Application of Machine Learning Based Surrogate Model for Prediction of Sectional Temperature of Radially Cooled Gas Turbine Blades

2021 ◽  
Author(s):  
Rishabh Shrivastava ◽  
Nisha Tamar ◽  
Amit Grover ◽  
Debdulal Das
Keyword(s):  
Machine Learning ◽  
Linear Regression ◽  
Mean Absolute Error ◽  
Turbine Blades ◽  
Absolute Error ◽  
Inlet Temperature ◽  
Turbine Inlet Temperature ◽  
Gas Turbine Blades ◽  
Thermal Calculations ◽  
Cooling Air

Abstract Accurate thermal prediction of gas turbine blades is essential to ensure successful operation throughout the design life. Large Gas turbines operate in different conditions based on customer requirements, due to which turbine blades are subjected to variations in thermal loading conditions. Simulating this behavior using conventional finite element modeling involves detailed and time-consuming analyses for calculation of blade temperature, which can be further utilized to assess cyclic and creep life. This paper deals with developing and utilizing machine learning based surrogate models to predict the sectional temperature (output) of a radially cooled blade. The surrogate models are developed to predict the output using turbine inlet temperature, hot gas mass flow, cooling air temperature and cooling air mass flow as input to the machine learning (ML) model. All thermal parameters for ML model have been obtained from CFD based 3D thermal calculations. A comparative study is presented between linear regression, decision tree, random forest, and gradient boost ML models, to select the model with the least mean absolute error. Additionally, hyperparameter optimization is performed using grid search to minimize the error. The results show that the linear regression-based model outputs the least mean absolute error of 6.5°C and the highest dependence of the output is on the turbine inlet temperature, followed by the cooling air temperature. The findings show a good agreement between the predicted output of the surrogate model and multi-dimensional physics based thermal calculations, while offering a considerate reduction in analysis time.

Machine Learning Based Classification Model for Predicting Cyclic Failure in Radially Cooled Gas Turbine Blades

2021 ◽  
Author(s):  
Rishabh Shrivastava ◽  
Ankush Kapoor ◽  
Stuti Kaushal ◽  
Amit Yadav ◽  
Pavankumar Vodnala
Keyword(s):  
Machine Learning ◽  
Random Forest ◽  
Gas Turbine ◽  
Flow Rate ◽  
Turbine Blades ◽  
Operating Conditions ◽  
Classification Model ◽  
Inlet Temperature ◽  
Turbine Inlet Temperature ◽  
Gas Turbine Blades

Abstract Gas turbine blades and vanes face very severe operating conditions - high temperature and pressure which necessitates the creation of complex cooling and component designs, resulting in high computational cost. The ability to predict cyclic failure in these components is therefore a critical activity that has been historically performed using 3D commercial finite element (FE) codes for baseload conditions. However, these codes take substantial time and resources which restricts their application in failure prediction at variable operating conditions. Newer data-driven techniques such as machine learning (ML) provide a valuable tool that can be utilized to predict the occurrence of cyclic failure for these conditions with minimal time and resource requirement. In this paper, a machine learning based surrogate model is developed to predict the cyclic failure of a radially cooled turbine blade. The features used as input to machine learning model are turbine inlet temperature, coolant inlet temperature, hot gas mass flow rate, cooling air mass flow rate and blade materials. The output for the model is a binary variable depicting the incident of component failure. 70% of the FE data points are used to train the ML model while the remaining are used for testing. A comparative study between Logistic Regression, Random Forest, K-nearest neighbor, and Support Vector Machine (SVM) was performed to select the most accurate algorithm for the classification model. Finally, the results show that the Random Forest and SVM algorithms predicts failure with the highest f-1 score of 0.92. The model also demonstrates that Turbine Inlet temperature has the highest importance amongst the input features followed by blade material. Additionally, this methodology offers a tremendous advantage for failure prediction by reducing analysis time from multiple hours to a few seconds, rendering this technique especially beneficial for time sensitive business decisions in the gas turbine industry.

scholarly journals Detection and Severity Evaluation of Combined Rail Defects Using Deep Learning

Vibration ◽  
2021 ◽  
Vol 4 (2) ◽  
pp. 341-356
Author(s):  
Jessada Sresakoolchai ◽  
Sakdirat Kaewunruen
Keyword(s):  
Neural Network ◽  
Machine Learning ◽  
Deep Learning ◽  
Mean Absolute Error ◽  
Absolute Error ◽  
Machine Learning Techniques ◽  
Rolling Stock ◽  
Raw Data ◽  
Learning Techniques ◽  
Combined Defects

