scholarly journals Condition-Based Maintenance for Normal Behaviour Characterisation of Railway Car-Body Acceleration Applying Neural Networks

2021 ◽  
Vol 13 (21) ◽  
pp. 12265
Author(s):  
Pablo Garrido Martínez-Llop ◽  
Juan de Dios Sanz Bobi ◽  
Álvaro Solano Jiménez ◽  
Jorge Gutiérrez Sánchez
Keyword(s):  
Neural Networks ◽  
Dynamic Behaviour ◽  
Condition Based Maintenance ◽  
Vehicle Body ◽  
Average Error ◽  
Normal Behaviour ◽  
Car Body ◽  
Successful Model ◽  
Optimized Model ◽  
Potential Failure

Recently, passenger comfort and user experience are becoming increasingly relevant for the railway operators and, therefore, for railway manufacturers as well. The main reason for this to happen is that comfort is a clear differential value considered by passengers as final customers. Passengers’ comfort is directly related to the accelerations received through the car-body of the train. For this reason, suspension and damping components must be maintained in perfect condition, assuring high levels of comfort quality. An early detection of any potential failure in these systems derives in a better maintenance inspections’ planification and in a more sustainable approach to the whole train maintenance strategy. In this paper, an optimized model based on neural networks is trained in order to predict lateral car-body accelerations. Comparing these predictions to the values measured on the train, a normal characterisation of the lateral dynamic behaviour can be determined. Any deviation from this normal characterisation will imply a comfort loss or a potential degradation of the suspension and damping components. This model has been trained with a dataset from a specific train unit, containing variables recorded every second during the year 2017, including lateral and vertical car-body accelerations, among others. A minimum average error of 0.034 m/s2 is obtained in the prediction of lateral car-body accelerations. This means that the average error is approximately 2.27% of the typical maximum estimated values for accelerations in vehicle body reflected in the EN14363 for the passenger coaches (1.5 m/s2). Thus, a successful model is achieved. In addition, the model is evaluated based on a real situation in which a passenger noticed a lack of comfort, achieving excellent results in the detection of atypical accelerations. Therefore, as it is possible to measure acceleration deviations from the standard behaviour causing lack of comfort in passengers, an alert can be sent to the operator or the maintainer for a non-programmed intervention at depot (predictive maintenance) or on board (prescriptive maintenance). As a result, a condition-based maintenance (CBM) methodology is proposed to avoid comfort degradation that could end in passenger complaints or speed limitation due to safety reasons for excessive acceleration. This methodology highlights a sustainable maintenance concept and an energy efficiency strategy.

Multi-Objective Crashworthiness Optimization of Composite Hat-Shape Energy Absorber Using GMDH-Type Neural Networks and Genetic Algorithms

2010 ◽  
Author(s):  
Naser Tavassoli ◽  
Bahram Notghi ◽  
Abolfazl Darvizeh ◽  
Mansour Darvizeh
Keyword(s):  
Neural Networks ◽  
Energy Absorption ◽  
Failure Criterion ◽  
Design Procedure ◽  
Vehicle Body ◽  
Group Method ◽  
Nsga Ii ◽  
Car Body ◽  
Multi Objective ◽  
Conflicting Objectives

Reducing the weight of car body and increasing the crashworthiness capability of car body are two important objectives of car design. In this paper, a multi-objective optimization for optimal composite hat-shape energy absorption system is presented At the first, the behaviors of the hat shape under impact, as simplified model of side member of a vehicle body, are studied by the finite element method using commercial software ABAQUS. Two meta-models based on the evolved group method of data handling (GMDH) type neural networks are then achieved for modeling of both the absorbed energy (E) and the Tsai-Hill Failure Criterion (TS) with respect to geometrical design variables using those training and testing data obtained models. The obtained polynomial neural meta-models are finally used in a multi-objective optimum design procedure using NSGA-II with a new diversity preserving mechanism for Pareto based optimization of hat-shape. Two conflicting objectives such as maximizing the energy absorption capability (E), minimizing the Tsai-Hill Failure Criterion are considered in this work.

