Research on intelligent fault diagnosis method for nuclear power plant based on correlation analysis and deep belief network

Distinctive physical insights driven from machine learning modelling of nuclear power plant severe accident scenario propagation

The severe accident scenario propagation studies of nuclear power plants (NPPs) have been one of the most critical factors in deploying nuclear power for decades. During an NPP accident, the accident scenario can change during its propagation from the initiating event to a series of accident sub-scenarios. Hence, having time-wise updated information about the current type of accident sub-scenario can help plant operators mitigate the accident propagation and underlying consequences. In this work, we demonstrate the capability of machine learning (Decision Tree) to help researchers and design engineers in finding distinctive physical insights between four different types of accident scenarios based on the pressure vessel's maximum external surface temperature at a particular time. Although the four accidents we included in this study are considered some of the most extensively studied NPPs accident scenarios for decades, our findings shows that decision tree classification could define remarkable distinct differences between them with reliable statistical confidence.

Analysis of College Students' Network Moral Behavior by the History of Ideological and Political Education under Deep Learning

The research on the history of ideological and political education (IPE) is the basis for deepening it, and it is also of great help to higher education. The diversity of network information also easily leads to poor guidance for college students who are not strong in discrimination. This study adopts the method of a questionnaire survey to investigate the common moral anomie among college students in the network space. The survey data are sorted and classified and then input into the recurrent neural network structure for data analysis using deep learning (DL) algorithms. The results are fed back to the investigators intuitively and understandably. The results show that some college students have some problems, such as lack of network moral knowledge, vague values, moral behavior anomia, spatial knowledge and behavior inconsistency, and moral and emotional indifference. DL algorithms are added to the analysis process to make the findings more objective. These conclusions provide reference suggestions for subsequent research on college students' online moral behavior in the context of IPE history.

Monitoring of Temperature Measurements for Different Flow Regimes in Water and Galinstan with Long Short-Term Memory Networks and Transfer Learning of Sensors

Temperature sensing is one of the most common measurements of a nuclear reactor monitoring system. The coolant fluid flow in a reactor core depends on the reactor power state. We investigated the monitoring and estimation of the thermocouple time series using machine learning for a range of flow regimes. Measurement data were obtained, in two separate experiments, in a flow loop filled with water and with liquid metal Galinstan. We developed long short-term memory (LSTM) recurrent neural networks (RNNs) for sensor predictions by training on the sensor’s own prior history, and transfer learning LSTM (TL-LSTM) by training on a correlated sensor’s prior history. Sensor cross-correlations were identified by calculating the Pearson correlation coefficient of the time series. The accuracy of LSTM and TL-LSTM predictions of temperature was studied as a function of Reynolds number (Re). The root-mean-square error (RMSE) for the test segment of time series of each sensor was shown to linearly increase with Re for both water and Galinstan fluids. Using linear correlations, we estimated the range of values of Re for which RMSE is smaller than the thermocouple measurement uncertainty. For both water and Galinstan fluids, we showed that both LSTM and TL-LSTM provide reliable estimations of temperature for typical flow regimes in a nuclear reactor. The LSTM runtime was shown to be substantially smaller than the data acquisition rate, which allows for performing estimation and validation of sensor measurements in real time.

A Correlation-Based Feature Selection Algorithm for Operating Data of Nuclear Power Plants

Nuclear power plant operating data are characterized by a large variety, strong coupling, and low data value density. When using machine learning techniques for fault diagnosis and other related research, feature selection enables dimensionality reduction while maintaining the physical meaning of the original features, thus improving the computational efficiency and generalization ability of the learning model. In this paper, a correlation-based feature selection algorithm is developed to implement feature selection of nuclear power plant operating data. The proposed algorithm is verified by experiments and compared with traditional correlation-based feature selection algorithms. The experiments and comparison results show that the proposed algorithm is effective in realizing the dimensionality reduction of nuclear power plant operating data.

Prediction of the Loss of Feed Water Fault Signatures Using Machine Learning Techniques

Fault diagnosis occurrence and its precise prediction in nuclear power plants are extremely important in avoiding disastrous consequences. The inherent limitations of the current fault diagnosis methods make machine learning techniques and their hybrid methodologies possible solutions to remedy this challenge. This study sought to develop, examine, compare, and contrast three robust machine learning methodologies of adaptive neurofuzzy inference system, long short-term memory, and radial basis function network by modeling the loss of feed water event using RELAP5. The performance indices of residual plots, mean absolute percentage error, root mean squared error, and coefficient of determination were used to determine the most suitable algorithms for accurately diagnosing the loss of feed water transient signatures. The study found out that the adaptive neurofuzzy inference system model outperformed the other schemes when predicting the temperature of the steam generator tubes, the radial basis function network scheme was best suited in forecasting the mass flow rate at the core inlet, while the long short-term memory algorithm was best suited for the estimation of the severities of the loss of the feed water fault.

Comprehensive analysis and recommendation of feature evaluation measures for intrusion detection

The revolutionary advances in network technologies have spearheaded the design of advanced cyberattacks to surpass traditional security defense with dreadful consequences. Recently, Intrusion Detection System (IDS) is considered as a pivotal element in network security infrastructures to achieve solid line of protection against cyberattacks. The prime challenges presented to IDS are curse of high dimensionality and class imbalance that tends to increase the detection time and degrade the efficiency of IDS. As a result, feature selection plays an important role in enabling to identify the most significant features for intrusion detection. Although, several feature evaluation measures are being proposed for feature selection in literature, there is no consensus on which measures are best for intrusion detection. Therein, this work aims at recommending the most appropriate feature evaluation measure for building an efficient IDS. In this direction, four filter-based feature evaluation measures that stem from different theories such as Consistency, Correlation, Information and Distance are investigated for their potential implications in enhancing the detection ability of IDS model for different classes of attacks. Along with this, the influence of the selected features on classification accuracy of an IDS model is analyzed using four different categories of classifiers namely, K-nearest neighbors (KNN), Random Forest (RF), Support Vector Machine (SVM) and Deep Belief Network (DBN). Finally, a two-step statistical significance test is conducted on the experimental results to determine which feature evaluation measure contributes statistically significant difference in IDS performance. All the experimental comparisons are performed on two benchmark intrusion detection datasets, NSL-KDD and UNSW-NB15. In these experiments, consistency measure has best influenced the IDS model in improving the detection ability with regard to detection rate (DR), false alarm rate (FAR), kappa statistics (KS) and identifying the most significant features for intrusion detection. Also, from the analysis results, it is revealed that RF is the ideal classifier to be used in conjunction with any of these four feature evaluation measures to achieve better detection accuracy than others. From the statistical results, we recommend the use of consistency measure for designing an efficient IDS in terms of DR and FAR.