• 대한전기학회:학술대회논문집
• /
• 대한전기학회 2006년도 심포지엄 논문집 정보 및 제어부문
• /
• pp.135-137
• /
• 2006

To improve device yield and throughput, faults in plasma processing equipment should be quickly and accurately diagnosed. Despite many useful information of ex-situ sensor measurements, their applications to recognize plasma faultshave not been investigated. In this study, a new technique to identify fault causes by recognizing X-ray photoelectron spectroscopy (XPS) using neural network and continuous wavelet transformation (CWT). The presented technique was evaluated with the plasma etch data. A totalof 17 experiments were conducted for model construction. Model performance was investigated from the perspectives of training error, testing error, and recognition accuracy with respect to various thresholds. CWT-based BPNN models demonstrated a higher prediction accuracy of about 26%. Their advantages over pure XPS-based models were conspicuous in all three measures at small networks.

• 한국기계가공학회지
• /
• 제4권3호
• /
• pp.57-62
• /
• 2005

Neural networks, which have learning and self-organizing abilities, can be advantageously used in the pattern recognition. Neural network techniques have been widely used in monitoring and diagnosis, and compare favourable with traditional statistical pattern recognition algorithms, heuristic rule-based approaches, and fuzzy logic approaches. In this study the fault diagnosis of the face-milling using the artificial neural network was investigated. After training, the sample which measure load current was monitored by constant output results.

• 한국압력기기공학회 논문집
• /
• 제13권1호
• /
• pp.48-53
• /
• 2017

We diagnosed the defect using the data obtained from the nuclear power plant simulator. In this paper, we diagnosed faults in the nuclear power plant system for discovery instead of the traditional single-component or device unit. We created the six fault scenarios and used a fault simulator to obtain the fault data. It was extracted pattern from acquired failure data. Neural network model was trained and simple pattern matching algorithm was applied. We presented a simulation result and confirmed that the applied algorithm works correctly.

• 대한전기학회논문지:전력기술부문A
• /
• 제55권10호
• /
• pp.410-417
• /
• 2006

Generally, the training set of maximum $m{\times}L(m+f)$ patterns in the pattern recognition method is required for the real-time bus reconfiguration strategy when a main transformer fault occurs in the distribution substation. Accordingly, to make the application of pattern recognition method possible, the size of the training set must be reduced as efficient level. This Paper proposes a methodology which obtains the minimized training set by applying the pattern clustering method to load patterns of the main transformers and feeders during selected period and to obtain bus reconfiguration strategy based on it. The MaxMin distance clustering algorithm is adopted as the pattern clustering method. The proposed method reduces greatly the number of load patterns to be trained and obtain the satisfactory pattern matching success rate because that it generates the typical pattern clusters by appling the pattern clustering method to load patterns of the main transformers and feeders during selected period. The proposed strategy is designed and implemented in Visual C++ MFC. Finally, availability and accuracy of the proposed methodology and the design is verified from diversity simulation reviews for typical distribution substation.

• 한국정밀공학회지
• /
• 제19권3호
• /
• pp.52-58
• /
• 2002

The purpose of condition monitoring and fault diagnosis is to detect faults occurring in machinery in order to improve the level of safety in plants and reduce operational and maintenance costs. The recognition performance is important not only to gain a high recognition rate bur a1so to minimize the diagnosis failures error rate by using off effective rejection module. We examined the problem of performance evaluation for the rejection scheme considering the accuracy of individual c1asses in order to increase the recognition performance. We use the Smith's method among the previous studies related to rejection method. Nearest neighbor classifier is used for classifying the machine conditions from the vibration signals. The experiment results for the performance evaluation of rejection show the modified optimum rejection method is superior to others.

• Journal of Electrical Engineering and Technology
• /
• 제10권6호
• /
• pp.2326-2333
• /
• 2015

Stator turn faults in permanent magnet synchronous motors (PMSMs) are more dangerous than those in induction motors (IMs) because of the presence of spinning rotor magnets that can be turned off at will. Condition monitoring and fault detection and diagnosis of the PMSM have been receiving a growing amount of attention among scientists and engineers in the past few years. The aim of this study is to propose a new detection technique of stator winding faults in a three-phase PMSM. This technique is based on the image analysis and recognition of the stator current Concordia patterns, and will allow the identification of turn faults in the stator winding as well as its correspondent fault index severity. A test bench of a vector controlled PMSM motor behaviors under short circuited turn in two phases stator windings has been built. Some experimental results of the phase to phase short circuits have been performed for diagnosis purpose.