scholarly journals Predicting Sooting Propensity of Oxygenated Fuels Using Artificial Neural Networks

Processes ◽  
2021 ◽  
Vol 9 (6) ◽  
pp. 1070
Author(s):  
Abdul Gani Abdul Jameel
Keyword(s):  
Neural Networks ◽  
Artificial Neural Networks ◽  
Functional Groups ◽  
Average Error ◽  
Ann Model ◽  
Oh Groups ◽  
Unseen Data ◽  
Learning Capabilities ◽  
Artificial Neural ◽  
Oxygenated Fuels

The self-learning capabilities of artificial neural networks (ANNs) from large datasets have led to their deployment in the prediction of various physical and chemical phenomena. In the present work, an ANN model was developed to predict the yield sooting index (YSI) of oxygenated fuels using the functional group approach. A total of 265 pure compounds comprising six chemical classes, namely paraffins (n and iso), olefins, naphthenes, aromatics, alcohols, and ethers, were dis-assembled into eight constituent functional groups, namely paraffinic CH3 groups, paraffinic CH2 groups, paraffinic CH groups, olefinic –CH=CH2 groups, naphthenic CH-CH2 groups, aromatic C-CH groups, alcoholic OH groups, and ether O groups. These functional groups, in addition to molecular weight and branching index, were used as inputs to develop the ANN model. A neural network with two hidden layers was used to train the model using the Levenberg–Marquardt (ML) training algorithm. The developed model was tested with 15% of the random unseen data points. A regression coefficient (R2) of 0.99 was obtained when the experimental values were compared with the predicted YSI values from the test set. An average error of 3.4% was obtained, which is less than the experimental uncertainty associated with most reported YSI measurements. The developed model can be used for YSI prediction of hydrocarbon fuels containing alcohol and ether-based oxygenates as additives with a high degree of accuracy.

scholarly journals Use of Nondestructive Testing of Ultrasound and Artificial Neural Networks to Estimate Compressive Strength of Concrete

Buildings ◽  
2021 ◽  
Vol 11 (2) ◽  
pp. 44
Author(s):  
Fernando A. N. Silva ◽  
João M. P. Q. Delgado ◽  
Rosely S. Cavalcanti ◽  
António C. Azevedo ◽  
Ana S. Guimarães ◽  
...  
Keyword(s):  
Neural Networks ◽  
Compressive Strength ◽  
Artificial Neural Networks ◽  
Cross Sections ◽  
Influential Factors ◽  
Average Error ◽  
Ann Model ◽  
Cement Ratio ◽  
Water Cement Ratio ◽  
Artificial Neural

The work presents the results of an experimental campaign carried out on concrete elements in order to investigate the potential of using artificial neural networks (ANNs) to estimate the compressive strength based on relevant parameters, such as the water–cement ratio, aggregate–cement ratio, age of testing, and percentage cement/metakaolin ratios (5% and 10%). We prepared 162 cylindrical concrete specimens with dimensions of 10 cm in diameter and 20 cm in height and 27 prismatic specimens with cross sections measuring 25 and 50 cm in length, with 9 different concrete mixture proportions. A longitudinal transducer with a frequency of 54 kHz was used to measure the ultrasonic velocities. An ANN model was developed, different ANN configurations were tested and compared to identify the best ANN model. Using this model, it was possible to assess the contribution of each input variable to the compressive strength of the tested concretes. The results indicate an excellent performance of the ANN model developed to predict compressive strength from the input parameters studied, with an average error less than 5%. Together, the water–cement ratio and the percentage of metakaolin were shown to be the most influential factors for the compressive strength value predicted by the developed ANN model.

scholarly journals How to Predict Seawater Temperature for Sustainable Marine Aquaculture (Student Abstract)

