Moon-Ghu Park

Time-optimal control of nuclear reactor power with adaptive proportional-integral-feedforward gains

A time-optimal control method which consists of coarse and fine control stages is described. During the coarse control stage, the maximum control effort (time-optimal) is used to direct the system toward the switching boundary, which is set near the desired power level. At this boundary, the controller is switched to the fine control stage in which an adaptive proportional-integral-feedforward controller is used to compensate for any unmodeled reactivity feedback effects. This fine control is also introduced to obtain a constructive method for determining the (adaptive) feedback gains against the sampling effect. The feedforward control term is included to suppress the over- or undershoot. The estimation and feedback of the temperature-induced reactivity are also discussed. >

Design of a nonlinear model-based controller with adaptive PI gains for robust control of a nuclear reactor

Abstract A Model-Based Controller with Adaptive Proportional-Integral gains (MCAPI) is developed for a trajectory tracking control of a single-input, single-output nonlinear system. The robustness of tracking error dynamics is guaranteed by the feedback of estimated uncertainty and the performance specification given by the adaptation of PI gains using the second method of Lyapunov. The newly developed MCAPI method is applied to the power tracking control of a nuclear reactor and the simulation results show great improvement in tracking performance compared with the conventional model-based control methods.

INVESTIGATION OF REACTOR CONDITION MONITORING AND SINGULARITY DETECTION VIA WAVELET TRANSFORM AND DE-NOISING

Wavelet theory was applied to detect a singularity in a reactor power signal. Compared to Fourier transform, wavelet transform has localization properties in space and frequency. Therefore, using wavelet transform after de-noising, singular points can easily be found. To test this theory, reactor power signals were generated using the HANARO(a Korean multi-purpose research reactor) dynamics model consisting of 39 nonlinear differential equations contaminated with Gaussian noise. Wavelet transform decomposition and de-noising procedures were applied to these signals. It was possible to detect singular events such as a sudden reactivity change and abrupt intrinsic property changes. Thus, this method could be profitably utilized in a real-time system for automatic event recognition(e.g., reactor condition monitoring).

Reconstruction of core axial power shapes using the alternating conditional expectation algorithm

Abstract We have introduced the alternating conditional expectation (ACE) algorithm in reconstructing 20-node axial core power shapes from five-level in-core detector powers. The core design code, Reactor Operation and Control Simulation (ROCS), calculates 3-dimensional power distributions for various core states, and the reference core-averaged axial power shapes and corresponding simulated detector powers are utilized to synthesize the axial power shape. By using the ACE algorithm, the optimal relationship between a dependent variable, the plane power, and independent variables, five detector powers, is determined without any preprocessing. A total of ∼3490 data sets per each cycle of YongGwang Nuclear (YGN) power plant units 3 and 4 is used for the regression. Continuous analytic function corresponding to each optimal transformation is calculated by simple regression model. The reconstructed axial power shapes of ∼21,200 cases are compared to the original ROCS axial power shapes. Also, to test the validity and accuracy of the new method, its performance is compared with that of the Fourier fitting method (FFM), a typical method of the deterministic approach. For a total of 21,204 data cases, the averages of root mean square (rms) error, axial peak error ( ΔF z ), and axial shape index error ( Δ ASI) of new method are calculated as 0.81%, 0.51% and 0.00204, while those of FFM are 2.29%, 2.37% and 0.00264, respectively. We also evaluated the wide range of axial power profiles from the xenon-oscillation. The results show that the newly developed method is far superior to FFM; average rms and axial peak error are just ∼35 and ∼20% of those of FFM, respectively. ©

Self-tuning control of a nuclear reactor using a Gaussian function neural network

A self-tuning control method is described for a nuclear reactor system that requires only a set of input-output measurements. The use of an artificial neural network in nonlinear model-based adaptive control, both as a plant model and a controller, is investigated. A neural network called a Gaussian function network is used for one-step-ahead predictive control to track the desired plant output. The effectiveness of the controller is demonstrated by the application of the method to the power tracking control of the Korea Multipurpose Research Reactor.

GLRT based fault detection in sensor drift monitoring system

In a nuclear power plant (NPP), periodic sensor calibrations are required to assure sensors are operating correctly. However, only a few faulty sensors are found to be calibrated. For the safe operation of an NPP and the reduction of unnecessary calibration, on-line calibration monitoring is needed. This paper presents an on-line sensor drift monitoring technique, based on a Generalized Likelihood Ratio Test (GLRT), for detecting and estimating mean shifts in sensor signal. Also, principal component-based Auto-Associative support vector regression (AASVR) is proposed for the sensor signal validation of the NPP. Response surface methodology (RSM) is employed to efficiently determine the optimal values of SVR hyperparameters. The proposed model was confirmed with actual plant data of Kori NPP Unit 3. The results show that the accuracy of the model and the fault detection performance of the GLRT are very competitive.

Multi-resolution analysis for determining control rod drop time

This study describes a multi-resolution analysis (MRA) to determine the onset and end drop times of control rods. The measurement test of the drop times of control rods is normally performed during the start-up test of each reactor cycle since it is a crucial safety function to guarantee the reactor safe shutdown. The MRA with wavelet transform is particularly useful in analyzing the onset transients of rod drop as a means of capturing the unique attributes of such signals in an efficient way. This approach also allows the automated determination of rod drop time which reduces the uncertainty induced by ad hoc heuristic approaches. The test signal is generated by adding the random noise measured from real rod drop tests subtracting the wavelet-filtered noise free signal from the noisy signal leaving the noise. The signal is similar to both high sharp spikes noise and sine wave noise from the real voltage trace generated during the rod drop test. The effectiveness of the method is demonstrated by the MRA process.

Signal validation based on PCSVR and EULM

In a nuclear power plant (NPP), periodic sensor calibrations are required to assure sensors are operating correctly. However, only a few faulty sensors are found to be calibrated. For the safe operation of an NPP and the reduction of unnecessary calibration, on-line calibration monitoring is needed. In the previous study, principal component-based Auto-Associative support vector regression (PCSVR) was proposed for the sensor signal validation of the NPP. In this paper the error uncertainty limit monitoring (EULM) is integrated with PCSVR for the failure detection. This paper describes the design of an AASVR-based sensor validation system for a power generation system. Response surface methodology (RSM) is employed to efficiently determine the optimal values of SVR hyperparameters. The residuals between the estimated signals and the measured signals are inputted to the EULM to detect whether the sensors are failed or not. The proposed sensor monitoring algorithm was verified through applications to the turbine 1st chamber pressure in pressurized water reactor (PWR).

KERNEL-BASED NOISE FILTERING OF NEUTRON DETECTOR SIGNALS

This paper describes recently developed techniques for effective filtering of neutron detector signal noise. In this paper, three kinds of noise filters are proposed and their performance is demonstrated for the estimation of reactivity. The tested filters are based on the unilateral kernel filter, unilateral kernel filter with adaptive bandwidth and bilateral filter to show their effectiveness in edge preservation. Filtering performance is compared with conventional low-pass and wavelet filters. The bilateral filter shows a remarkable improvement compared with unilateral kernel and wavelet filters. The effectiveness and simplicity of the unilateral kernel filter with adaptive bandwidth is also demonstrated by applying it to the reactivity measurement performed during reactor start-up physics tests.