Assessment of the Robustness of Gas Turbine Diagnostics Tools Based on Neural Networks

2006 ◽  
Author(s):  
R. Bettocchi ◽  
M. Pinelli ◽  
P. R. Spina ◽  
M. Venturini ◽  
G. A. Zanetta
Keyword(s):  
Neural Networks ◽  
Gas Turbine ◽  
Measurement Errors ◽  
Health State ◽  
Artificial Intelligence Technique ◽  
Boundary Load ◽  
Set Up ◽  
Operating Points ◽  
Range Of Variation ◽  
Selection Of

The paper deals with the set-up and the application of an Artificial Intelligence technique based on Neural Networks (NNs) to gas turbine diagnostics, in order to evaluate its capabilities and its robustness. The data used for both training and testing the NNs were generated by means of a Cycle Program, calibrated on a Siemens V94.3A gas turbine. Such data are representative of operating points characterized by different boundary, load and health state conditions. The analyses carried out are aimed at the selection of the most appropriate NN structure for gas turbine diagnostics, by evaluating NN robustness with respect to: • interpolation capability and accuracy in the presence of data affected by measurement errors; • extrapolation capability in the presence of data lying outside the range of variation adopted for NN training; • accuracy in the presence of input data corrupted by bias errors; • accuracy when one input is not available. This situation is simulated by replacing the value of the unavailable input with its nominal value.

A Multiple Operating Point Minimization Technique for Gas Turbine Operating State Determination

2002 ◽  
Author(s):  
M. Pinelli ◽  
P. R. Spina ◽  
M. Venturini
Keyword(s):  
Gas Turbine ◽  
Field Application ◽  
Operating Point ◽  
Health State ◽  
Characteristic Parameters ◽  
Estimation Errors ◽  
State Determination ◽  
Minimization Technique ◽  
Machine Health ◽  
Operating Points

Gas turbine operating state determination can be performed using Gas Path Analysis (GPA) techniques, which use measurements taken on the machine to calculate the characteristic parameters that are indices of the machine health state. The number and type of characteristic parameters that can be evaluated depend on the number and type of the available measured variables. Thus, when there are not enough measured variables to determine all the characteristic parameters, some of them have to be estimated independently of the actual gas turbine health state. In this way, variations due to aging or deterioration which, in the actual machine, may occur on these last characteristic parameters, cause estimation errors on the characteristic parameters assumed as problem unknowns. In the field application of GPA techniques the available instrumentation is often inadequate to ensure reliable operating state analysis. This problem may be partially overcome using a multiple operating point minimization technique. This consists of the determination of the characteristic parameters that minimize the sum of the square differences between measured and computed values of the measurable variables in multiple operating points. In this way the lack of data is overcome by data obtained in different operating points. This paper describes a procedure for gas turbine operating state determination based on a multiple operating point minimization technique and presents a study aimed at selecting the best set and number of operating points that have to be used.

Neural Networks for the Diagnostics of Gas Turbine Engines

1996 ◽  
Author(s):  
G. Torella ◽  
G. Lombardo
Keyword(s):  
Neural Networks ◽  
Gas Turbine ◽  
Back Propagation ◽  
Learning Rate ◽  
Turbine Engines ◽  
Gas Turbine Engines ◽  
Turbine Engine ◽  
Classical Architecture ◽  
Set Up ◽  
The One

The paper describes the activities carried out for developing and testing Back Propagation Neural Networks (BPNN) for the gas turbine engine diagnostics. One of the aims of this study was to analyze the problems encountered during training using large number of patterns. Each pattern contains information about the engine thermodynamic behaviour when there is a fault in progress. Moreover the research studied different architectures of BPNN for testing their capability to recognize patterns even when information is noised. The results showed that it is possible to set-up and optimize suitable and robust Neural Networks useful for gas turbine diagnostics. The methods of Gas Path Analysis furnish the necessary data and information about engine behaviour. The best architecture, among the ones studied, is formed by 13, 26 and 47 neurons in the input, hidden and output layer respectively. The investigated Nets have shown that the best encoding of faults is the one using a unitary diagonal matrix. Moreover the calculation have identified suitable laws of learning rate factor (LRF) for improving the learning rate. Finally the authors used two different computers. The first one has a classical architecture (sequential, vectorial and parallel). The second one is the Neural Computer, SYNAPSE-1, developed by Siemens.

