Man Gyun Na

Auto-tuned PID controller using a model predictive control method for the steam generator water level

In this paper, proportional-integral-derivative (PID) control gains are automatically tuned by using a model predictive control (MPC) method. The MPC has received much attention as a powerful tool for the control of industrial process systems. An MPC-based PID controller can be derived from the second-order linear model of a process. The steam generator is usually described by the well-known fourth-order linear model, which consists of the mass capacity, reverse dynamics, and mechanical oscillation terms. However the important terms in this linear model are the mass capacity and reverse dynamics terms, both of which can be described by a second-order linear system. The proposed auto-tuned PID controller was applied to a linear model of steam generators. The parameters of a linear model for steam generators are very different according to the power levels. The PID gains of the proposed controller are tuned automatically. Also, the proposed controller showed fast water level tracking and small shrink and swell performance by changing only the input-weighting factor according to the power level for the water-level deviation and sudden steam flow disturbances supposed to investigate the tracking performance and swell and shrink characteristics.

Design of a genetic fuzzy controller for the nuclear steam generator water level control

The nuclear steam generator is a nonminimum-phase system, which is caused by the swell and shrink effects. Since its inverse system has unstable dynamics, it is difficult to train the fuzzy controller via the conventional backpropagation of the system output errors. In this paper, a genetic algorithm is applied for the simultaneous design of membership functions and rule sets for a fuzzy control method for a steam generator water level. The genetic fuzzy controller for the steam generator is a fuzzy logic controller which is tuned offline by the genetic algorithm using the water level, feedwater flowrate, and steam flowrate signals of the steam generator. The symmetric Gaussian membership functions based on the flowrate and water level errors are applied. The proposed genetic fuzzy controller has a generalized and simplified rule base. The same genetic algorithm that is used to optimize the genetic fuzzy controller tunes a conventional proportional-integral (P-I) controller, and the performance of two controllers is compared. The genetic fuzzy controller shows good response that its swell and shrink phenomena are smaller and its response is faster than those of a well-tuned P-I controller are.

Design of a fuzzy model predictive power controller for pressurized water reactors

In this paper, a fuzzy model predictive control method is applied to design an automatic controller for thermal power control in pressurized water reactors. The future reactor power is predicted by using the fuzzy model identified by a subtractive clustering method of a fast and robust algorithm. The objectives of the proposed fuzzy model predictive controller are to minimize both the difference between the predicted reactor power and the desired one, and the variation of the control rod positions. Also, the objectives are subject to maximum and minimum control rod positions and maximum control rod speed. The genetic algorithm that is useful to accomplish multiple objectives is used to optimize the fuzzy model predictive controller. A three-dimensional nuclear reactor analysis code is used to verify the proposed controller for a nuclear reactor. From results of numerical simulation to check the performance of the proposed controller at the 5%/min ramp increase or decrease of a desired load and its 10% step increase or decrease which are design requirements, it was found that the nuclear power level controlled by the proposed fuzzy model predictive controller could track the desired power level very well.

Prediction of major transient scenarios for severe accidents of nuclear power plants

It is very difficult for nuclear power plant operators to predict and identify the major severe accident scenarios following an initiating event by staring at temporal trends of important parameters. In this regard, a probabilistic neural network (PNN) that has been applied well to the classification problems is used in order to classify accidents into groups of initiating events such as loss of coolant accidents (LOCA), total loss of feedwater (TLOFW), station blackout (SBO), and steam generator tube rupture (SGTR). Also, a fuzzy neural network (FNN) is designed to identify their major severe accident scenarios after the initiating events. The inputs to PNN and FNN are initial time-integrated values obtained by integrating measurement signals during a short time interval after reactor scram. An automatic structure constructor for the fuzzy neural network automatically selects the input variables from the time-integrated values of many measured signals, and optimizes the number of rules and its related parameters. In cases that an initiating event develops into a severe accident, this may happen when plant operators do not follow the appropriate accident management guidance or plant safety systems do not work, the proposed algorithm showed accurate classification of initiating events. Also, it well predicted timings for important occurrences during severe accident progression scenarios, which is very helpful to perform severe accident management.

Design of an adaptive predictive controller for steam generators

The water level control of a nuclear steam generator is very important to secure the sufficient cooling inventory for the nuclear reactor and, at the same time, to prevent the damage of turbine blades. The dynamics of steam generators is very different according to power levels and changes as time goes on. The generalized predictive control method is used to solve an optimization problem for the finite future time steps at current time and to implement only the first control input among the solved optimal control inputs of several time steps. A recursive parameter estimation algorithm estimates on-line the mathematical model of steam generator every time step to generate the linear controller design model. In this work, by combining these generalized predictive control method and recursive parameter estimation algorithm, a new controller is designed to control the water level of nuclear steam generators. It is shown through application to a linear model and a nonlinear model of steam generators that the proposed controller has good performance.

