• International Journal of Automotive Technology
• /
• v.8 no.5
• /
• pp.637-650
• /
• 2007

A new on-line fault detection and isolation(FDI) scheme has been proposed for engines using an adaptive neural network classifier; this paper investigates the robustness of this scheme by evaluating in a wide range of operational modes. The neural classifier is made adaptive to cope with the significant parameter uncertainty, disturbances, and environmental changes. The developed scheme is capable of diagnosing faults in the on-line mode and can be directly implemented in an on-board diagnosis system(hardware). The robustness of the FDI for the closed-loop system with crankshaft speed feedback is investigated by testing it for a wide range of operational modes, including robustness against fixed and sinusoidal throttle angle inputs, change in load, change in an engine parameter, and all changes occurring simultaneously. The evaluations are performed using a mean value engine model(MVEM), which is a widely used benchmark model for engine control system and FDI system design. The simulation results confirm the robustness of the proposed method for various uncertainties and disturbances.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.53 no.7
• /
• pp.103-109
• /
• 2016

Corrective control for asynchronous sequential machines is a novel automatic control theory that compensates illegal behavior or adverse effects of faults in the operation of existent asynchronous machines. In this paper, we propose a scheme of diagnosing and tolerating faults occurring to input channels of corrective control systems. The corrective controller can detect faults occurring in the input channel to the controlled machine, whereas those faults happening in the external input channel cannot be detected. The proposed scheme involves an outer operator which, upon receiving the state feedback, diagnoses a fault and sends an appropriate command signal to the controller for tolerating faults in the external input channel.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.19 no.3
• /
• pp.151-158
• /
• 2024

Although on-device artificial intelligence (AI) has gained attention to diagnosing machine faults in real time, most previous studies did not consider the model retraining and redeployment processes that must be performed in real-world industrial environments. Our study addresses this challenge by proposing an on-device AI-based real-time machine fault diagnosis system that utilizes continual learning. Our proposed system includes a lightweight convolutional neural network (CNN) model, a continual learning algorithm, and a real-time monitoring service. First, we developed a lightweight 1D CNN model to reduce the cost of model deployment and enable real-time inference on the target edge device with limited computing resources. We then compared the performance of five continual learning algorithms with three public bearing fault datasets and selected the most effective algorithm for our system. Finally, we implemented a real-time monitoring service using an open-source data visualization framework. In the performance comparison results between continual learning algorithms, we found that the replay-based algorithms outperformed the regularization-based algorithms, and the experience replay (ER) algorithm had the best diagnostic accuracy. We further tuned the number and length of data samples used for a memory buffer of the ER algorithm to maximize its performance. We confirmed that the performance of the ER algorithm becomes higher when a longer data length is used. Consequently, the proposed system showed an accuracy of 98.7%, while only 16.5% of the previous data was stored in memory buffer. Our lightweight CNN model was also able to diagnose a fault type of one data sample within 3.76 ms on the Raspberry Pi 4B device.

• Journal of Korean Society of Industrial and Systems Engineering
• /
• v.40 no.1
• /
• pp.57-64
• /
• 2017

In order to reduce damages to major railroad components, which have the potential to cause interruptions to railroad services and safety accidents and to generate unnecessary maintenance costs, the development of rolling stock maintenance technology is switching from preventive maintenance based on the inspection period to predictive maintenance technology, led by advanced countries. Furthermore, to enhance trust in accordance with the speedup of system and reduce maintenances cost simultaneously, the demand for fault diagnosis and prognostic health management technology is increasing. The objective of this paper is to propose a highly reliable learning model using various machine learning algorithms that can be applied to critical rolling stock components. This paper presents a model for railway rolling stock component fault diagnosis and conducts a mechanical failure diagnosis of motor components by applying the machine learning technique in order to ensure efficient maintenance support along with a data preprocessing plan for component fault diagnosis. This paper first defines a failure diagnosis model for rolling stock components. Function-based algorithms ANFIS and SMO were used as machine learning techniques for generating the failure diagnosis model. Two tree-based algorithms, RadomForest and CART, were also employed. In order to evaluate the performance of the algorithms to be used for diagnosing failures in motors as a critical railroad component, an experiment was carried out on 2 data sets with different classes (includes 6 classes and 3 class levels). According to the results of the experiment, the random forest algorithm, a tree-based machine learning technique, showed the best performance.

