scholarly journals Gas Turbine and Sensor Fault Diagnosis With Nested Artificial Neural Networks

2004 ◽  
Author(s):  
N. Xiradakis ◽  
Y. G. Li
Keyword(s):  
Neural Network ◽  
Neural Networks ◽  
Artificial Neural Networks ◽  
Gas Turbine ◽  
Diagnostic System ◽  
Network System ◽  
Fault Recovery ◽  
Sensor Fault ◽  
Engine Component ◽  
Neural Network System

Accurate gas turbine diagnosis relies on accurate measurements from sensors. Unfortunately, sensors are prone to degradation or failure during gas turbine operations. In this paper a stack of decentralised artificial neural networks are introduced and investigated as an approach to approximate the measurement of a failed sensor once it is detected. Such a system is embedded into a nested neural network system for gas turbine diagnosis. The whole neural network diagnostic system consists of a number of feedforward neural networks for engine component diagnosis, sensor fault detection and isolation; and a stack of decentralised neural networks for sensor fault recovery. The application of the decentralised neural networks for the recovery of any failed sensor has the advantage that the configuration of the nested neural network system for engine component diagnosis is relatively simple as the system does not take into account sensor failure. When a sensor fails, the biased measurement of the failed sensor is replaced with a recovered measurement approximated with the measurements of other healthy sensors. The developed approach has been applied to an engine similar to the industrial 2-shaft engine, GE LM2500+, whose performance and training samples are simulated with an aero-thermodynamic modelling tool — Cranfield University’s TURBOMATCH computer program. Analysis shows that the use of the stack of decentralised neural networks for sensor fault recovery can effectively recover the measurement of a failed sensor. Comparison between the performance of the diagnostic system with and without the decentralised neural networks shows that the sensor recovery can improve the performance of the neural network engine diagnostic system significantly when a sensor fault is present.

scholarly journals CONNECTIONIST TECHNIQUES FOR THE IDENTIFICATION AND SUPPRESSION OF INTERFERING UNDERLYING FACTORS

2003 ◽  
Vol 17 (08) ◽  
pp. 1447-1466 ◽  
Author(s):  
EMILIO CORCHADO ◽  
COLIN FYFE
Keyword(s):  
Neural Networks ◽  
Artificial Neural Networks ◽  
Difficult Problem ◽  
Network System ◽  
Visual Inputs ◽  
Neural Network System ◽  
Underlying Causes ◽  
Artificial Neural ◽  
Vertical Bars ◽  
The Individual

We consider the difficult problem of identification of independent causes from a mixture of them when these causes interfere with one another in a particular manner: those considered are visual inputs to a neural network system which are created by independent underlying causes which may occlude each other. The prototypical problem in this area is a mixture of horizontal and vertical bars in which each horizontal bar interferes with the representation of each vertical bar and vice versa. Previous researchers have developed artificial neural networks which can identify the individual causes; we seek to go further in that we create artificial neural networks which identify all the horizontal bars from only such a mixture. This task is a necessary precursor to the development of the concept of "horizontal" or "vertical".

The Health Monitoring of Bridges using Artificial Neural Networks

2001 ◽  
Vol 17 (3) ◽  
pp. 157-166
Author(s):  
Pei-Ling Liu ◽  
Shyh-Jang Sun
Keyword(s):  
Neural Network ◽  
Neural Networks ◽  
Monitoring System ◽  
Health Monitoring ◽  
Network System ◽  
Bridge Structure ◽  
Constant Mass ◽  
Simulation Data ◽  
Neural Network System ◽  
The Neural Networks

ABSTRACTThis study develops a neural network system to monitor the safety of a bridge structure. A truck of constant mass is driven at constant speed through the target bridge. Then, the maximal and minimal values of the bridge elongations are processed by a monitoring system to evaluate the current condition of the bridge. The monitoring system is composed of parallel backpropagation neural networks. Each neural network monitors a part of the bridge. The neural networks are trained using simulation data. The numerical example shows that the monitoring system is effective in the damage detection of the bridge.

scholarly journals Neural Network System Design for Predicting MIN Reliability

2020 ◽  
Vol 9 (7) ◽  
pp. 1216-1221
Keyword(s):  
Neural Network ◽  
Neural Networks ◽  
System Design ◽  
Interconnection Networks ◽  
Network System ◽  
Multistage Interconnection Networks ◽  
Number Of Layers ◽  
Neural Network System ◽  
Improved Performance ◽  
Number Of Iterations

fforts have been made to examine and study different path and multi-path Multistage Interconnection Networks (MIN) possessing regular or irregular topology. Numerous strategies for establishing fault-tolerance in MINs have also been studied. These studies have provided us help to understand the strength and weakness of the existing static and dynamic and regular and irregular MINs. Application of Neural Networks leads to the development of MINs with improved performance and study of its Reliability In this paper ANN based system has been developed which will help in the study of metrics required for enhancing and predicting the reliability of MINs. In this paper Number of iterations are conducted to improve the ANN based system to predict the reliability of MINs by changing the number of neurons and the number of layers.