Design of an adaptive observer-based controller for the water level of steam generators

Abstract The water level contributions from the mass capacity, reverse dynamics, and mechanical oscillations are estimated by applying an adaptive observer. In an adaptive observer, both parameters and state variables of the system are estimated simultaneously. The cost function used in control design compensates the reverse dynamics so that the controller may be insensitive to the reverse dynamics. The time-varying problem is resolved by estimating the parameters at every time step. By estimating the flow errors along with the states and the parameters, a control algorithm is derived to treat the time-varying property, reverse dynamics, and flow measurement errors. The proposed algorithm is compared with the conventional P-I controller. It is found that the adaptive observer-based controller generates faster responses and smaller swell, shrink, and overshoot than the P-I controller does.

A model predictive controller for load-following operation of PWR reactors

The basic concept of a model predictive control method is to solve on-line, at each time step, an optimization problem for a finite future interval and to implement only the first optimal control input as the current control input. It is a suitable control strategy for time-varying systems, in particular, because the parameter estimator identifies a controller design model recursively at each time step, and also the model predictive controller recalculates an optimal control input at each time step by using newly measured signals. The proposed controller is applied to the integrated power level and axial power distribution controls for a Korea Standard Nuclear Power Plant (KSNP). The power level and the axial shape index are controlled by two kinds of the five regulating control rod banks and the two part-strength control rod banks together with the automatic adjustment of boric acid concentration. The three-dimensional reactor analysis code, Multipurpose Analyzer for Static and Transient Effects of Reactor, which models the KSNP, is interfaced to the proposed controller to verify the proposed controller for controlling the reactor power level and the axial shape index. It is known from numerical simulations that the proposed controller exhibits very fast tracking responses.

Detection and Diagnostics of Loss of Coolant Accidents Using Support Vector Machines

It is very difficult for operators to predict the progression of a loss of coolant accident (LOCA) because nuclear plant operators are provided with only partial information during the accident or they may have insufficient time to analyze the data despite being provided with considerable information. Therefore, its break location should be identified and the break size should be predicted accurately in order to provide the operators and technical support personnel with important and valuable information needed to successfully manage the accident. In this paper, support vector machines (SVMs) are used to identify the break location of a LOCA and predict the break size using the support vector classification (SVC) and support vector regression (SVR), which are well-known application areas of SVMs. The SVR models to predict the break size were optimized using a genetic algorithm. The inputs to the SVMs are the time-integrated values obtained by integrating the measurement signals in a short time interval after a reactor scram. The results showed that the proposed algorithm identified the break locations of LOCAs without fault and predicted the break size accurately.

PREDICTION OF RESIDUAL STRESS FOR DISSIMILAR METALS WELDING AT NUCLEAR POWER PLANTS USING FUZZY NEURAL NETWORK MODELS

A fuzzy neural network model is presented to predict residual stress for dissimilar metal welding under various welding conditions. The fuzzy neural network model, which consists of a fuzzy inference system and a neuronal training system, is optimized by a hybrid learning method that combines a genetic algorithm to optimize the membership function parameters and a least squares method to solve the consequent parameters. The data of finite element analysis are divided into four data groups, which are split according to two end-section constraints and two prediction paths. Four fuzzy neural network models were therefore applied to the numerical data obtained from the finite element analysis for the two end-section constraints and the two prediction paths. The fuzzy neural network models were trained with the aid of a data set prepared for training (training data), optimized by means of an optimization data set and verified by means of a test data set that was different (independent) from the training data and the optimization data. The accuracy of fuzzy neural network models is known to be sufficiently accurate for use in an integrity evaluation by predicting the residual stress of dissimilar metal welding zones.

PREDICTION OF THE REACTOR VESSEL WATER LEVEL USING FUZZY NEURAL NETWORKS IN SEVERE ACCIDENT CIRCUMSTANCES OF NPPS

Safety-related parameters are very important for confirming the status of a nuclear power plant. In particular, the reactor vessel water level has a direct impact on the safety fortress by confirming reactor core cooling. In this study, the reactor vessel water level under the condition of a severe accident, where the water level could not be measured, was predicted using a fuzzy neural network (FNN). The prediction model was developed using training data, and validated using independent test data. The data was generated from simulations of the optimized power reactor 1000 (OPR1000) using MAAP4 code. The informative data for training the FNN model was selected using the subtractive clustering method. The prediction performance of the reactor vessel water level was quite satisfactory, but a few large errors were occasionally observed. To check the effect of instrument errors, the prediction model was verified using data containing artificially added errors. The developed FNN model was sufficiently accurate to be used to predict the reactor vessel water level in severe accident situations where the integrity of the reactor vessel water level sensor is compromised. Furthermore, if the developed FNN model can be optimized using a variety of data, it should be possible to predict the reactor vessel water level precisely.