical issues. Finally, a mixed separation thickening process and a numerical example are given to illustrate the presented results.

Reduced basis method for managed pressure drilling based on a model with local nonlinearities

AbstractTo circumvent restrictions of conventional drilling methods, such as slow control actions and inability to drill depleted reservoirs, a drilling method called managed pressure drilling (MPD) has been developed. In MPD, single‐phase flow processes can be modeled as a feedback interconnection of a high‐order linear system and a low‐order nonlinear system. These nonlinearities appear locally both inside and at the boundaries of the computational domain. To obtain a fast simulation platform for real‐time purposes (eg, online model‐based controller implementation), model order reduction is required for MPD. However, the local nonlinearities render applying model order reduction techniques challenging. In this study, a new approach is proposed to deal with such nonlinearities within the reduced basis (RB) context and it is successfully tested on a model for MPD. Contrary to the classical RB technique, the proposed approach not only does not generate nonphysical spikes at the locations of these local nonlinearities but also yields high speedup factors. The obtained reduced‐order model can be used for efficient online simulation and controller design for drilling systems with MPD.

Interval estimation and optimization for motion trajectory of overhead crane under uncertainty

The parameter uncertainty has an important effect on the motion planning of overhead cranes, especially in relation to its industrial safety of production activities. Thus, a novel uncertain estimation-and-optimization strategy is proposed for motion planning of overhead cranes with uncertainty in this paper. The main work of this paper includes the following aspects. First, the overhead crane is simplified as a double pendulum model and the corresponding motion planning is described as an optimal control problem with uncertainty. Second, uncertainties are expressed as interval parameters where only the upper and lower bounds are required without probability information and a bounds estimation problem for optimal control with uncertainty is established; the solution contains all possible values. Third, the bounds estimation problem is solved by a surrogate model-based method, the motion trajectory intervals of overhead cranes are obtained. Fourth, in order to reduce the influence of uncertainty on the motion planning of overhead cranes, an optimization method is introduced to reduce the sensitivity to uncertainty. Finally, the numerical examples show that high accurate interval estimation results are obtained with a reasonable computational cost, and the sensitivity of motion trajectory to uncertainty is reduced obviously with the help of optimization. The proposed strategy provides a guidance for uncertain analysis and online controller design of overhead cranes.

Doubly Fed Induction Generator (DFIG)-Based Wind Farm Control Framework for Primary Frequency and Inertial Response Application

This paper presents a new wind farm control framework for inertial and primary frequency response for a high wind integrated power system. The proposed architecture is unique in the sense that the methodology can be used for frequency regulation support during subsynchronous and super-synchronous operation of the wind turbines (farm). The architecture work with existing wind farm controllers thus avoiding any additional replacement or tuning. The methodology depends on reduced order modeling based on model order reduction (MOR) and subsequent online controller design. The approach is tested on a smaller wind farm and further evaluated on a larger reduced power grid with 39 buses and ten generators. The results show that the proposed architecture provides greater flexibility in wind farm control towards frequency oscillations.

Online Learning ARMA Controllers With Guaranteed Closed-Loop Stability.

This paper presents a novel online block adaptive learning algorithm for autoregressive moving average (ARMA) controller design based on the real data measured from the plant. The method employs ARMA input-output models both for the plant and the resulting closed-loop system. In a sliding window, the plant model parameters are identified first offline using a supervised learning algorithm minimizing an ε -insensitive and regularized identification error, which is the window average of the distances between the measured plant output and the model output for the input provided by the controller. The optimal controller parameters are then determined again offline for another sliding window as the solution to a constrained optimization problem, where the cost is the ε -insensitive and regularized output tracking error and the constraints that are linear inequalities of the controller parameters are imposed for ensuring the closed-loop system to be Schur stable. Not only the identification phase but also the controller design phase uses the input-output samples measured from the plant during online learning. In the developed online controller design method, the controller parameters can always be kept in a parameter region providing Schur stability for the closed-loop system. The ε -insensitiveness provides robustness against disturbances, so does the regularization better generalization performance in the identification and the control. The method is tested on benchmark plants, including the inverted pendulum and dc motor models. The method is also tested on an emulated and also a real dc motor by online block adaptive learning ARMA controllers, in particular, Proportional-Integral-Derivative controllers.