Application of Neural Network in Letter Recognition Using the Perceptron Method

2021 ◽  
Vol 11 (02) ◽  
pp. 53-65
Author(s):  
Bagus Suteja
Keyword(s):  
Neural Network ◽  
Neural Networks ◽  
Letter Recognition ◽  
Network System ◽  
Learning Method ◽  
Neural Network System

Modeling with neural networks is the learning and adjustment of an object. The perceptron method is a learning method with supervision in a neural network system. In designing a neural network that needs to be considered is the number of specifications that will be identified. A neural network consists of a number of neurons and a number of inputs. To identify some letters, it takes several neurons to distinguish them. These neurons will generate a combination value that is used to identify the letters. so that the resulting network must have parameters that can be set by changing through the rules of learning with supervision.

Gas turbine gas-path fault identification using nested artificial neural networks

2018 ◽  
Vol 90 (6) ◽  
pp. 992-999 ◽  
Author(s):  
Amare D. Fentaye ◽  
Aklilu T. Baheta ◽  
Syed Ihtsham Ul-Haq Gilani
Keyword(s):  
Neural Network ◽  
Neural Networks ◽  
Artificial Neural Networks ◽  
Gas Turbine ◽  
Diagnostic Technique ◽  
Identification Accuracy ◽  
Fault Identification ◽  
Content Type ◽  
Artificial Neural ◽  
Fault Diagnostic

Purpose The purpose of this paper is to present a quantitative fault diagnostic technique for a two-shaft gas turbine engine applications. Design/methodology/approach Nested artificial neural networks (NANNs) were used to estimate the progressive deterioration of single and multiple gas-path components in terms of mass flow rate and isentropic efficiency indices. The data required to train and test this method are attained from a thermodynamic model of the engine under steady-state conditions. To evaluate the tolerance of the method against measurement uncertainties, Gaussian noise values were considered. Findings The test results revealed that this proposed method is capable of quantifying single, double and triple component faults with a sufficiently high degree of accuracy. Moreover, the authors confirmed that NANNs have derivable advantages over the single structure-based methods available in the public domain, particularly over those designed to perform single and multiple faults together. Practical implications This method can be used to assess engine’s health status to schedule its maintenance. Originality/value For complicated gas turbine diagnostic problems, the conventional single artificial neural network (ANN) structure-based fault diagnostic technique may not be enough to get robust and accurate results. The diagnostic task can rather be better done if it is divided and shared with multiple neural network structures. The authors thus used seven decentralized ANN structures to assess seven different component fault scenarios, which enhances the fault identification accuracy significantly.

scholarly journals Discussion forH-Matrices and It’s Application

2014 ◽  
Vol 2014 ◽  
pp. 1-6
Author(s):  
Guichun Han ◽  
Huishuang Gao ◽  
Haitao Yang
Keyword(s):  
Neural Network ◽  
Neural Networks ◽  
Network System ◽  
Numerical Examples ◽  
Diagonally Dominant Matrices ◽  
Neural Network System ◽  
Diagonally Dominant ◽  
The Stability

NonsingularH-matrices and positive stable matrices play an important role in the stability of neural network system. In this paper, some criteria for nonsingularH-matrices are obtained by the theory of diagonally dominant matrices and the obtained result is introduced into identifying the stability of neural networks. So the criteria for nonsingularH-matrices are expanded and their application on neural network system is given. Finally, the effectiveness of the results is illustrated by numerical examples.

scholarly journals Predicting various forms of endometriosis using artificial neural networks

2020 ◽  
Vol 35 (4) ◽  
pp. 143-149
Author(s):  
N. G. Sazonova ◽  
T. A. Makarenko ◽  
A. N. Narkevich
Keyword(s):  
Neural Network ◽  
Neural Networks ◽  
Differential Diagnosis ◽  
Artificial Neural Networks ◽  
Mathematical Models ◽  
Computer System ◽  
Clinical Manifestations ◽  
Diagnostic System ◽  
Artificial Neural ◽  
Computer Diagnostic System