2020 ◽  
Vol 34 (10) ◽  
pp. 13887-13888
Author(s):  
Masahito Okuno ◽  
Takanobu Otsuka
Keyword(s):  
Neural Networks ◽  
Seawater Temperature ◽  
Prediction Method ◽  
Accurate Prediction ◽  
Average Error ◽  
Marine Aquaculture ◽  
Global Demand ◽  
Environment Control ◽  
Marine Products ◽  
Using Data

The increasing global demand for marine products has turned attention to marine aquaculture. In marine aquaculture, appropriate environment control is important for a stable supply. The influence of seawater temperature on this environment is significant and accurate prediction is therefore required. In this paper, we propose and describe the implementation of a seawater prediction method using data acquired from real aquaculture areas and neural networks. Our evaluation experiment showed that hourly next-day prediction has an average error of about 0.2 to 0.4 ◦C and daily prediction of up to one week has an average error of about 0.2 to 0.5 ◦C. This is enough to meet actual worker need, which is within 1 ◦C error, thus confirming that our seawater prediction method is suitable for actual sites.

A comprehensive method for studying the degree of thermal damage to steel elements of vehicles using field methods

2021 ◽  
Vol 93 ◽  
pp. 137-151
Author(s):  
Yu. D. Motorygin ◽  
◽  
G. A. Sikorova ◽  
Keyword(s):  
High Temperatures ◽  
Thermal Damage ◽  
Vehicle Body ◽  
Fuel Load ◽  
Complex Method ◽  
Main Task ◽  
Field Methods ◽  
Car Body ◽  
Comprehensive Method ◽  
Independent Field

Introduction. Every year, the number of registered motor vehicles in Russia increases, with an increase of 30 % over the past 10 years. The desire to extend the service life of existing vehicles leads to an increase in breakdowns that lead to fire. The task of identifying places where there was an emergency situation that led to a fire is difficult, especially in the case of a complete burnout of the fuel load of the car. Goals and objectives. The purpose of the work is to create a methodology by which it is possible to conduct studies of cold-formed steel products of the vehicle body to determine the zone of the greatest thermal damage. Each method determines one of the characteristics that changes when the steel is exposed to high temperatures. The main task of the work is to test the possibility of applying these methods in a certain sequence to determine the zone that has undergone more annealing. Research methods. The complex method includes induction thickness measurement, which allows measuring the thickness of a layer of high-temperature oxide; a method for measuring the coercive force or demagnetization currents, as well as a method for measuring microhardness. The latter method will determine the decrease in the hardness of the vehicle body after exposure to high temperatures. Results and its discussion. The obtained data on the degree of thermal damage to cold-deformed steel products of the car body by three independent field methods will increase the reliability of the combustion start zone, which in turn will determine the search zone for the cause of ignition. Conclusion. Determining the specific cause of the fire during the maintenance, repair and operation of vehicles will allow you to exclude similar emergency modes in the future. The introduction of a comprehensive method for investigating the condition of an object after a fire will significantly increase the level of reliability of the results obtained when searching for a breakdown that led to a fire. Keywords: fire source, car body, thermal changes, field methods, complex technique, magnetic properties, scale thickness, microhardness

scholarly journals Latent Multi-Task Architecture Learning

2019 ◽  
Vol 33 ◽  
pp. 4822-4829 ◽  
Author(s):  
Sebastian Ruder ◽  
Joachim Bingel ◽  
Isabelle Augenstein ◽  
Anders Søgaard
Keyword(s):  
Neural Networks ◽  
Recent Work ◽  
Deep Neural Networks ◽  
Synthetic Data ◽  
Approaches To Learning ◽  
Average Error ◽  
Task Learning ◽  
Task Architecture ◽  
Parameter Sharing

