Loading...

(Development of Efficient Methods for Design of an Operator Aided Tool for Identification and Forecasting of Transients in PWRs (Case Study: BNPP

Moshkbar-Bakhshayesh, Khalil | 2014

1075 Viewed
  1. Type of Document: Ph.D. Dissertation
  2. Language: Farsi
  3. Document No: 46371 (46)
  4. University: Sharif University of Technology
  5. Department: Energy Engineering
  6. Advisor(s): Ghofrani, Mohammad Bagher
  7. Abstract:
  8. This thesis introduces a new method for identification and forecasting of future states of nuclear power plants (NPPs) parameters. The proposed method consists of four steps. First, the type of transients is recognized by the modular identifier which has been developed using the latest advances of error back propagation (EBP) learning algorithm. In second step, for more robustness of modular identifier against noisy input data, auto-regressive integrated moving average (ARIMA) method is used. A hybrid network is then used to forecast the selected parameters of the identified transient. ARIMA model is used to estimate the linear component of the selected parameters. The neural network developed by EBP learning algorithm is then used to estimate the nonlinear component of the selected parameters. Prediction of parameters is obtained by adding the estimated linear and nonlinear components. Finally, for transients out of collocated knowledge (unknown transients), a semi-supervised approach namely transductive support vector machine (TSVM) is used to cluster the type of transients. To analyze the ability of the proposed method, Bushehr nuclear power plant (BNPP) parameters are forecasted. Results show good agreement of forecasted parameters with final safety analysis report (FSAR). Noticeable advantages of the proposed method are: More robustness against noisy data, improvement balance between memorization and generalization, reduction of false identification, sole dependency of identification on the sign of each output signal, selection of the plant variables for transients training independent of each other, extendibility for identification of more transients without unfavorable effects, forecasting any quantifiable parameter without necessity to know its correlation with other, possibility for prediction of parameters in long temporal dependencies, and clustering of unlabeled transients.Conceptual design of an operator aided tool (OAT) is developed based on the proposed method.This OAT can be used as a support system for the NPPs operators to estimate uture states of the plant and to perform appropriate actions before transient progresses into critical states
  9. Keywords:
  10. Artificial Neural Network ; Autoregressive Integrated Moving Average (ARIMA) ; Support Vector Machine (SVM) ; Bushehr Nuclear Power Plant ; Transient Identification ; Parameters Forecasting ; Transductive Support Vector Machine (TSVM)Method ; Operator Aided Tool

 Digital Object List

 Bookmark