An approach to fault diagnosis of reciprocating compressor valves using Teager–Kaiser energy operator and deep belief networks (original) (raw)
Related papers
Proceedings of the Institution of Mechanical Engineers, Part C: Journal of Mechanical Engineering Science, 2017
In this paper, a hybrid deep belief network is proposed to diagnose single and combined faults of suction and discharge valves in a reciprocating compressor. This hybrid integrates the deep belief network structured by multiple stacked restricted Boltzmann machines for pre-training and simplified fuzzy ARTMAP (SFAM) for fault classification. In the pre-training procedure, an algorithm for selecting local receptive fields is used to group the most similar features into the receptive fields of which top values are the units of each layer, and then restricted Boltzmann machine is applied to these units to construct a network. Unsupervised learning is also carried out for each restricted Boltzmann machine layer in this procedure to compute the network weights and biases. Finally, the network output is fed into SFAM to perform fault classification. In order to diagnose the valve faults, three signal types of vibration, pressure, and current are acquired from a two-stage reciprocating air...
Reciprocating compressors are widely used in industry for various purposes and faults occurring in them can degrade performance, consume additional energy, and even cause severe damage to the machine. This paper will develop an automated approach to condition classification of a reciprocating compressor based on vibration measurements. Both the time domain and frequency domain techniques have been applied to the vibration signals and a large number of candidate features have been obtained based on previous studies. A subset selection method has then been used to configure a probabilistic neural network (PNN), with high computational efficiency, for effective fault classifications. The results show that a 95.50% correct classification between four different faulty cases is the best result when using a subset of frequency feature, whereas a 93.05% rate is the best for the subset from the time domain.
Intelligent Valve Fault Diagnosis Approach for Reciprocating Compressor Based on Acoustic Signals
2021
Reciprocating compressor is one of the critical components in petrochemical, process and gas storage /transportation industry. Due to the dynamic operating conditions, valve failures of the compressors are one of the most frequent failures that causes around 36% of the unplanned shutdowns for reciprocating compressors. For reliable operations of the compressor, a conditionbased fault diagnosis model is proposed in this study. Majority of the existing studies are based on vibration, pressure and other intrusive sensors, which interferes with the system dynamics. Hence, non-intrusive acoustic sensor-based signal analysis technique is tested in this study. Due to the non-stationary nature of the acoustic signal obtained from the working compressor, it is quite difficult to extract fault-specific information. Hence, Minimum Entropy Deconvolution Adjusted (MEDA)-Empirical Mode Decomposition (EMD) approach is proposed for valve-fault diagnosis of reciprocating compressor. After enhancing ...
Journal of Physics: Conference Series, 2011
Reciprocating compressors are widely used in industry for various purposes and faults occurring in them can degrade their performance, consume additional energy and even cause severe damage to the machine. Vibration monitoring techniques are often used for early fault detection and diagnosis, but it is difficult to prescribe a given set of effective diagnostic features because of the wide variety of operating conditions and the complexity of the vibration signals which originate from the many different vibrating and impact sources. This paper studies the use of genetic algorithms (GAs) and neural networks (NNs) to select effective diagnostic features for the fault diagnosis of a reciprocating compressor. A large number of common features are calculated from the time and frequency domains and envelope analysis. Applying GAs and NNs to these features found that envelope analysis has the most potential for differentiating three common faults: valve leakage, inter-cooler leakage and a loose drive belt. Simultaneously, the spread parameter of the probabilistic NN was also optimised. The selected subsets of features were examined based on vibration source characteristics. The approach developed and the trained NN are confirmed as possessing general characteristics for fault detection and diagnosis.
Journal of Physics: Conference Series, 2012
Traditional vibration monitoring techniques have found it difficult to determine a set of effective diagnostic features due to the high complexity of the vibration signals originating from the many different impact sources and wide ranges of practical operating conditions. In this paper Principal Component Analysis (PCA) is used for selecting vibration feature and detecting different faults in a reciprocating compressor. Vibration datasets were collected from the compressor under baseline condition and five common faults: valve leakage, inter-cooler leakage, suction valve leakage, loose drive belt combined with intercooler leakage and belt loose drive belt combined with suction valve leakage. A model using five PCs has been developed using the baseline data sets and the presence of faults can be detected by comparing the and values from the features of fault vibration signals with corresponding thresholds developed from baseline data. However, the-statistic procedure produces a better detection as it can separate the five faults completely.