Various techniques have been developed to detect railway defects. One of the popular techniques is machine learning. This unprecedented study applies deep learning, which is a branch of machine learning techniques, to detect and evaluate the severity of rail combined defects. The combined defects in the study are settlement and dipped joint. Features used to detect and evaluate the severity of combined defects are axle box accelerations simulated using a verified rolling stock dynamic behavior simulation called D-Track. A total of 1650 simulations are run to generate numerical data. Deep learning techniques used in the study are deep neural network (DNN), convolutional neural network (CNN), and recurrent neural network (RNN). Simulated data are used in two ways: simplified data and raw data. Simplified data are used to develop the DNN model, while raw data are used to develop the CNN and RNN model. For simplified data, features are extracted from raw data, which are the weight of rolling stock, the speed of rolling stock, and three peak and bottom accelerations from two wheels of rolling stock. In total, there are 14 features used as simplified data for developing the DNN model. For raw data, time-domain accelerations are used directly to develop the CNN and RNN models without processing and data extraction. Hyperparameter tuning is performed to ensure that the performance of each model is optimized. Grid search is used for performing hyperparameter tuning. To detect the combined defects, the study proposes two approaches. The first approach uses one model to detect settlement and dipped joint, and the second approach uses two models to detect settlement and dipped joint separately. The results show that the CNN models of both approaches provide the same accuracy of 99%, so one model is good enough to detect settlement and dipped joint. To evaluate the severity of the combined defects, the study applies classification and regression concepts. Classification is used to evaluate the severity by categorizing defects into light, medium, and severe classes, and regression is used to estimate the size of defects. From the study, the CNN model is suitable for evaluating dipped joint severity with an accuracy of 84% and mean absolute error (MAE) of 1.25 mm, and the RNN model is suitable for evaluating settlement severity with an accuracy of 99% and mean absolute error (MAE) of 1.58 mm.

Development and Fabrication of Silicon Nitride Components for Ceramic Gas Turbine

1997 ◽  
Author(s):  
Keisuke Makino ◽  
Ken-Ichi Mizuno ◽  
Toru Shimamori
Keyword(s):  
High Temperature ◽  
Silicon Nitride ◽  
Gas Turbine ◽  
Gas Turbines ◽  
Turbine Blades ◽  
High Strength ◽  
Inlet Temperature ◽  
Turbine Inlet Temperature ◽  
Spark Plug ◽  
Power Turbine

NGK Spark Plug Co., Ltd. has been developing various silicon nitride materials, and the technology for fabricating components for ceramic gas turbines (CGT) using theses materials. We are supplying silicon nitride material components for the project to develop 300 kW class CGT for co-generation in Japan. EC-152 was developed for components that require high strength at high temperature, such as turbine blades and turbine nozzles. In order to adapt the increasing of the turbine inlet temperature (TIT) up to 1,350 °C in accordance with the project goals, we developed two silicon nitride materials with further unproved properties: ST-1 and ST-2. ST-1 has a higher strength than EC-152 and is suitable for first stage turbine blades and power turbine blades. ST-2 has higher oxidation resistance than EC-152 and is suitable for power turbine nozzles. In this paper, we report on the properties of these materials, and present the results of evaluations of these materials when they are actually used for CGT components such as first stage turbine blades and power turbine nozzles.

scholarly journals Analysis of the energetic/environmental performances of gas turbine plant: Effect of thermal barrier coatings and mass of cooling air

2009 ◽  
Vol 13 (1) ◽  
pp. 147-164 ◽  
Author(s):  
Ion Ion ◽  
Anibal Portinha ◽  
Jorge Martins ◽  
Vasco Teixeira ◽  
Joaquim Carneiro
Keyword(s):  
Thermal Conductivity ◽  
Gas Turbine ◽  
Thermal Barrier Coatings ◽  
Turbine Blades ◽  
Heat Transfer Analysis ◽  
Thermal Barrier ◽  
Inlet Temperature ◽  
Barrier Coatings ◽  
Cooling Air Flow ◽  
Cooling Air