• Journal of the Institute of Electronics Engineers of Korea SD
• /
• v.43 no.9 s.351
• /
• pp.59-63
• /
• 2006

With the increasing complexity of VLSI devices, more complex faults have appeared. Many methods for diagnosing the single stuck-at fault have been studied. Often multiple defects on a foiling chip better reflect the reality. So, we propose an efficient diagnosis algorithm for multiple stuck-at faults. By using vectorwise intersections as an important metric of diagnosis, the proposed algorithm can diagnose multiple defects using single stuck-at fault simulator. In spite of multiple fault diagnosis, the number of candidate faults is also drastically reduced. For fault identification, positive calculations and negative calculations based on variable weights are used for the matching algorithm. Experimental results for ISCAS85 and full-scan version of ISCAS89 benchmark circuits prove the efficiency of the proposed algorithm.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.18 no.3
• /
• pp.367-372
• /
• 2008

In this paper we proposed a model of a fault diagnosis expert system with high reliability to compare identical well-functioning motors. The purpose of the survey was to determine if any differences exit among these identical motors and to identify exactly what these differences were, if in fact they were found. Using measured data for many identical brushless dc motors, this study attempted to find out whether normal and fault can be classified by each other. Measured data was analyzed using the State Transition Model (STM). Based on a proposed STM method, the effect of a different normal state is minimized and the detection of fault is improved in identical motor system. Experimental results are presented to prove that STM method could be a useful tool for diagnosing the condition of identical BLDE motors.

• 제어로봇시스템학회:학술대회논문집
• /
• 2005.06a
• /
• pp.1439-1444
• /
• 2005

Induction motors are the most commonly used electrical drives because they are rugged, mechanically simple, adaptable to widely different operating conditions, and simple to control. The most common faults in squirrel-cage induction motors are bearing, stator and rotor faults. Surveys conducted by the IEEE and EPRI show that the most common fault in induction motor is bearing failure (${\sim}$40% of failure). Thence, this paper addresses experimental results for diagnosing faults with different rolling element bearing damage via motor current spectral analysis. Rolling element bearings generally consist of two rings, an inner and outer, between which a set of balls or rollers rotate in raceways. We set the experimental test bed to detect the rolling-element bearing misalignment of 3 type induction motors with normal condition bearing system, shaft deflection system by external force and a hole drilled through the outer race of the shaft end bearing of the four pole test motor. This paper takes the initial step of investigating the efficacy of current monitoring for bearing fault detection by incipient bearing failure. The failure modes are reviewed and the characteristics of bearing frequency associated with the physical construction of the bearings are defined. The effects on the stator current spectrum are described and related frequencies are also determined. This is an important result in the formulation of a fault detection scheme that monitors the stator currents. We utilized the FFT, Wavelet analysis and averaging signal pattern by inner product tool to analyze stator current components. The test results clearly illustrate that the stator signature can be used to identify the presence of a bearing fault.

• Journal of Power Electronics
• /
• v.18 no.5
• /
• pp.1567-1576
• /
• 2018

The reliability of grid-connected inverters can be improved by algorithms capable of diagnosing faults in LCL filters. A fault diagnostic method during inverter startup is proposed. The proposed method can accurately generate and monitor information on the peak value and the location of the peak frequency component of the step response of a damped LCL filter. To identify faults, the proposed method compares the evaluated response with the response of a healthy higher-order damped LCL filter. The frequency components in the filter voltage response are first analytically obtained in closed form, which yields the expected trends for the filter faults. In the converter controller, the frequency components in the filter voltage response are computed using an appropriately designed fast Fourier transform and compared with healthy LCL response parameters using a finite state machine, which is used to sequence the proposed startup diagnostics. The performance of the proposed method is validated by comparing analytical results with the simulation and experimental results for a three-phase grid-connected inverter with a damped LCL filter.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.10 no.2
• /
• pp.283-288
• /
• 2015

The possible erroneous output data of the sensor nodes can cause the performance limit or the degradation of the reliability in the whole wireless sensor networks(WSN). In this paper, we propose a new sensor node scheme with multiple sensors and a new fault diagnostic algorithm. The algorithm can increase the reliability of the whole WSNs by utilizing measurements of the multiple sensors on the node and by determining the validity of the date by comparing the value of each sensor. It can increase the cost and complexity of the node, but is suitable for the area where the high reliability is critical.

• Transactions of the Korean Society of Mechanical Engineers
• /
• v.18 no.2
• /
• pp.323-338
• /
• 1994

This paper outlines a framework for performing intelligent sensor validation for a diagnostic expert system while reasoning under uncertainty. The emphasis is on the algorithmic preprocess technique. A companion paper focusses on heuristic post-processing. Sensor validation plays a vital role in the ability of the overall system to correctly detemine the state of a plant monitored by imperfect sensors. Especially, several theoretical developments were made in understanding uncertain sensory data in statistical aspect. Uncertain information in sensory values is represented through probability assignments on three discrete states, "high", "normal", and "low", and additional sensor confidence measures in Algorithmic Sv.Upper and lower warning limits are generated from the historical learning sets, which represents the borderlines for heat rate degradation generated in the Algorithmic SV initiates a historic data base for better reference in future use. All the information generated in the Algorithmic SV initiate a session to differentiate the sensor fault from the process fault and to make an inference on the system performance. This framework for a diagnostic expert system with sensor validation and reasonig under uncertainty applies in HEATXPRT$^{TM}$, a data-driven on-line expert system for diagnosing heat rate degradation problems in fossil power plants.