2016
In the current economic challenge, methods to accurately predict system failure has become a holy grail in maintenance with the goal to reduce the cost of unavailability due to unscheduled shutdown. This has led to the current research with the aim to achieve a more accurate fault diagnosis for rotating machinery using a neural network (NN) with principal component analysis (PCA) as a pre-processing step to fuse multiple sensor data. The multisensor data fusion has been proven to improve the fault detection ability for machinery compared to single source condition monitoring. In this paper, an NN-based methodology is presented, where PCA is applied as preprocessing step to detect the rotating machinery faults during operation. The effectiveness of the proposed model is illustrated by a case study on two shaft industrial gas turbine where the real-time performance monitoring data collected from the plant and used to train and test the proposed algorithm. The analysis results show tha...
Intelligent fault diagnostic model for rotating machinery
2017 IEEE International Conference on Industrial Engineering and Engineering Management (IEEM), 2017
The aim of this paper is to present an intelligent fault diagnostic to assess the changes and detect malfunctions in rotating machinery using real-time data. The developed model interprets performance condition monitoring data and determines machine health status with the use of Artificial Neural Networks (ANN). The ANN networks were trained for principle performance parameters from which actual system performance can be predicted based on given set of input parameters. The validity of the proposed model was evaluated through a case study on twin-shaft 18700 kW industrial gas turbine engine to detect a fault happened in engine bell-mouth. The results showed the networks trained using Levenberg-Marquardt (LM) training function can achieve more accurate results compared to Bayesian regulation (BR) and scaled conjugate gradient (SCG) training functions. In addition, the results also showed that both power output parameter and the fuel flow rate can be effectively used to monitor the performance of gas turbine.
Neurocomputing
The effective fault diagnosis of rotating machinery is critical to ensure the continuous operation of equipment and is more economical than scheduled maintenance. Traditional signal processing-based and artificial intelligence-based methods, such as wavelet packet transform and support vector machine, have been proved effective in fault diagnosis of rotating machinery, which prevents unexpected machine breakdowns due to the failure of significant components. However, these methods have several disadvantages that make them unable to automatically and effectively extract valid fault features for the effective fault diagnosis of rotating machinery. A novel adaptive learning rate deep belief network combined with Nes-terov momentum is developed in this study for rotating machinery fault diagnosis. Nesterov momentum is adopted to replace traditional momentum to enable declining in advance and to improve training performance. Then, an individual adaptive learning rate method is used to select a suitable step length for accelerating descent. To confirm the utility of the proposed deep learning network architecture, two examinations are implemented on datasets from gearbox and locomotive bearing test rigs. Results indicate that the method achieves impressive performance in fault pattern recognition. Comparisons with existing methods are also conducted to demonstrate that the proposed method is more accurate and robust.
Mathematical Methods and Modeling in Machine Fault Diagnosis
Mathematical Problems in Engineering, 2014
Modern mathematics has often been utilized as an effective tool to model mechanical equipment so that their dynamic characteristics can be studied analytically. This will help identify potential failures of mechanical equipment by observing change in the equipment's dynamic parameters. On the other hand, dynamic signals are also important and provide reliable information about the equipment's working status. Modern mathematics has also provided us with a systematic way to design and implement various signal processing methods, which are used to analyze these dynamic signals, and to enhance intrinsic signal components that are directly related to machine failures. This special issue is aimed at stimulating not only new insights on mathematical methods for modeling but also recently developed signal processing methods, such as sparse decomposition with potential applications in machine fault diagnosis. The papers included in this special issue provide a glimpse into some of the research and applications in the field of machine fault diagnosis through applications of the modern mathematical methods.
Technological Evolution in the Fault Diagnosis of Rotating Machinery: A Review
Background: Rotating machineries are widely used in a various type of industrial applications. For reliable working of mechanical system, fault diagnosis of rotating element is the most important and crucial task. Over the past decades, fault diagnosis of mechanical systems has progressed with the evolution of machinery in terms of complexity and scale. High scale machinery require health monitoring and fault diagnosis to guarantee their reliable performance. Research on fault diagnosis of rotary machines has grown rapidly in the recent years. Objective of this paper is to review the recent research trends in the field of rotating machinery fault diagnosis in terms of these aspects: vibration generation mechanism, fault mechanism, signal acquisition and signal processing and fault diagnostics, artificial neural network, support vector machine etc.