Optimized Operating Point Selection for Gas Turbine Health State Analysis by Using a Multi-Point Technique

2003 ◽  
Author(s):  
M. Pinelli ◽  
P. R. Spina ◽  
M. Venturini
Keyword(s):  
Gas Turbine ◽  
Operating Point ◽  
Health State ◽  
Characteristic Parameters ◽  
Point Selection ◽  
Estimation Errors ◽  
State Determination ◽  
Minimization Technique ◽  
Operating Points ◽  
State Analysis

Gas turbine operating state determination can be performed using Gas Path Analysis (GPA) techniques, which use measurements taken on the machine to calculate the characteristic parameters that are indices of the machine health state. The number and type of characteristic parameters that can be evaluated depend on the number and type of the available measured variables. Thus, when there are not enough measured variables to determine all the characteristic parameters, some of them have to be estimated independently of the actual gas turbine health state. In this way, variations due to aging or deterioration which, in the actual machine, may occur on these last characteristic parameters, cause estimation errors on the characteristic parameters assumed as problem unknowns. The available instrumentation in field applications is often inadequate to ensure reliable operating state analysis when GPA-based techniques are used. This problem may be partially overcome using a multiple operating point minimization technique. This consists of the determination of the characteristic parameters that minimize the sum of the square differences between measured and computed values of the measurable variables in multiple operating points. In this way the lack of data is overcome by data obtained in different operating points. This paper describes a procedure for gas turbine operating state determination based on a multiple operating point minimization technique and presents a study aimed at selecting the best set and number of operating points that should be used.

Platelet Anti-Aggregating Activity and Tolerance of Clopidogrel in Atherosclerotic Patients

1996 ◽  
Vol 76 (06) ◽  
pp. 0939-0943 ◽  
Author(s):  
B Boneu ◽  
G Destelle ◽  
Keyword(s):  
Platelet Aggregation ◽  
Large Scale ◽  
Bleeding Time ◽  
Antithrombotic Activity ◽  
Ambulatory Patients ◽  
Patients At Risk ◽  
Set Up ◽  
Ischemic Events ◽  
Large Scale Clinical Trial ◽  
Selection Of

SummaryThe anti-aggregating activity of five rising doses of clopidogrel has been compared to that of ticlopidine in atherosclerotic patients. The aim of this study was to determine the dose of clopidogrel which should be tested in a large scale clinical trial of secondary prevention of ischemic events in patients suffering from vascular manifestations of atherosclerosis [CAPRIE (Clopidogrel vs Aspirin in Patients at Risk of Ischemic Events) trial]. A multicenter study involving 9 haematological laboratories and 29 clinical centers was set up. One hundred and fifty ambulatory patients were randomized into one of the seven following groups: clopidogrel at doses of 10, 25, 50,75 or 100 mg OD, ticlopidine 250 mg BID or placebo. ADP and collagen-induced platelet aggregation tests were performed before starting treatment and after 7 and 28 days. Bleeding time was performed on days 0 and 28. Patients were seen on days 0, 7 and 28 to check the clinical and biological tolerability of the treatment. Clopidogrel exerted a dose-related inhibition of ADP-induced platelet aggregation and bleeding time prolongation. In the presence of ADP (5 \lM) this inhibition ranged between 29% and 44% in comparison to pretreatment values. The bleeding times were prolonged by 1.5 to 1.7 times. These effects were non significantly different from those produced by ticlopidine. The clinical tolerability was good or fair in 97.5% of the patients. No haematological adverse events were recorded. These results allowed the selection of 75 mg once a day to evaluate and compare the antithrombotic activity of clopidogrel to that of aspirin in the CAPRIE trial.