On-line direct control design for nonlinear systems

An approach to design a feedback controller for nonlinear systems directly from experimental data is presented. Improving over a recently proposed technique, which employs exclusively a batch of experimental data collected in a preliminary experiment, here the control law is updated and refined during real-time operation, hence enabling an on-line learning capability. The theoretical properties of the described approach, in particular closed-loop stability and tracking accuracy, are discussed. Finally, the experimental results obtained with a water tank laboratory setup are presented.

A methodology for identification and control of electro-mechanical actuators

Mechatronic systems are fully-integrated engineering systems that are composed of mechanical, electronic, and computer control sub-systems. These integrated systems use electro-mechanical actuators to cause the required motion. Therefore, the design of appropriate controllers for these actuators are an essential step in mechatronic system design. In this paper, a three-stage methodology for real-time identification and control of electro-mechanical actuator plants is presented, tested, and validated. First, identification models are constructed from experimental data to approximate the plants’ response. Second, the identified model is used in a simulation environment for the purpose of designing a suitable controller. Finally, the designed controller is applied and tested on the real plant through Hardware-in-the-Loop (HIL) environment. The described three-stage methodology provides the following practical contributions:•Establishes an easy-to-follow methodology for controller design of electro-mechanical actuators.•Combines off-line and on-line controller design for practical performance.•Modifies the HIL concept by using physical plants with computer control (rather than virtual plants with physical controllers).Simulated and experimental results for two case studies, induction motor and vehicle drive system, are presented in order to validate the proposed methodology. These results showed that electromechanical actuators can be identified and controlled using an easy-to-duplicate and flexible procedure.

A novel trajectory-based online controller design approach to fault accommodation in NREL’s 5MW wind turbine systems

This paper presents a real-time mechanism to tolerate faults occurring in a wind turbine (WT) system. This system is composed of a FAST coded simulator designed by the U.S. National Renewable Energy Laboratory. The demonstrated mechanism lies under the taxonomy of active fault-tolerant control (FTC) systems, namely online redesign based approach. In the proposed approach, we do not use any a priori information about the model of the turbine in real-time. In fact, we use online measurements generated by the WT. Based on the given control specifications, and the observed measurement an occurred fault is accommodated by reconfiguring the online controller such that the WT generates rated power even under faulty conditions. Second, no explicit fault diagnosis (FD) module is used in this approach. As a result, issues of model uncertainty, false alarms, etc. associated with an integrated FD and controller reconfiguration approach to FTC systems are not experienced here.

Distributed Particle Swarm Optimization for limited-time adaptation with real robots

SUMMARYEvaluative techniques offer a tremendous potential for online controller design. However, when the optimization space is large and the performance metric is noisy, the overall adaptation process becomes extremely time consuming. Distributing the adaptation process reduces the required time and increases robustness to failure of individual agents. In this paper, we analyze the role of the four algorithmic parameters that determine the total evaluation time in a distributed implementation of a Particle Swarm Optimization (PSO) algorithm. For an obstacle avoidance case study using up to eight robots, we explore in simulation the lower boundaries of these parameters and propose a set of empirical guidelines for choosing their values. We then apply these guidelines to a real robot implementation and show that it is feasible to optimize 24 control parameters per robot within 2 h, a limited amount of time determined by the robots' battery life. We also show that a hybrid simulate-and-transfer approach coupled with a noise-resistant PSO algorithm can be used to further reduce experimental time as compared to a pure real-robot implementation.

