ANN modelling of CO2 refrigerant cooling system COP in a smart warehouse
مدل سازی ANN سیستم خنک کننده کولر CO2 COP در یک انبار هوشمند-2020
Industrial cooling systems consume large quantities of energy with highly variable power demand. To reduce environmental impact and overall energy consumption, and to stabilize the power requirements, it is recommended to recover surplus heat, store energy, and integrate renewable energy production. To control these operations continuously in a complex energy system, an intelligent energy management system can be employed using operational data and machine learning. In this work, we have developed an artificial neural network based technique for modelling operational CO2 refrigerant based industrial cooling systems for embedding in an overall energy management system. The operating temperature and pressure measurements, as well as the operating frequency of compressors, are used in developing operational model of the cooling system, which outputs electrical consumption and refrigerant mass flow without the need for additional physical measurements. The presented model is superior to a generalized theoretical model, as it learns from data that includes individual compressor type characteristics. The results show that the presented approach is relatively precise with a Mean Average Percentage Error (MAPE) as low as 5%, using low resolution and asynchronous data from a case study system. The developed model is also tested in a laboratory setting, where MAPE is shown to be as low as 1.8%.
Keywords: Industrial cooling systems | Carbon dioxide refrigerant | Artificial neural networks | Coefficient of performance | Energy storage | Smart warehouse
AI-based Framework for Deep Learning Applications in Grinding
چارچوبی مبتنی بر هوش مصنوعی برای کاربردهای یادگیری عمیق در شبکه سازی-2020
Rejection costs for a finish-machined gearwheel with grinding burn can rise to the order of 10,000 euros each. A reduction in costs by reducing rejection rate by only 5-10 pieces per year already amortizes costs for data-acquisition hardware for online process monitoring. The grinding wheel wear, one of the major influencing factors responsible for the grinding burn, depends on a large number of influencing variables like cooling lubricant, feed rate, circumferential wheel speed and wheel topography. In the past, machine learning algorithms such as Support Vector Machines (SVM), Hidden Markov Models (HMM) and Artificial Neural Networks (ANN) have proven effective for the predictive analysis of process quality. In addition to predictive analysis, AI-based applications for process control may raise the resilience of machining processes. Using machine learning methods may also lead to a heavy reduction of cost amassed due to a physical inspection of each workpiece. With this contribution, information from previous works is leveraged and an AI-based framework for adaptive process control of a cylindrical grinding process is introduced. For the development of such a framework, three research objectives have been derived: First, the dynamic wheel wear needs to be modelled and measured, because of its strong impact on the resulting workpiece quality. Second, models to predict the quality features of the produced workpieces depending on process setup parameters and materials used have to be established. Here, special focus is set on deriving models that are independent of a specific wheel-workpiece-pair. The opportunity to use such a model in a variety of grinding configurations gives the production line consistent process support. Third, the resilience of analytical models regarding graceful degradation of sensors needs to be tackled, since the stability of such systems has to be guaranteed to be used in productive environments. Process resilience against human errors and sensor failures leads to a minimization of rejection costs in production. To do so, a framework is presented, where virtual sensors, upon the failure or detection of an erroneous signal from physical sensors, will be activated and provide signals to the downstream smart systems until the process is completed or the physical sensor is changed.
Keywords: Cylindrical Grinding | Wheel Wear | Virtual Sensors | Process Resilience | Artificial Intelligence
Modeling of forward osmosis process using artificial neural networks (ANN) to predict the permeate flux
مدل سازی فرآیند اسمزوز رو به جلو با استفاده از شبکه های عصبی مصنوعی (ANN) برای پیش بینی شار نفوذ-2020
Artificial neural networks (ANN) are black box models that are becoming more popular than transport-based models due to their high accuracy and less computational time in predictions. The literature shows a lack of ANN models to evaluate the forward osmosis (FO) process performance. Therefore, in this study, a multi-layered neural network model is developed to predict the permeate flux in forward osmosis. The developed model is tested for its generalization capability by including lab-scale experimental data from several published studies. Nine input variables are considered including membrane type, the orientation of membrane, molarity of feed solution and draw solution, type of feed solution and draw solution, crossflow velocity of the feed solution, and the draw solution and temperature of the feed solution and the draw solution. The development of optimum network architecture is supported by studying the impact of the number of neurons and hidden layers on the neural network performance. The optimum trained network shows a high R2 value of 97.3% that is the efficiency of the model to predict the targeted output. Furthermore, the validation and generalized prediction capability of the model is tested against untrained published data. The performance of the ANN model is compared with a transport-based model in the literature. A simple machine learning technique such as a multiple linear regression (MLR) model is also applied in a similar manner to be compared with the ANN model. ANN demonstrates its ability to form a complex relationship between inputs and output better than MLR.