Using Neural Networks in Controlling Low- and Medium-Capacity Gas-Turbine Plants

2019 ◽  
Vol 90 (11) ◽  
pp. 737-740 ◽  
Author(s):  
B. V. Kavalerov ◽  
I. V. Bakhirev ◽  
G. A. Kilin
Keyword(s):  
Neural Networks ◽  
Gas Turbine

STUDY OF INACCURACY IN THE OPERATION OF THE MAIN CONTROLLER OF STB MACHINE

2019 ◽  
Keyword(s):  
Normal Distribution ◽  
Stable Process ◽  
Random Errors ◽  
Actual Length ◽  
Experimental Data Processing ◽  
Set Up ◽  
Main Regulator ◽  
Transmission Gear ◽  
Selection Of

At production of fabrics, including fabrics for agricultural purpose, an important role is played by the cor-rect adjustment of operation of machine main regulator. The quality of setup of machine main controller is determined by the proper selection of rotation angle of warp beam weaving per one filling thread. In the pro-cess of using the regulator as a result of mistakes in adjustment, wear of transmission gear and backlashes in connections of details there are random changes in threads length. The purpose of the article is the research of property of random errors of basis giving by STB machine regulator. Mistakes can be both negative, and positive. In case of emergence only negative or only positive mistakes operation of the machine becomes im-possible as there will be a consecutive accumulation of mistakes. As a result of experimental data processing for stable process of weaving and the invariable diameter of basis threads winding of threads it is revealed that the random error of giving is set up as linear function of the accidental length having normal distribution. Measurements of accidental deviations in giving of a basis by the main regulator allowed to construct a curve of normal distribution of its actual length for one pass of weft thread. The presented curve of distribution of random errors in giving of a basis is the displaced curve of normal distribution of the accidental sizes. Also we define the density of probability of normal distribution of basis giving errors connected with a margin er-ror operation of the main regulator knowing of which allows to plan ways of their decrease that is important for improvement of quality of the produced fabrics.

scholarly journals A New Hyper-Parameter Optimization Method for Power Load Forecast Based on Recurrent Neural Networks

Algorithms ◽  
2021 ◽  
Vol 14 (6) ◽  
pp. 163
Author(s):  
Yaru Li ◽  
Yulai Zhang ◽  
Yongping Cai
Keyword(s):  
Neural Networks ◽  
Parameter Optimization ◽  
Recurrent Neural Networks ◽  
Critical Role ◽  
Optimization Method ◽  
Bayesian Optimization ◽  
Power Load ◽  
Gradient Calculation ◽  
Truncated Newton ◽  
Selection Of

The selection of the hyper-parameters plays a critical role in the task of prediction based on the recurrent neural networks (RNN). Traditionally, the hyper-parameters of the machine learning models are selected by simulations as well as human experiences. In recent years, multiple algorithms based on Bayesian optimization (BO) are developed to determine the optimal values of the hyper-parameters. In most of these methods, gradients are required to be calculated. In this work, the particle swarm optimization (PSO) is used under the BO framework to develop a new method for hyper-parameter optimization. The proposed algorithm (BO-PSO) is free of gradient calculation and the particles can be optimized in parallel naturally. So the computational complexity can be effectively reduced which means better hyper-parameters can be obtained under the same amount of calculation. Experiments are done on real world power load data,where the proposed method outperforms the existing state-of-the-art algorithms,BO with limit-BFGS-bound (BO-L-BFGS-B) and BO with truncated-newton (BO-TNC),in terms of the prediction accuracy. The errors of the prediction result in different models show that BO-PSO is an effective hyper-parameter optimization method.

Sign in / Sign up

Export Citation Format

Share Document