Real-time Predictive Control Research Articles

Statistical evaluation of a radar rainfall system for sewer system management

Urban areas are faced with mounting demands for managing waste and stormwater for a cleaner environment. Rainfall information is a critical component in efficient management of urban drainage systems. A major water quality impact affecting receiving waterbodies is the discharge of untreated waste and stormwater during precipitation, termed wet weather flow. Elimination or reduction of wet weather flow in metropolitan sewer districts is a major goal of environmental protection agencies and often requires considerable capital improvements. Design of these improvements requires accurate rainfall data in conjunction with monitored wastewater flow data. Characterizing the hydrologic/hydraulic performance of the sewer using distant rain gauges can cause oversizing and wasted expenditures. Advanced technology has improved our ability to measure accurately rainfall over large areas. Weather radar, when combined with rain gauge measurements, provides detailed information concerning rainfall intensities over specific watersheds. Knowing how much rain fell over contributing areas during specific periods aids in characterizing inflow and infiltration to sanitary and combined sewers, calibration of sewer system models, and in operation of predictive real-time control measures. Described herein is the design of a system for managing rainfall information for sewer system management, along with statistical analysis of 60 events from a large metropolitan sewer district. Analysis of the lower quartile rainfall events indicates that the expected average difference is 25.61%. Upper quartile rainfall events have an expected average difference of 17.25%. Rain gauge and radar accumulations are compared and evaluated in relation to specific needs of an urban application. Overall, the events analyzed agree to within ± 8% based on the median average difference between gauge and radar.

Global Predictive Real-Time Control of Sewers Allowing Surcharged Flows

A global predictive real-time control strategy minimizing overflow volumes from combined sewers during rainfalls is presented. For an optimal use of controlled sewer transport and storage capacities, the proposed strategy allows surcharged flows. Flows and piezometric heads in the sewer are computed according to flow inputs by a hydraulic simulation model. The optimal operation of the regulators controlling these flow inputs is determined on a finite control horizon using the generalized reduced gradient optimization algorithm. The control strategy was applied to the 23 rain events that occurred during the summer of 1989 on the urban area drained by the Marigot interceptor in Laval, Canada. In this application, the admitted intensity of surcharges was varied to assess this parameter impact on total overflow volumes. A comparison between performances of the proposed strategy and a local reactive control was also carried out. Results obtained indicate that the global predictive control can reduce overflow v...

Model Reference Adaptive Predictive Control for a Variable-Frequency Oil-Cooling Machine

This paper develops methodologies and techniques for the design, analysis, and implementation of a model reference adaptive predictive temperature controller for a variable-frequency oil-cooling machine, suited for cooling high-speed machine tools. The oil-cooling process is modeled experimentally as a first-order system model with a time delay and its system parameters are identified using the recursive least-square method. Based on this model, a model reference adaptive predictive controller is proposed for achieving set-point tracking and robustness. A real-time model reference adaptive predictive control algorithm is then presented and implemented utilizing a stand-alone digital signal processor TMS320F243 from Texas Instruments Incorporated. The experimental results show that the proposed control method is proven capable of giving satisfactory performance under set-point changes, fixed loads, and load changes.

CHOICE OF RBF MODEL STRUCTURE FOR PREDICTING GREENHOUSE INSIDE AIR TEMPERATURE

The application of the radial basis function neural network to greenhouse inside air temperature modelling has been previously investigated by the authors. In those studies, the inside air temperature is modelled as a function of the inside relative humidity and of the outside temperature and solar radiation. Several training and learning methods were compared and the application of the Levenberg-Marquardt optimisation method was found to be the best way to determine the neural network parameters. A second-order model structure previously selected in the context of dynamic temperature models identification, was used. The model is intended to be incorporated in a real-time predictive greenhouse environmental control strategy. It is now relevant to question if the model structure used so far, selected in a different modelling framework, is the most correct in some sense. In this paper the usefulness of correlation-based model validity tests is addressed in order to answer the question mentioned above.

Real-time predictive control of photoresist film thickness uniformity

With the trends toward larger wafer size and the linewidth going below 100 nm, one of the challenges is to control the resist thickness and uniformity to a tight tolerance in order to minimize the thin-film interference effect on the critical dimension. In this paper, we propose a new approach to improve resist thickness control and uniformity through the softbake process. Using an array of thickness sensors, a multizone bakeplate, and advanced control strategy, the temperature distribution of the bakeplate is manipulated in real time to reduce resist thickness nonuniformity. The bake temperature is also constrained to prevent the decomposition of a photoactive compound in the resist. We have experimentally obtained a repeatable improvement in resist thickness uniformity from wafer-to-wafer and across individual wafers. Thickness nonuniformity of less than 10 /spl Aring/ has been obtained. On average, there is 10 /spl times/ improvement in the thickness uniformity as compared to conventional softbake process.