Keywords: Artificial neural network | Forward osmosis | Water treatment | Desalination | Machine learning
Fatigue life prediction of metallic materials considering mean stress effects by means of an artificial neural network
پیش بینی طول عمر خستگی مواد فلزی با توجه به میانگین اثرات استرس با استفاده از شبکه عصبی مصنوعی-2020
The mean stress effect plays an important role in the fatigue life predictions, its influence significantly changes high-cycle fatigue behaviour, directly decreasing the fatigue limit with the increase of the mean stress. Fatigue design of structural details and mechanical components must account for mean stress effects in order to guarantee the performance and safety criteria during their foreseen operational life. The purpose of this research work is to develop a new methodology to generate a constant life diagram (CLD) for metallic materials, based on assumptions of Haigh diagram and artificial neural networks, using the probabilistic Stüssi fatigue S-N fields. This proposed methodology can estimate the safety region for high-cycle fatigue regimes as a function of the mean stress and stress amplitude in regions where tensile loading is predominance, using fatigue S-N curves only for two stress R-ratios. In this approach, the experimental fatigue data of the P355NL1 pressure vessel steel available for three stress R-ratios (−1, −0.5, 0), are used. A multilayer perceptron network has been trained with the back-propagation algorithm; its architecture consists of two input neurons (σm, N) and one output neuron (σa). The suggested CLD based on trained artificial neural network algorithm and probabilistic Stüssi fatigue fields applied to dog-bone shaped specimens made of P355NL1 steel showed a good agreement with the high-cycle fatigue experimental data, only using the stress R-ratios equal to 0 and −0.5. Furthermore, a procedure for estimating the fatigue resistance reduction factor, Kf , for the fatigue life prediction of structural details (stress R-ratios equal to 0, 0.15 and 0.3) in extrapolation regions is suggested and used to generate the Kf results for stress R-ratios from −1 to 0.3, based on machine learning artificial neural network algorithm.
Keywords: Fatigue | Artificial neural network | Back-propagation algorithm | Stüssi model | Constant life diagram
Adaptive indirect neural network model for roughness in honing processes
مدل شبکه عصبی غیرمستقیم سازگار برای زبری در فرآیندهای آب بندی-2020
Honing processes provide a crosshatch pattern that allows oil flow, for example in combustion engine cylinders. This paper provides an adaptive neural network model for predicting roughness as a function of process parameters. Input variables are three parameters from the Abbott-Firestone curve, Rk, Rpk and Rvk. Output parameters are grain size, density of abrasive, pressure, linear speed and tangential speed. The model consists of applying a direct and an indirect model consecutively, with one convergence parameter and one error parameter. The indirect model has one network with 48 neurons and the direct model has three networks having 25, 9 and 5 neurons respectively. The adaptive one allows selecting discrete values for some variables like grain size or density.
Keywords: Honing | Surface roughness | Artificial neural networks | Adaptive control
Predicting and explaining corruption across countries: A machine learning approach
پیش بینی و توضیح فساد در سراسر کشور: رویکرد یادگیری ماشینی-2020
In the era of Big Data, Analytics, and Data Science, corruption is still ubiquitous and is perceived as one of the major challenges of modern societies. A large body of academic studies has attempted to identify and explain the potential causes and consequences of corruption, at varying levels of granularity, mostly through theoretical lenses by using correlations and regression-based statistical analyses. The present study approaches the phenomenon from the predictive analytics perspective by employing contemporary machine learning techniques to discover the most important corruption perception predictors based on enriched/enhanced nonlinear models with a high level of predictive accuracy. Specifically, within the multiclass classification modeling setting that is employed herein, the Random Forest (an ensemble-type machine learning algorithm) is found to be the most accurate prediction/classification model, followed by Support Vector Machines and Artificial Neural Networks. From the practical standpoint, the enhanced predictive power of machine learning algorithms coupled with a multi-source database revealed the most relevant corruption-related information, contributing to the related body of knowledge, generating actionable insights for administrator, scholars, citizens, and politicians. The variable importance results indicated that government integrity, property rights, judicial effectiveness, and education index are the most influential factors in defining the corruption level of significance
Keywords: Corruption perception | Machine learning | Predictive modeling | Random forest | Society policies and regulations |Government integrity | Social development
Using of Artificial Neural Networks (ANNs) to predict the thermal conductivity of Zinc Oxide–Silver (50%–50%)/Water hybrid Newtonian nanofluid
استفاده از شبکه های عصبی مصنوعی (ANN) برای پیش بینی هدایت حرارتی روی اکسید-نقره روی (50٪ -50٪) / نانوسیال نیوتنی ترکیبی آب-2020
In this study, after generating experimental data points of Zinc Oxide (ZnO)–Silver (Ag) (50%–50%)/Water nanofluid, an algorithm is proposed to calculate the best neuron number in the Artificial Neural Network (ANN), and the performance and correlation coefficient for ANN has been calculated. Then, using the fitting method, a surface is fitted on the experimental data, and the correlation coefficient and performance of this method have been calculated. Finally, the absolute values of errors in both methods have been compared. It can be seen that the best neuron number in the hidden layer is 7 neurons. We concluded that both methods could predict the behavior of nanofluid, but the fitting method had smaller errors. Also, the ANN method had better ability in predicting the thermal conductivity of nanofluid based on the volume fraction of nanoparticles and temperature. Finally, we found that, in ANN, all outputs, the maximum absolute value of error is 0.0095, and the train performance is 1.6684e-05.
