Incipient fault detection and diagnosis using artificial neural networks

The objective of this paper is to present the development and numerical testing of a robust fault detection and identification (FDI) system using artificial neural networks (ANNs), for incipient (slowly developing) faults occurring in process systems. The challenge in using ANNs in FDI systems arises because of one's desire to detect faults of varying severity, faults from noisy sensors, and multiple simultaneous faults. To address these issues, it becomes essential to have a learning algorithm that ensures quick convergence to a high level of accuracy. A recently developed accelerated learning algorithm, namely a form of an adaptive back propagation (ABP) algorithm, is used for this purpose. The ABP algorithm is used for the development of an FDI system for a process composed of a direct current motor, a centrifugal pump, and the associated piping system. Simulation studies indicate that the FDI system has significantly high sensitivity to incipient fault severity, while exhibiting insensitivity to sensor noise. For multiple simultaneous faults, the FDI system detects the fault with the predominant signature. The major limitation of the developed FDI system is encountered when it is subjected to simultaneous faults with similar signatures. During such faults, the inherent limitation of pattern-recognition-based FDI methodsmore » becomes apparent. Thus, alternate, more sophisticated FDI methods become necessary to address such problems. Even though the effectiveness of pattern-recognition-based FDI methods using ANNs has been demonstrated, further testing using real-world data is necessary.« less

The diagnosis of incipient fault is important for power transformer condition monitoring. Incipient faults are monitored by conventional and artificial intelligence based models. Key gases, percentage value of gases, and ratio of the Doernenburg, Roger, IEC methods are input variables to artificial intelligence models, which affects the accuracy of incipient fault diagnosis, so selection of the most influencing relevant input variable is an important research area. With this main objective, Waikato Environment for Knowledge Analysis software is applied to 360 simulated samples having different operating lives to find the most influencing input parameters for incipient fault diagnosis in the gene expression programming model. The Waikato Environment for Knowledge Analysis identifies%C2H2,%C2H4,%CH4, C2H6/C2H2, C2H2/C2H4, CH4/H2, C2H4/C2H6, and C2H2/CH4 as the most relevant input variables in incipient fault diagnosis, and it is used for fault diagnosis using different artificial intelligence methods, i.e., artificial neural networks, fuzzy logic, support vector machines, and gene expression programming. The compared results shows that gene expression programming gives better results than the artificial neural network, fuzzy logic, and support vector machine with accuracy variation from 98.15 to 100%, proving the gene expression programming method can be used in transformer fault diagnosis research.

The traditional fault diagnosis expert system is dependent on knowledge acquisition of the experts. Knowledge acquisition is recognized as the problem of expert system. In addition, there are also some limitations of adaptive capacity, learning ability and real-time. And artificial neural network with good fault-tolerance and associative memory function, as well as very strong self-adaptive, self-learning ability, just can make up for the limitations of traditional expert system. This paper will construct a new expert system with the artificial neural network into and fault tree. Besides fault tree and neural network, this article mainly introduces the system model of fault diagnosis of the fire control computer and sensor subsystem, the method and process of fault diagnosis. In this expert system, we use the object-oriented production rule to represent the knowledge, which solves the bottleneck problem of the diagnostic knowledge acquisition effectively. The inferential process begins with the abnormal event and finally finds all of the possible faults and the faulty component. For some possible faulty components, which have large number of fault samples, the neural network model can be used to diagnose. The training network of fault samples employs the BP neural network. Finally, simulation training results show that the fault diagnosis expert system based on the combination of fault tree and neural network is rational and effective in fault diagnosis of the fire control system, realizes perfectly the combination of new knowledge and old one, and can grasp the state of systems dynamically.

On-line fault diagnosis system support for reactive scheduling in multipurpose batch chemical plants

In the development of highly automated and unmanned vehicles, the auto industry is faced with a new set of problems associated with the need to standardize elements of automated driving systems and determine the rules for their testing and certification. Due to the acceleration of scientific and technological progress and the development of new technologies, previously used standards began to become obsolete as the practice of their application accumulated, and sometimes even before the approval of a particular technical regulation. In these conditions, a promising technical and legal solution may be the use of fuzzy logic in tools for aggregation and management of expert knowledge in the processes of certification and testing of highly automated vehicles and elements of their automated control systems. Fuzzy decision-making models use typical fuzzy situations that form catalogs of scenarios for testing vehicles and their systems. In this way, an expert system knowledge base can be formed in which knowledge engineers apply a set of test scenario or experiment parameters for testing and simulation. Determining the parameters of a new scenario, their similarity with previously formalized scenarios and the decision to include a particular scenario in the scenario catalog remains with the experts. The approaches used at the present stage to the formalization of expert knowledge are not acceptable for creating knowledge bases controlled by big data management systems or artificial neural networks, which are in the near future in the development of expert systems. The article proposes a method by which a knowledge engineer administering an expert system can automate the creation of a catalog of “fuzzy” test scenarios and simulation modeling of systems, by automatically searching for the maximum value of the states of these systems belonging to maps and catalogs of scenarios, with a probability specified by the expert, with using machine learning methods.