Coordinating fuzzy control of the sintering process

Abstract The sintering process is an important step in the iron and steel making process and features strong complexity, nonlinearity and large time-delay. This paper presents a coordinating control method that two fuzzy controllers are designed to control the burning through point (BTP) and the bunker-level. First, the BTP control is the main control task with the sinter strand speed as the control variable, and a predictive fuzzy control scheme is developed where an intermediate process model is used to perform on-line controller design to meet a closed-loop performance specification that is associated with desired process output. Then, a fuzzy controller is designed to control the charging bunker-level via adjusting the strand speed. The two fuzzy controllers are proposed for these two closed-loops and it is shown that the satisfactory response of the closed-loop system can be achieved by handling the strand speed. Results of simulation and actual running illustrate the performance of the proposed method.

MIMO adaptive vibration control of smart structures with quickly varying parameters: Neural networks vs classical control approach

This paper presents experimental adaptive identification and control of smart structures using neural networks based on system classification technique. An inverted L-structure with surface-bonded piezoceramic sensors/actuators is used for analysis. The state space, as well as matrix fraction description presentation, from control input voltages to output sensor voltage, is established in multivariable form. It is observed that the computational time required for online parameter identification and controller design is generally quite high. For the system, whose parameters change abruptly with large amplitudes, classical adaptive control techniques give poor transient behavior and sometimes instability. Also, for obtaining the ideal closed-loop performance, linear quadratic regulator cannot be re-designed in real-time for changed parameters of the smart structures, even if these parameters are identified in real time. Closed-loop identification of system parameters and control gains using system classification-based neural networks is proposed and implemented. A preliminary experimental study is also done to see the effectiveness of the proposed technique over classical control methods.

A Fast Adaptive Memetic Algorithm for Online and Offline Control Design of PMSM Drives

A fast adaptive memetic algorithm (FAMA) is proposed which is used to design the optimal control system for a permanent-magnet synchronous motor. The FAMA is a memetic algorithm with a dynamic parameter setting and two local searchers adaptively launched, either one by one or simultaneously, according to the necessities of the evolution. The FAMA has been tested for both offline and online optimization. The former is based on a simulation of the whole system--control system and plant--using a model obtained through identification tests. The online optimization is model free because each fitness evaluation consists of an experimental test on the real motor drive. The proposed algorithm has been compared with other optimization approaches, and a matching analysis has been carried out offline and online. Excellent results are obtained in terms of optimality, convergence, and algorithmic efficiency. Moreover, the FAMA has given very robust results in the presence of noise in the experimental system.

Adaptive feed-forward control of thermal heating process

A novel adaptive inverse control (AIC) scheme is presented to facilitate on-line controller design of unknown nonlinear dynamic systems. The scheme is based on a control oriented model known as the U-model. The use of U-model alleviates the computational complexity of on-line nonlinear controller design that arises when using other modelling frame works such as NARMAX model. The effectiveness of the proposed scheme is illustrated by utilizing a laboratory scale nonlinear heating process as a test system.

Pole Placement Controller Design Based on Fuzzy Process Model

An explicit self-tuning controller based on the Takagi-Sugeno fuzzy model of the process is proposed. The fuzzy model is represented as a linear regression model whose parameters are functions of some of the process variables. Such a model can be considered as a linear time-varying model whose parameter values are known at every moment. The pole placement design procedure modified for time-varying systems is applied to obtain the polynomial controller parameters that provide the desired closed-loop poles. The proposed algorithm is very simple, and thus suitable for on-line controller design in adaptive control systems.

On-line system identification and control design of an extrusion cooking process: Part II. Model predictive and inferential control design

A non-linear model-based predictive control law to regulate specific mechanical energy (SME) using screw speed is developed. Operating variable setpoints are determined using an inferential model that correlates SME with product density and melt moisture content. Additionally, a ratio control strategy is used to regulate melt moisture content and an output noise filter is added to attenuate sensor noise. When combined with an on-line system identification procedure, the resulting system provides good servo and regulatory control response that is robust to modeling errors and disturbances. This paper represents the second of two papers that describe our investigation into the use of on-line system identification for automatic tuning and adaptive control of high-shear twin-screw extrusion processes. The resulting control design addresses the need for a comprehensive system that can be used to regulate extrudate density in extrusion cooking and puffing processes.