Keywords: Artificial Neural Networks (ANNs) | Thermal conductivity | Hybrid Newtonian nanofluid
A review of learning in biologically plausible spiking neural networks
مروری بر یادگیری در شبکه های عصبی اسپایک بیولوژیکی قابل قبول-2020
Artificial neural networks have been used as a powerful processing tool in various areas such as pattern recognition, control, robotics, and bioinformatics. Their wide applicability has encouraged researchers to improve artificial neural networks by investigating the biological brain. Neurological research has significantly progressed in recent years and continues to reveal new characteristics of biological neurons. New technologies can now capture temporal changes in the internal activity of the brain in more detail and help clarify the relationship between brain activity and the perception of a given stimulus. This new knowledge has led to a new type of artificial neural network, the Spiking Neural Network (SNN), that draws more faithfully on biological properties to provide higher processing abilities. A review of recent developments in learning of spiking neurons is presented in this paper. First the biological background of SNN learning algorithms is reviewed. The important elements of a learning algorithm such as the neuron model, synaptic plasticity, information encoding and SNN topologies are then presented. Then, a critical review of the state-of-the-art learning algorithms for SNNs using single and multiple spikes is presented. Additionally, deep spiking neural networks are reviewed, and challenges and opportunities in the SNN field are discussed.
Keywords: Spiking neural network (SNN) | Learning | Synaptic plasticity
Extreme learning machine for a new hybrid morphological/linear perceptron
دستگاه یادگیری شدید برای مورفولوژی جدید ترکیبی / پرسپترون خطی-2020
Morphological neural networks (MNNs) can be characterized as a class of artificial neural networks that perform an operation of mathematical morphology at every node, possibly followed by the application of an activation function. Morphological perceptrons (MPs) and (gray-scale) morphological associative memories are among the most widely known MNN models. Since their neuronal aggregation functions are not differentiable, classical methods of non-linear optimization can in principle not be directly applied in order to train these networks. The same observation holds true for hybrid morphological/ linear perceptrons and other related models. Circumventing these problems of non-differentiability, this paper introduces an extreme learning machine approach for training a hybrid morphological/linear perceptron, whose morphological components were drawn from previous MP models. We apply the resulting model to a number of well-known classification problems from the literature and compare the performance of our model with the ones of several related models, including some recent MNNs and hybrid morphological/linear neural networks.
Keywords: Mathematical morphology | Lattice computing | Morphological neural networks | Hybrid morphological/linear perceptron | Extreme learning machine | Classification
Neural network aided development of a semi-empirical interatomic potential for titanium
شبکه عصبی به توسعه پتانسیل متقابل نیمه تجربی تیتانیوم کمک کرده است-2020
Artificial neural networks, utilizing machine learning techniques to uncover subtle and complex patterns in big data problems, are able to condense large amounts of computationally expensive density functional theory and ab initio results into classical force field potentials. However, in order to produce a computationally efficient network, with minimal network architecture, a structural fingerprint whose components are highly correlated to the per atom energy is necessary. In this paper, we demonstrate the effectiveness a structural fingerprint motivated by the highly successful MEAM formalism by creating an artificial neural network containing a single hidden layer of 20 nodes which provides a semi-empirical force field potential for elemental titanium. This potential is suitable for dynamic calculations of α-, β-, and ω-titanium at a variety of temperatures. This potential is able to achieve a number of results in agreement with DFT calculations which surpass classical potential formalisms with comparable computational performance.
Keywords: Machine learning | Neural networks | Titanium