• 대한전기학회논문지:전력기술부문A
• /
• 제53권12호
• /
• pp.637-644
• /
• 2004

This paper proposes an expert system with the knowledge learning capability which can enhance the safety and effectiveness of substation operation in the automated substation as well as existing substation by inferring multiple events such as main transformer fault, busbar fault and main transformer work schedule under multiple inference mode and multiple objective mode and by considering totally the switch status and the main transformer operating constraints. Especially inference mode includes the local minimum tree search method and pattern recognition method to enhance the performance of real-time bus reconfiguration strategy. The inference engine of the expert system consists of intuitive inferencing part and logical inferencing part. The intuitive inferencing part offers the control strategy corresponding to the event which is most similar to the real event by searching based on a minimum distance classification method of pattern recognition methods. On the other hand, logical inferencing part makes real-time control strategy using real-time mode(best-first search method) when the intuitive inferencing is failed. Also, it builds up a knowledge base or appends a new knowledge to the knowledge base using pattern learning function. The expert system has main transformer fault, main transformer maintenance work and bus fault processing function. It is implemented as computer language, Visual C++ which has a dynamic programming function for implementing of inference engine and a MFC function for implementing of MMI. Finally, it's accuracy and effectiveness is proved by several event simulation works for a typical substation.

• Journal of Electrical Engineering and Technology
• /
• 제9권5호
• /
• pp.1729-1738
• /
• 2014

The offline verification of state of charge estimation, power estimation, fault diagnosis and emergency control of battery management system (BMS) is one of the key technologies in the field of electric vehicle battery system. It is difficult to test and verify the battery management system software in the early stage, especially for algorithms such as system state estimation, emergency control and so on. This article carried out the virtual environment modeling for verification of battery management system. According to the input/output parameters of battery management system, virtual environment is determined to run the battery management system. With the integration of the developed BMS model and the external model, the virtual environment model has been established for battery management system in the vehicle's working environment. Through the virtual environment model, the effectiveness of software algorithm of BMS was verified, such as battery state parameters estimation, power estimation, fault diagnosis, charge and discharge management, etc.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 1997년도 하계학술대회 논문집 D
• /
• pp.924-926
• /
• 1997

A high impedance fault(HIF) in a power system could be due to a downed conductor, and is a dangerous situation because the current may be too small to be detected by conventional means. In this paper, HIF(High impedance fault) and LIF(Low impedance fault) detection methods were reviewed. No single defection method can detect all electrical conditions resulting from downed conductor faults, because high impedance fault have arc phenomena, asymmetry and randomness. Neural network are well-suited for solving difficult signal processing and pattern recognition problem. This paper presents the application of artificial neural network(ANN) to detect the HIF and LIF. Test results show that the neural network was able to identify the high impedance fault by real-time operation. Furthermore, neural network was able to discriminate the HIF from the LIF.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제16권9호
• /
• pp.2927-2941
• /
• 2022

The primary task of machine fault diagnosis is to judge whether the current state is normal or damaged, so it is a typical binary classification problem with mutual exclusion. Mutually exclusive events and out-of-domain detection have one thing in common: there are two types of data and no intersection. We proposed a fusion model method to improve the accuracy of machine fault diagnosis, which is based on the mutual exclusivity of events and the commonality of out-of-distribution detection, and finally generalized to all binary classification problems. It is reported that the performance of a convolutional neural network (CNN) will decrease as the recognition type increases, so the variational auto-encoder (VAE) is used as the primary model. Two VAE models are used to train the machine's normal and fault sound data. Two reconstruction probabilities will be obtained during the test. The smaller value is transformed into a correction value of another value according to the mutually exclusive characteristics. Finally, the classification result is obtained according to the fusion algorithm. Filtering normal data features from fault data features is proposed, which shields the interference and makes the fault features more prominent. We confirm that good performance improvements have been achieved in the machine fault detection data set, and the results are better than most mainstream models.