Introduction. Endometriosis is a difficult-to-diagnose pathology due to the diversity of clinical manifestations and the lack of high-precision markers necessary for rapid noninvasive diagnosis and timely administration of pathogenetically justified treatment.The aim of this work was to develop a computer system that allows us to assess the probability of endometriosis with various localizations in women, based on artificial neural networks.Material and Methods. The neural network mathematical models were constructed and tested based on data from 110 patients with morphologically pre-confirmed endometriosis. Patients were divided into training and test samples. The models were built based on anamnestic data and results of proteomic and enzyme immunoassays in blood plasma samples.Results and Discussion. In the course of the study, four mathematical models of neural networks were constructed to predict the presence or absence of endometriosis in a woman and its localization if present. Based on these mathematical models, a computer system “Differential diagnosis of endometriosis” was developed. This system allowed to assess the probability and localization of endometriosis in a patient based on parameters obtained as a result of neural network training.Conclusion. The developed computer diagnostic system allowed predicting the presence of endometriosis and its localization with a probability over 80%, depending on the predicted localization, based on data about the patient and the results of her examination. This system may be used for differential diagnosis of endometriosis from other diseases of the female reproductive system, as well as for differential diagnosis of various endometriosis localizations.

Creep Life Prediction for Aero Gas Turbine Hot Section Component Using Artificial Neural Networks

2013 ◽  
Vol 136 (3) ◽  
Author(s):  
M. F. Abdul Ghafir ◽  
Y. G. Li ◽  
L. Wang
Keyword(s):  
Neural Network ◽  
Neural Networks ◽  
Artificial Neural Networks ◽  
Gas Turbine ◽  
Life Prediction ◽  
Creep Life ◽  
Model Based ◽  
Creep Life Prediction ◽  
Artificial Neural ◽  
Prediction Approach

Accurate and reliable component life prediction is crucial to ensure safety and economics of gas turbine operations. In pursuit of such improved accuracy and reliability, model-based creep life prediction methods have become more complicated and demand higher computational time. Therefore, there is a need to find an alternative approach that is able to provide a quick solution to creep life prediction for production engines while at the same time maintain the same accuracy and reliability as that of the model-based methods. In this paper, a novel creep life prediction approach using artificial neural networks is introduced as an alternative to the model-based creep life prediction approach to provide a quick and accurate estimation of gas turbine creep life. Multilayer feed forward backpropagation neural networks have been utilized to form three neural network–based creep life prediction architectures known as the range-based, functional-based, and sensor-based architectures. The new neural network creep life prediction approach has been tested with a model single-spool turboshaft gas turbine engine. The results show that good generalization can be achieved in all three neural network architectures. It was also found that the sensor-based architecture is better than the other two in terms of accuracy, with 98% of the post-test samples possessing prediction errors within ±0.4%.

Optimum Design of Structures for Earthquake Loading by a Cellular Evolutionary Algorithm and Neural Networks

2012 ◽  
pp. 306-322 ◽  
Author(s):  
Saeed Gholizadeh
Keyword(s):  
Neural Network ◽  
Genetic Algorithm ◽  
Neural Networks ◽  
Optimum Design ◽  
Optimization Algorithm ◽  
Time History ◽  
Network System ◽  
Hybrid Neural Network ◽  
Neural Network System ◽  
Nonlinear Time History

The present chapter deals with optimum design of structures for earthquake induced loads by taking into account nonlinear time history structural response. As the structural seismic optimization is a time consuming and computationally intensive task, in this chapter, a methodology is proposed to reduce the computational burden. The proposed methodology consists of an efficient optimization algorithm and a hybrid neural network system to effectively predict the nonlinear time history responses of structures. The employed optimization algorithm is a modified cellular genetic algorithm which reduces the required generation numbers compared with the standard genetic algorithm. Also, the hybrid neural network system is a combination of probabilistic and generalized regression neural networks. Numerical results demonstrate the computational merits of the proposed methodology for seismic design optimization of structures.

Hierarchical Neural Networks for Modelling Adaptive Financial Systems

2006 ◽  
pp. 109-123
Author(s):  
Masoud Mohammadian ◽  
Mark Kingham
Keyword(s):  
Neural Network ◽  
Neural Networks ◽  
Interest Rate ◽  
Interest Rates ◽  
Network System ◽  
Financial Systems ◽  
Neural Network System ◽  
Hierarchical Neural Network ◽  
Hierarchical Neural Networks ◽  
Accuracy Of Prediction

In this chapter, an intelligent hierarchical neural network system for prediction and modelling of interest rates in Australia is developed. A hierarchical neural network system is developed to model and predict 3 months’ (quarterly) interest-rate fluctuations. The system is further trained to model and predict interest rates for 6-month and 1-year periods. The proposed system is developed with first four and then five hierarchical neural networks to model and predict interest rates. Conclusions on the accuracy of prediction using hierarchical neural networks are also reported.

Sign in / Sign up

Export Citation Format

Share Document