Zirconia stabilized with 8 wt.% Y2O3 is the most common material to be applied in thermal barrier coatings owing to its excellent properties: low thermal conductivity, high toughness and thermal expansion coefficient as ceramic material. Calculation has been made to evaluate the gains of thermal barrier coatings applied on gas turbine blades. The study considers a top ceramic coating Zirconia stabilized with 8 wt.% Y2O3 on a NiCoCrAlY bond coat and Inconel 738LC as substrate. For different thickness and different cooling air flow rates, a thermodynamic analysis has been performed and pollutants emissions (CO, NOx) have been estimated to analyze the effect of rising the gas inlet temperature. The effect of thickness and thermal conductivity of top coating and the mass flow rate of cooling air have been analyzed. The model for heat transfer analysis gives the temperature reduction through the wall blade for the considered conditions and the results presented in this contribution are restricted to a two considered limits: (1) maximum allowable temperature for top layer (1200?C) and (2) for blade material (1000?C). The model can be used to analyze other materials that support higher temperatures helping in the development of new materials for thermal barrier coatings.

Machine learning and Grad-Cam based vascular aging assessment using photoplethysmogram (Preprint)

2021 ◽  
Author(s):  
Hangsik Shin
Keyword(s):  
Machine Learning ◽  
Correlation Coefficient ◽  
Age Estimation ◽  
Mean Squared Error ◽  
Mean Absolute Error ◽  
Absolute Error ◽  
Coefficient Of Determination ◽  
Vascular Aging ◽  
Squared Error ◽  
Vascular Age

BACKGROUND Arterial stiffness due to vascular aging is a major indicator for evaluating cardiovascular risk. OBJECTIVE In this study, we propose a method of estimating age by applying machine learning to photoplethysmogram for non-invasive vascular age assessment. METHODS The machine learning-based age estimation model that consists of three convolutional layers and two-layer fully connected layers, was developed using segmented photoplethysmogram by pulse from a total of 752 adults aged 19–87 years. The performance of the developed model was quantitatively evaluated using mean absolute error, root-mean-squared-error, Pearson’s correlation coefficient, coefficient of determination. The Grad-Cam was used to explain the contribution of photoplethysmogram waveform characteristic in vascular age estimation. RESULTS Mean absolute error of 8.03, root mean squared error of 9.96, 0.62 of correlation coefficient, and 0.38 of coefficient of determination were shown through 10-fold cross validation. Grad-Cam, used to determine the weight that the input signal contributes to the result, confirmed that the contribution to the age estimation of the photoplethysmogram segment was high around the systolic peak. CONCLUSIONS The machine learning-based vascular aging analysis method using the PPG waveform showed comparable or superior performance compared to previous studies without complex feature detection in evaluating vascular aging. CLINICALTRIAL 2015-0104

scholarly journals P112 Prediction of free Ceftriaxone concentration in children: disease and maturation do matter

2019 ◽  
Vol 104 (6) ◽  
pp. e64.2-e64
Author(s):  
H-Y Shi ◽  
X Huang ◽  
Q Li ◽  
Wu Y-E ◽  
MW Khan ◽  
...  
Keyword(s):  
Linear Regression ◽  
Mean Absolute Error ◽  
Predictive Ability ◽  
Predictive Performance ◽  
Absolute Error ◽  
Regression Equations ◽  
Free Concentration ◽  
In Vivo Binding ◽  
Mean Prediction Error

BackgroundTo evaluate the predictive ability of the existing formula to measure free ceftriaxone levels in children, and optimize the formula by adding disease and maturation factors.MethodsFifty children receiving ceftriaxone were evaluated, and the predictive performance of the different equations were assessed by mean absolute error (MAE), mean prediction error (MPE) and linear regression of predicted vs. actual free levels.ResultsThe average free ceftriaxone concentration was 2.11 ± 9.51µg/ml. The predicted free concentration was 1.15 ± 4.39µg/ml with the in vivo binding equation, which increased to 1.58 ± 7.73µg/ml and 2.01 ± 9.53µg/ml when adjusted for age (disease adapted equation), and age and albumin (disease-maturation equation) respectively. The average MAE values were 0.48 (in vivo banding equation), 0.34 (disease adapted equation) and 0.41 (disease maturation equation). The average MPE values were -0.41 (in vivo binding equation), 0.14 (disease adapted equation) and 0.09 (disease maturation equation). The respective linear regression equations and coefficients were y=1.8647x+1.0731(R2=0.7398), y=1.1455x+0.8414(R2=0.8674), and y=0.9664x(R2=0.8641) for the in vivo binding, disease adapted and disease maturation equations respectively.ConclusionCompared to the in vivo binding equation, the disease adapted and disease maturation equations showed lower MAE and MPE values, and the latter showed the lowest MPE value. In addition, the slope of the disease maturation equation was closer to 1 compared to the other two. Therefore, the optimized disease maturation equation should be used to measure free ceftriaxone levels in children.Disclosure(s)Nothing to disclose.