Predicting the Greenhouse Inside Air Temperature with RBF Neural Networks

The application of the radial basis function neural network to greenhouse inside air temperature modelling has been previously investigated by the authors. In those studies, the inside air temperature is modelled as a function of the inside relative humidity and of the outside temperature and solar radiation. A second-order model structure previously selected in the context of dynamic temperature models identification, is used. Several training and learning methods were compared and the application of the Levenberg-Marquardtoptimisation method was found to be the best way to determine the neural network parameters. Such a type of model is intended to be incorporated in a real-time predictive greenhouse environmental control strategy, which implies that prediction horizons greater than one time step will be necessary. In this paper the radial basis function neural network will be compared to conventional auto-regressive with exogenous inputs models, on the prediction of the greenhouse inside air temperature, considering prediction horizons greater than one time step.

Global predictive real-time control of Quebec Urban Community's westerly sewer network

Quebec Urban Community (QUC) has selected Global Predictive Real-Time Control (GP-RTC) as the most efficient approach to achieve environmental objectives defined by the Ministry of Environment. QUC wants to reduce combined sewer overflows (CSOs) frequency to the St Lawrence river to two events per summer period in order to reclaim the use of Jacques-Cartier Beach for recreational activities and sports of primary contact. QUC's control scheme is based on the Certainty Equivalent Control Open Loop Feedback (CEOLF) strategy which permits one to introduce, at each control period, updated measurements and meteorological predictions. A non-linear programming package is used to find the flow set points that minimise a multi-objective (cost) function, subjected to linear equality and inequality constraints representing the physical and operational constraints on the sewer network. Implementation of GP-RTC on QUC's westerly network was performed in the summer of 1999 and was operational by mid-August. Reductions in overflow volumes with GP-RTC compared to static control are attributed to the optimal use of two existing tunnels as retention facilities as well as the maximal use of the wastewater treatment plant (WWTP) capacity.

Convergence of constrained model-based predictive control for batch processes

The convergence property of constrained model-based predictive control for batch processes (BMPC) is investigated. BMPC is a recently developed control technique that combines iterative learning control with real-time predictive control. It is proven for a general class of linear constrained systems that the tracking error converges to zero as the run number increases.

Modelling large urban drainage systems in real time

The interactions between rainfall and urban drainage systems (UDSs) are complex and must be considered as a whole in order to maximise control efficiency whilst at the same time achieving environmentally acceptable solutions. More rigorous standards, as a result of recent EU and UK legislation, are increasingly encouraging intervention in system management rather than more traditional passive procedures. To achieve these goals a global predictive real-time control (RTC) strategy is required, in which real-time flow prediction plays an important part in the provision of necessary first-hand information on system status in both current and predictive modes. This paper describes one such strategy, which differs from existing methods in the following ways: the novel way in which the UDS is represented; the algorithm used for model parameter identification; the strategies associated with the system output prediction; and the transfer function model used to represent the system. This transfer function model is a conceptually parameterised transfer function (CPTF) model, which by its nature falls into the category of lumped, dynamic, linear and conceptual although its structure takes the form of a non-conceptual transfer function model. The modelling approach is described as the RHINOS (Real-time urban Hydrological INfrastructure and Output modelling Strategy).

Real-time urban drainage system modelling using weather radar rainfall data

This paper introduces an urban drainage system (UDS) modelling ;approach called RHINOS (Real-time urban Hydrological Infrastructure and Output modelling Strategy) developed specifically for use in predictive realtime control (RTC)' applications. RHINOS uses quantitative distributed rainfall forecasts to predict pipe flows at locations throughout the pipe network. The paper highlights, with the use of a case study, the contribution weather radar systems can make to the RTC of UDS. The RHINOS procedure differs from existing urban hydrological modelling approaches in several ways including: (i) the way in which the UDS is represented; (ii) the parameter identification/estimation algorithm, and (iii) the techniques developed for flow prediction. RHINOS has been developed to produce supplementary information to any sewer telemetry system and enable global predictive control strategies to be implemented.

Modelling large urban drainage systems in real time

The interactions between rainfall and urban drainage systems (UDSs) are complex and must be considered as a whole in order to maximise control efficiency whilst at the same time achieving environmentally acceptable solutions. More rigorous standards, as a result of recent EU and UK legislation, are increasingly encouraging intervention in system management rather than more traditional passive procedures. To achieve these goals a global predictive real-time control (RTC) strategy is required, in which real-time flow prediction plays an important part in the provision of necessary first-hand information on system status in both current and predictive modes. This paper describes one such strategy, which differs from existing methods in the following ways: the novel way in which the UDS is represented; the algorithm used for model parameter identification; the strategies associated with the system output prediction; and the transfer function model used to represent the system. This transfer function model is a conceptually parameterised transfer function (CPTF) model, which by its nature falls into the category of lumped, dynamic, linear and conceptual although its structure takes the form of a non-conceptual transfer function model. The modelling approach is described as the RHINOS (Real-time urban Hydrological INfrastructure and Output modelling Strategy).