Multi-task learning (MTL) allows deep neural networks to learn from related tasks by sharing parameters with other networks. In practice, however, MTL involves searching an enormous space of possible parameter sharing architectures to find (a) the layers or subspaces that benefit from sharing, (b) the appropriate amount of sharing, and (c) the appropriate relative weights of the different task losses. Recent work has addressed each of the above problems in isolation. In this work we present an approach that learns a latent multi-task architecture that jointly addresses (a)–(c). We present experiments on synthetic data and data from OntoNotes 5.0, including four different tasks and seven different domains. Our extension consistently outperforms previous approaches to learning latent architectures for multi-task problems and achieves up to 15% average error reductions over common approaches to MTL.

scholarly journals Chestnut (Castanea sativa Mill.) cultivar classification: an artificial neural network approach

2020 ◽  
Vol 48 (1) ◽  
pp. 366-377 ◽  
Author(s):  
Yeşim Benal ÖZTEKİN ◽  
Alper TANER ◽  
Hüseyin DURAN
Keyword(s):  
Neural Networks ◽  
Transfer Functions ◽  
Geometric Mean ◽  
Back Propagation ◽  
Castanea Sativa ◽  
Average Error ◽  
Ann Model ◽  
Neural Network Approach ◽  
Castanea Sativa Mill ◽  
Artificial Neural

The present study investigated the possible use of artificial neural networks (ANN) to classify five chestnut (Castanea sativa Mill.) varieties. For chestnut classification, back-propagation neural networks were framed on the basis of physical and mechanical parameters. Seven physical and mechanical characteristics (geometric mean diameter, sphericity, volume of nut, surface area, shell thickness, shearing force and strength) of chestnut were determined. It was found that these characteristics were statistically different and could be used in the classification of species. In the developed ANN model, the design of the network is 7-(5-6)-1 and it consists of 7 input, 2 hidden and 1 output layers. Tansig transfer functions were used in both hidden layers, while linear transfer functions were used in the output layer. In ANN model, R2 value was obtained as 0.99999 and RMSE value was obtained as 0.000083 for training. For testing, R2 value was found as 0.99999 and RMSE value was found as 0.00031. In the approximation of values obtained with ANN model to the values measured, average error was found as 0.011%. It was found that the results found with ANN model were very compatible with the measured data. It was found that the ANN model obtained can classify chestnut varieties in a fast and reliable way.

scholarly journals Optimizing the Geometry of Flexure System Topologies Using the Boundary Learning Optimization Tool

2018 ◽  
Vol 2018 ◽  
pp. 1-14 ◽  
Author(s):  
Ali Hatamizadeh ◽  
Yuanping Song ◽  
Jonathan B. Hopkins
Keyword(s):  
Neural Networks ◽  
Artificial Neural Networks ◽  
Optimization Algorithm ◽  
Geometric Parameters ◽  
Average Error ◽  
Computational Tool ◽  
Model Training ◽  
Constraint Topologies ◽  
Synthesis Approach

We introduce a new computational tool called the Boundary Learning Optimization Tool (BLOT) that identifies the boundaries of the performance capabilities achieved by general flexure system topologies if their geometric parameters are allowed to vary from their smallest allowable feature sizes to their largest geometrically compatible feature sizes for given constituent materials. The boundaries generated by the BLOT fully define the design spaces of flexure systems and allow designers to visually identify which geometric versions of their synthesized topologies best achieve desired combinations of performance capabilities. The BLOT was created as a complementary tool to the freedom and constraint topologies (FACT) synthesis approach in that the BLOT is intended to optimize the geometry of the flexure topologies synthesized using the FACT approach. The BLOT trains artificial neural networks to create models of parameterized flexure topologies using numerically generated performance solutions from different design instantiations of those topologies. These models are then used by an optimization algorithm to plot the desired topology’s performance boundary. The model-training and boundary-plotting processes iterate using additional numerically generated solutions from each updated boundary generated until the final boundary is guaranteed to be accurate within any average error set by the user. A FACT-synthesized flexure topology is optimized using the BLOT as a simple case study.

Sign in / Sign up

Export Citation Format

Share Document