scholarly journals Measurement Method for Evaluating the Lockdown Policies during the COVID-19 Pandemic

2020 ◽  
Vol 17 (15) ◽  
pp. 5574 ◽  
Author(s):  
Mohammed Al Zobbi ◽  
Belal Alsinglawi ◽  
Omar Mubin ◽  
Fady Alnajjar
Keyword(s):  
Machine Learning ◽  
Infection Rate ◽  
Global Economy ◽  
Data Analytics ◽  
Measurement Method ◽  
Mean Absolute Error ◽  
Absolute Error ◽  
Decision Makers ◽  
Machine Learning Classification ◽  
Data Analyses

Coronavirus Disease 2019 (COVID-19) has affected day to day life and slowed down the global economy. Most countries are enforcing strict quarantine to control the havoc of this highly contagious disease. Since the outbreak of COVID-19, many data analyses have been done to provide close support to decision-makers. We propose a method comprising data analytics and machine learning classification for evaluating the effectiveness of lockdown regulations. Lockdown regulations should be reviewed on a regular basis by governments, to enable reasonable control over the outbreak. The model aims to measure the efficiency of lockdown procedures for various countries. The model shows a direct correlation between lockdown procedures and the infection rate. Lockdown efficiency is measured by finding a correlation coefficient between lockdown attributes and the infection rate. The lockdown attributes include retail and recreation, grocery and pharmacy, parks, transit stations, workplaces, residential, and schools. Our results show that combining all the independent attributes in our study resulted in a higher correlation (0.68) to the dependent value Interquartile 3 (Q3). Mean Absolute Error (MAE) was found to be the least value when combining all attributes.

scholarly journals Jump or kink: on super-efficiency in segmented linear regression breakpoint estimation

Biometrika ◽  
2020 ◽  
Author(s):  
Yining Chen
Keyword(s):  
Structural Change ◽  
Linear Regression ◽  
Convergence Rate ◽  
Pointwise Convergence ◽  
Mean Absolute Error ◽  
Absolute Error ◽  
The Other ◽  
Fixed Parameter ◽  
The Mean ◽  
Potential Remedy

Summary We consider the problem of segmented linear regression with a single breakpoint, with the focus on estimating the location of the breakpoint. If $n$ is the sample size, we show that the global minimax convergence rate for this problem in terms of the mean absolute error is $O(n^{-1/3})$. On the other hand, we demonstrate the construction of a super-efficient estimator that achieves the pointwise convergence rate of either $O(n^{-1})$ or $O(n^{-1/2})$ for every fixed parameter value, depending on whether the structural change is a jump or a kink. The implications of this example and a potential remedy are discussed.

Development of the Advanced Closed-Cycle Gas Turbine System

2001 ◽  
Author(s):  
Miki Koyama ◽  
Toshio Mimaki
Keyword(s):  
Film Cooling ◽  
Cooling System ◽  
Early Stage ◽  
Turbine Blades ◽  
Thermal Conditions ◽  
Inlet Temperature ◽  
Internal Cooling ◽  
Turbine Inlet Temperature ◽  
Turbine Blade Cooling ◽  
Turbine Inlet

This aims to put the fruits of the R&D; “The Hydrogen Combustion Turbine” in WE-NET Phase I Program(1993-1998) to practical use at an early stage. The topping regenerating cycle was selected as the optimum cycle, with energy efficiency expected to be more than 60%(HHV) under the conditions of the turbine inlet temperature of 1973K(1700°C) and the pressure of 4.8MPa,in it. • As the turbine inlet temperature and pressure increase, issues to be resolved include the amount of NOx emissions and the durability of super alloys for turbine blades under such thermal conditions. In this respect, the development of the highly efficient methane-oxygen combustion technology, the turbine blade cooling technology, and the ultrahigh-temperature materials including thermal barrier coatings is being carried out. • In 1999, the results made it clear that there are little error among the three analytic programs used to verify the system efficiency, it was verified that the burning rate was going to arrive at over 98% from the methane-oxygen combustion test (under the atmospheric pressure). And the type of vane “Film cooling plus recycle type with internal cooling system” was selected as the most suitable vane.

Sign in / Sign up

Export Citation Format

Share Document