This paper investigates a fault diagnosis expert system base on artificial neural network (ANN-FD-ES) for mixed-signal circuits. We built this system by given a faulty circuit. Based on the transient response testing (TRT) on the mixed-signal circuits, both analog and digital signal characteristics are unified. The system includes knowledge base, reasoning machine, explanatory, and interface. The diagnostic portion of the system is based on neural-calculation approach, which is used to establish the knowledge base composed of the weigh values and threshold values. With this approach, conclusions are developed directly from the outputs of the net, rather than the domain knowledge in traditional expert system. Compared with the traditional expert system, this system is good at processing data. The experimental results show that the system not only improves the shortcoming of the traditional expert system such as the deficiency on knowledge acquisition and self-learning capability, but also achieves a satisfied fault diagnosis efficiency.

We aimed to validate the Multinational Association for Supportive Care in Cancer (MASCC) risk index, and compare it with the Talcott model and artificial neural network (ANN) in predicting the outcome of febrile neutropenia in a Chinese population. We prospectively enrolled adult cancer patients who developed febrile neutropenia after chemotherapy and risk classified them according to MASCC score and Talcott model. ANN models were constructed and temporally validated in prospectively collected cohorts. From October 2005 to February 2008, 227 consecutive patients were enrolled. Serious medical complications occurred in 22% of patients and 4% died. The positive predictive value of low risk prediction was 86% (95% CI = 81-90%) for MASCC score ≥ 21, 84% (79-89%) for Talcott model, and 85% (78-93%) for the best ANN model. The sensitivity, specificity, negative predictive value, and misclassification rate were 81%, 60%, 52%, and 24%, respectively, for MASCC score ≥ 21; and 50%, 72%, 33%, and 44%, respectively, for Talcott model; and 84%, 60%, 58%, and 22%, respectively, for ANN model. The area under the receiver-operating characteristic curve was 0.808 (95% CI = 0.717-0.899) for MASCC, 0.573 (0.455-0.691) for Talcott, and 0.737 (0.633-0.841) for ANN model. In the low risk group identified by MASCC score ≥ 21 (70% of all patients), 12.5% developed complications and 1.9% died, compared with 43.3%, and 9.0%, respectively, in the high risk group (p < 0.0001). The MASCC risk index is prospectively validated in a Chinese population. It demonstrates a better overall performance than the Talcott model and is equivalent to ANN model.

Chronic kidney disease (CKD) can be treated if it is detected early, but as the disease progresses, recovery becomes impossible. Eventually, renal replacement therapy such as transplantation or dialysis is necessary. Ultrasound is a test method with which to diagnose kidney cancer, inflammatory disease, nodular disease, chronic kidney disease, etc. It is used to determine the degree of inflammation using information such as the kidney size and internal echo characteristics. The degree of the progression of chronic kidney disease in the current clinical trial is based on the value of the glomerular filtration rate. However, changes in the degree of inflammation and disease can even be observed with ultrasound. In this study, from a total of 741 images, 251 normal kidney images, 328 mild and moderate CKD images, and 162 severe CKD images were tested. In order to diagnose CKD in clinical practice, three ROIs were set: the cortex of the kidney, the boundary between the cortex and medulla, and the medulla, which are areas examined to obtain information from ultrasound images. Parameters were extracted from each ROI using the GLCM algorithm, which is widely used in ultrasound image analysis. When each parameter was extracted from the three areas, a total of 57 GLCM parameters were extracted. Finally, a total of 58 parameters were used by adding information on the size of the kidney, which is important for the diagnosis of chronic kidney disease. The artificial neural network (ANN) was composed of 58 input parameters, 10 hidden layers, and 3 output layers (normal, mild and moderate CKD, and severe CKD). Using the ANN model, the final classification rate was 95.4%, the epoch needed for training was 38 times, and the misclassification rate was 4.6%.

Breast cancer is one of the most leading causes of cancer deaths among women. Early detection of cancer increases the survival rate of the affected women. Machine learning approaches that are used for classification of breast cancer usually takes a lot of processing time during the training process. This paper attempts to propose a Machine Learning approach for breast cancer detection in mammograms, which does not depend on the number of training samples. The paper aims to develop a core vector machine-based diagnosis system for breast cancer detection using the date from MIAS. The main motivation behind using this system is to reduce the computational and memory requirement for large training data and to improve the classification accuracy. The proposed method has four stages: 1) Pre-processing is done to extract the breast region using global thresholding and enhancement using histogram equalization; 2) identification of potential mass using Otsu thresholding; 3) feature extraction using Laws Texture energy measures; and 4) mass detection is done using Core vector machine (CVM) classifier. Comparative analysis was done with different existing algorithms: Artificial Neural Network (ANN), Support Vector Machine (SVM), and Fuzzy Support Vector Machines (FSVM). The results illustrate that the proposed Core Vector Machine (CVM) classifier produced a promising result in terms of sensitivity (96.9%), misclassification rate (0.0443) and accuracy (95.89%). The time taken for training process is 0.0443, which is less when compared with other machine learning algorithms. Performance analysis shows that CVM classifier is superior to other classifiers like ANN, SVM and FSVM. The computational time of the CVM classifier during the training process was also analysed and found to be better than other discussed algorithms. The results achieved show that CVM classifier is the best algorithm for breast mass detection in mammograms.