Neural network identification and control of an underwater vehicle

Real-time predictive control requires a forward model that is both accurate and fast. This paper introduces two nonlinear internal memory network architectures and compares their performance with a Multi-layer Perceptron (MLP) augmented with the use of spread encoding. The test plant is a single component from an Underwater Robotic Vehicle (URV), comprising a thruster encased in a steel frame and provided with buoyancy. This assemblv is free to move under water and is controlled for depth. The internal memory networks are of comparable accuracy to the MLP but more parsimonious, resulting, in a faster response which makes them better suited for on-line control. Although a particular case study is presented as the focus of this paper, the algorithms and methods developed have generic applicability.

Macroscopic Modeling of Freeway Traffic Using an Artificial Neural Network

Traffic flow on freeways is a complex process that often is described by a set of highly nonlinear, dynamic equations in the form of a macroscopic traffic flow model. However, some of the existing macroscopic models have been found to exhibit instabilities in their behavior and often do not track real traffic data correctly. On the other hand, microscopic traffic flow models can yield more detailed and accurate representations of traffic flow but are computationally intensive and typically not suitable for real-time implementation. Nevertheless, such implementations are likely to be necessary for development and application of advanced traffic control concepts in intelligent vehicle-highway systems. The development of a multilayer feed-forward artificial neural network model to address the freeway traffic system identification problem is presented. The solution of this problem is viewed as an essential element of an effort to build an improved freeway traffic flow model for the purpose of developing real-time predictive control strategies for dynamic traffic systems. To study the initial feasibility of the proposed neural network approach for traffic system identification, a three-layer feed-forward neural network model has been developed to emulate an improved version of a well-known higher-order continuum traffic model. Simulation results show that the neural network model can capture the traffic dynamics of this model quite closely. Future research will attempt to attain similar levels of performance using real traffic data.

Development of a knowledge-based approach for real-time predictive process control : T.H. Lee, C.C. Hang, S. Nungam, pp 847–850

Development of a knowledge-based approach for real-time predictive process control : T.H. Lee, C.C. Hang, S. Nungam, pp 847–850

Analysis and implementation of a real-time predictive current controller for permanent-magnet synchronous servo drives

A real-time current controller for PWM inverter-fed permanent-magnet synchronous motor drives is presented and analyzed. The proposed current control scheme is based on predictive control with a parallel integral loop added to compensate for the inaccuracy of the motor model and for the variations of motor parameters and DC voltage source. The proposed current control scheme is analyzed and its performance is evaluated by computer simulation. An EPROM-based implementation is presented in which calculations and pulsewidth modulation are executed by lookup tables resulting in high-speed operation. The controller performance is evaluated using a prototype l kW PM synchronous servo drive. Experimental results are given and discussed. >

Development of a Knowledge-Based Approach for Real-Time Predictive Process Control

In this paper, we present a knowledge-based approach to real-time predictive control which is applicable to a large class of industrial process control problems. The class of processes considered are those where dead-time IS significant and the .approach adopted is an integration of hard algorithms With knowledge-based methods. The paper also Includes the results of real-time experiments for level control In a coupled-tank system.

A knowledge-based approach to real-time predictive process control

This paper presents a knowledge-based approach to real-time predictive control which is applicable to a large class of industrial process control problems. The class of processes considered are those where dead-time is significant (common in the paper and pulp, mining and chemical process industries), and the approach adopted is an integration of hard algorithms with knowledge-based methods. In the paper, the appropriate hardware and software considerations to achieve the real-time knowledge-based framework are discussed, and the development of the knowledge-base for the predictive controller is presented. The paper also includes the results of real-time experiments for level control in a coupled-tank system.

A real-time predictive controller for automatic voltage regulation of a synchronous generator

The use of a predictive controller in automatic voltage regulation (AVR) of a synchronous generator is proposed for real-time control purposes. The controller consists of a fast-time linearized 2nd-order model of an alternator and control logic. The fast-time model is used to predict the time response of the controlled system (generator voltage output), and the control logic is responsible for choosing the necessary input to the controlled system. The principles of time-optimal control and linear regulator are used in controller design. The effectiveness of such a controller is examined by digital simulation study for the response of a synchronous machine, connected to an infinite bus under various input and initial conditions. The effects on the stability of the overall system are analysed.

A predictive time-optimal controller for second-order systems with time delay

This paper presents a method of interactive or real-time predictive control which can be implemented without a stored program computer. By restricting the process dy namic model to a second-order-plus-dead-time format, a configuration which is satisfactory for many chemical or petroleum processing systems, a time-optimal controller can be built entirely on a small analog/hybrid computer. For set-point changes, the controller utilizes an analog model of the process to search out the predicted optimum input switching sequence and continuously maintains the process on the time-optimal trajectory. Because of the predictive nature of the method there is no necessity to include the time-delay term in the high-speed model; the predicted switches in the process input are simply ad vanced in time by the amount of the time-delay.