Intelligent diagnosis for turbine blade faults using artificial neural networks and fuzzy logic

https://doi.org/10.1016/0952-1976(94)00082-XGet rights and content

Abstract

An on-line diagnostic system, which consists of three components: data acquisition, feature extraction, and pattern recognition, is described for detecting two main faults of turbine blades. Due to some significant noise in the signals, multiple vibration sensors are applied in order to obtain more-accurate results. The system identifies the existence of an unbalanced blade, or a losse blade, or the combination of these two faults. A feature-extraction algorithm, a frequency analyzer, was developed, and the features are formulated as the inputs of an artificial neural network using backpropagation. Fuzzy models are employed to dynamically update the training parameters, training rate, momentum, and steepness of the activation function, in order to speed up the training speed. The validity of the system was demonstrated with 90% discrimination accuracy in an experiment when multiple sensors were used. Moreover, the training time is reduced by 97.3% while all the three training parameters are dynamically updated.

References (23)

  • L.A. Zadeh

    Fuzzy sets

    Inform. Control.

    (1965)
  • T. Takagi et al.

    Derivation of fuzzy control rules from human operator's control actions

  • J. Hertz et al.

    Introduction to the Theory of Neural Computation

    (1991)
  • W.Y. Huang et al.

    Neural net and traditional classifiers

    Neural Informative Processing Systems

    (1988)
  • M. Black

    Vagueness: an exercise in logical analysis

    Philos. Sci.

    (1937)
  • T.D. Garvey

    A survey of AI approaches to the integration of information

  • R.C. Mann

    Multi-sensor integration using concurrent computing

  • S.S. Blackman

    Theoretical approaches to data association

  • R.C. Luo et al.

    Multisensor integration and fusion in intelligent systems

    IEEE Trans. Syst. Man Cybernet.

    (1989)
  • T. Pawlicki

    A neural architecture for evidence combination

  • G. Whittington et al.

    The application of a neural network model to sensor data fusion

  • Cited by (46)

    • Criticality analysis of petrochemical assets using risk based maintenance and the fuzzy inference system

      2019, Process Safety and Environmental Protection
      Citation Excerpt :

      The quantitative assessment of occupational safety risks is a safety feature (Pinto, 2014). The fuzzy inference system is the knowledge that describes the correct, timely or on-line diagnosis mechanism to detect pump failures (Azadeh et al., 2008) and turbine faults (Kuo, 1995). Ratnayake (2012) proposed a framework to integrate a large amount of data from different sources for optimal decision-making and inspection planning, and the focus is on static mechanical equipment in processes production plants.

    • A sparse auto-encoder-based deep neural network approach for induction motor faults classification

      2016, Measurement: Journal of the International Measurement Confederation
      Citation Excerpt :

      In the field of fault diagnosis, the majority of machine learning algorithms are supervised learning which needs a large amount of labeled data for training. Since neural networks possess strong representation ability due to its stacked hidden layers, they have been widely used as classifiers for machine fault diagnosis [10–12]. But the training of neural networks also needs a large amount of high-quality labeled data, and if the training samples are limited or cannot cover the testing distribution, the neural network may be easily overfitted which leads to a poor generalization especially for a complex classification problem.

    • Diagnostics of gear deterioration using EEMD approach and PCA process

      2015, Measurement: Journal of the International Measurement Confederation
      Citation Excerpt :

      Since the vibration measurements of the faulted machines normally exhibit high complexity, many researchers and engineers utilize the concepts of artificial intelligence and machine learning to identify the occurrence of machine malfunctions. Artificial neural network (ANN) and fuzzy logic techniques have been applied to classify the faults of gearbox systems and another rotating machinery [17,27–32]. Except for the ANN, the support vector machine (SVM) [15,20] and decision tree method [16,33] are also employed to classify the different fault types of gear transmission systems.

    • Application of a fuzzy inference system for functional failure risk rank estimation: RBM of rotating equipment and instrumentation

      2014, Journal of Loss Prevention in the Process Industries
      Citation Excerpt :

      Basically, a fuzzy membership function (MF) plots values of a crisp range between 0 and 1, which enables the degree of vagueness or uncertainty that has been associated with conventional crisp sets to be represented (Sharma et al., 2005). Researchers such as Bragli et al. (2003), Bertolini et al. (2009), Guo et al. (2009), Kuo (1995), Li, Chen, Daib, and Lib (2010), Mure and Demichela (2009), Pinto (2014), Ratnayake (2013), Schwartz, Kaufman, and Schwartz (2004), Seneviratne and Ratnayake (2013, 2012) and Suresh and Mujumdar (2004) have proposed risk prediction approaches incorporating fuzzy set theory. This manuscript illustrates the use of a fuzzy set theory-based consistent approach, estimating the FFR, to enhance the effectiveness of RBM planning.

    View all citing articles on Scopus
    View full text