Rising global temperatures tied to increases in greenhouse gas emissions are impacting high latitude regions, leading to changes in vegetation composition and feedbacks to climate through increased methane (CH4) emissions. In subarctic peatlands, permafrost collapse has led to shifts in vegetation species on landscape scales with high spatial heterogeneity. Our goal was to provide a baseline for vegetation distribution related to permafrost collapse and changes in biogeochemical processes. We collected unmanned aerial system (UAS) imagery at Stordalen Mire, Abisko, Sweden to classify vegetation cover types. A series of digital image processing routines were used to generate texture attributes within the image for the purpose of characterizing vegetative cover types. An artificial neural network (ANN) was developed to classify the image. The ANN used all texture variables and color bands (three spectral bands and six metrics) to generate a probability map for each of the eight cover classes. We used the highest probability for a class at each pixel to designate the cover type in the final map. Our overall misclassification rate was 32%, while omission and commission error by class ranged from 0% to 50%. We found that within our area of interest, cover classes most indicative of underlying permafrost (hummock and tall shrub) comprised 43.9% percent of the landscape. Our effort showed the capability of an ANN applied to UAS high-resolution imagery to develop a classification that focuses on vegetation types associated with permafrost status and therefore potentially changes in greenhouse gas exchange. We also used a method to examine the multiple probabilities representing cover class prediction at the pixel level to examine model confusion. UAS image collection can be inexpensive and a repeatable avenue to determine vegetation change at high latitudes, which can further be used to estimate and scale corresponding changes in CH4 emissions.

Artificial neural network (ANN) based prediction and optimization of an organic Rankine cycle (ORC) for diesel engine waste heat recovery

Load demand forecasting is a broad branch of electric power systems engineering. In the last few decades, hundreds of methods have been suggested by many brilliant researchers around the world to improve the current forecasting tools. These tools are categorized as: deterministic, probabilistic, stochastic, and artificial intelligent (AI) algorithms. Among these approaches, artificial neural networks (ANNs) proved themselves as very competitive and highly precise methods to accurately forecast energy. ANNs are divided into two main types called: feed-forward networks and recurrent/feedback networks, where each type of them has multiple sub-types. This study tries to improve the performance of any existing type/sub-type of ANNs by optimizing its configuration through using the biogeography-based optimization (BBO) algorithm. The number of input variables, layers, neurons, and the types of activation functions and training algorithm all are optimized. The goal is to preserve the simplicity, so only very simple multi-layer feed-forward ANNs are used instead of using time-series-based feed-forward/feedback ANNs. To prove the effectiveness of hybridizing ANNs with evolutionary algorithms (EAs), numerical simulations are carried-out on some Nova Scotia's loads. The results obtained from these optimally configured ANN s are highly significant, and thus they confirm that allegation.

Fault diagnosis support system for complex chemical plants

Incipient fault detection and diagnosis (FDD) is an important measure to improve the efficient, safe and stable operation of high-speed trains. This paper proposes a data-driven FDD method, namely deep slow feature analysis and belief rule base method (DSFA-BRB), for the running gears of high-speed trains. The method uses two kinds of statistics to perform fault detection on the multi-dimensional data of the running gears. In addition, the characteristics of more accurate data are extracted, which greatly reduces the complexity of constructing a diagnostic and quantitative model. Further, by constructing a BRB model combining expert knowledge and data, it is possible to avoid misjudgment caused by data incompleteness. Compared with the traditional methods, the DSFA-BRB algorithm has better performance in reducing fault alarm probability. Finally, the validity of the algorithm is verified by the actual running gears system.

Active vibration isolation technology has been widely used to reduce vibration transmission in many different engineering systems. Magnetic suspension isolator (MSI), as an active isolation actuator, has shown advantages including non‐contact, high response frequency, high reliability and long life‐span. However, its potential has not been fully explored due to the nonlinear and hysteretic behavior in a dynamic environment, and there is limited research work in the area. This paper proposes a new artificial neural network (ANN)‐based approach to model the dynamics of MSI. A modified genetic algorithm (MGA) is developed to train the ANN to improve the model accuracy. Results clearly show that the ANN model with the MGA approach outperforms the back propagation (BP) approach and the analytic method based on the least squares fitting method.