On-line system identification and control design of an extrusion cooking process: Part I. System identification

A systems analysis of an extrusion cooking process for puffed corn snack products revealed that the specific mechanical energy ( SME) and screw speed ( SS) was a desirable pairing of measured and manipulated variables, respectively for regulating extrudate density. To facilitate the design of an SME model-based control system, a discrete-time transfer function that depicts the dynamic response of motor load ( ML) to changes in SS is required. The research literature describes several off-line techniques for developing such transfer function models but no methods for on-line system identification were found. This paper represents the first of two articles that describe our investigations into the use of on-line system identification for automatic tuning and adaptive control of a high-shear twin-screw extrusion process. This paper reports results for using various system identification schemes in combination with relay-feedback as a way to derive, in real-time, a transfer function model that accurately depicts the dynamical behavior of an extrusion cooking process. A Wenger TX-52 co-rotating twin screw extruder was subjected to relay feedback during the processing of cornmeal for a breakfast cereal formula under different moisture and screw speed conditions. The data obtained from these experiments were used to derive first-, second- and third-order discrete-time transfer functions. An analysis of the resulting transfer functions revealed that a first-order lead-lag transfer function structure adequately described the dominant dynamic behavior of the process in all cases. Next, batch and recursive implementations of least-squares, extended least-squares, output error, maximum likelihood, Box–Jenkins and predictive error algorithms were used to derive parameters for the first-order transfer function. Overall, the batch output error method provided good transfer function estimates over the range of product and process conditions studied.

Manifold based feedback linearization of a 1-DOF beam with one-sided spring

Vibration reduction in a harmonically excited 1-DOF beam with one-sided spring is realized by control-ling the system state from a stable large amplitude 1/2 subharmonic response towards a coexisting unstable small amplitude harmonic response using feedback linearization. At the unstable harmonic response no control effort is needed because the unstable harmonic response is a long term solution of the uncontrolled system. To reduce control effort when stabilizing the unstable harmonic response, the stable manifold can be used within the control design because at the stable manifold the system state approaches the unstable harmonic response without control effort. Unfortunately, the calculation of this manifold acquires much off-line computational effort while its usage complicates the on-line control design. Therefore, the stable manifold is approximated by the stable eigenvectors of the monodromy matrix. Due to the local validity of the approximation, a two-stage control approach is used. In the first stage, the system state is controlled towards the unstable harmonic response to reach the region where the stable manifold can be approximated accurately by the stable eigenvectors. In the second stage, the system state is controlled towards the stable eigenvectors and approaches the unstable harmonic response with hardly any control effort

Predictive fuzzy control applied to the sinter strand process

A predictive fuzzy control scheme is developed where an intermediate process model, adaptively identified from observed data, is used to perform on-line controller design to meet a closed-loop performance specification that is associated with desired process output and predicted output. It is shown that the satisfactory response of the closed-loop system can be achieved by selection of the prediction horizon. The resulting separation of the adaptation system from the controller design enables learning convergency to be investigated. Simulation of the application of PFC to the control of a sinter strand process illustrates the ability of the method to control a complex industrial process.

Predictive Fuzzy Control for the Sinter Strand Process

A predictive fuzzy control scheme is developed where an intermediate process model, adaptively identified from observed data, is used to perform on-line controller design to meet a closed loop performance specification which is associated with desired process output and predicted output. It is shown that the satisfactory response of the closed-loop system can be achieved by selection of the prediction horizon. The resulting separation of the adaptation system from the controller design enables learning convergency to be investigated. Simulation of the application of PFC to the control of a sinter strand process illustrates the ability of the method to control a complex industrial process.