Linear Output Neurons Research Articles

Feed-Forward Neural Network-Based Prediction of Flutter Speed of 3 DOF 2D Wing Model

IntroductionThe application of artificial intelligence, particularly neural network models, has expanded throughout various fields of academic and industrial studies, recently. This research focuses to investigate aeroelastic phenomena and predict flutter speeds without expensive computational and experimental methods utilizing these algorithms, specifically the artificial neural network (ANN).MethodsFundamentally, the approach involves an ANN algorithm to navigate the complexities of aeroelastic systems, achieved by layering multiple ANNs. The network's neurons effectively interpret diverse flight data by this structure. The study also incorporates state-space representation and Theodorsen's unsteady aerodynamic theory to create a comprehensive dataset on aeroelastic flutter speed across different wing parameters. In order to forecast the flutter speeds, a feed-forward neural network (FFNN) model, which is an approach of ANN method, containing sigmoid hidden neurons and a linear output neuron has been proposed for the conditions above.ResultsThe proposed model accomplished a regression value of 0.92 for flutter speeds according to the experimental findings. As presented, the suggested FFNN model is successful and can be employed to predict the flutter speed estimation. The findings of the FFNN structure are validated with the previous work for aeroelastic flutter speed analysis in the literature.DiscussionFinally, the findings indicate that the FFNN model formulated in this study is highly accurate in predicting flutter speeds with the accordance of numerical results of aeroelastic structure modeled with unsteady aerodynamic theory. In addition, the correctness of the predicted flutter speeds demonstrates that analyzing without expensive tests and high computational costs is in the near future.

Turbofan engine health status prediction with artificial neural network

The main purpose of this study is to present the concept of the aircraft turbofan engine health status prediction with artificial neural network augmentation process. The main idea of engine health status prediction is based on the engine health status parameter broadly used in the aviation industry as well as propulsion technology being the performance and safety margin. As a result of research engine health status index is calculated in order to determine the engine degradation level. The calculated parameter is then used as a response parameter for the machine learning algorithm. The case study is based on the artificial neural network which was two-layer feedforward network with sigmoid hidden neurons and linear output neurons. Network performance is evaluated using mean squared error and regression analysis. The final results are analyzed using visualization plots such as regression fit plot and histogram of errors. The greatest achievement of this elaboration is the presentation of how the entire process of engine status prediction might be augmented with the use of an artificial neural network. What is the greatest scientific contribution of the article is the fact that there are no scientific studies available, which are based on the engine real-life operating data.

Investigation of Dyeing Characteristics of Merino Wool Fiber Dyed with Sustainable Natural Dye Extracted from Aesculus hippocastanum

Recently there has been growing interest in dyeing biomaterials using natural sustainable plant extracts classified as eco-friendly. The microwave-assisted method provides fast heating and energy efficiency, more homogenous heat distribution in dyeing baths, less use of chemicals, and less heat loss, resulting in this method being greener—more sustainable and ecological. Artificial neural networks (ANNs) are used to predict the dyeing properties of fibers, which are often complex and dependent on multiple variables. This saves time and reduces costs compared to trial-and-error methods. This study presents the green dyeing of merino wool fiber with natural dye extracted from Aesculus hippocastanum (horse chestnut) shells using the microwave-assisted method. Before dyeing, the merino wool fiber underwent a pre-mordanted process with aluminum potassium sulfate with different concentrations using the microwave-assisted method. Spectrophotometric analysis of the light, washing, and rubbing fastness of the dyed merino wool fibers was performed. The color strength, light, washing, and rubbing fastness of the dyed merino wool fiber were developed using the pre-mordanting process. After the pre-mordanting process, the light fastness of the samples improved from 1–2 to 3, the color change increased from 2 to 3–4, and the rubbing fastness developed from 2–3 to 4 according to mordant concentration, mordanting time, and dyeing time quantities. The spectrophotometric analysis results indicate that color coordinates vary based on mordant concentration, mordanting, and dyeing duration. Furthermore, the results proved that microwave energy significantly shortened the mordanting and dyeing duration, resulting in an eco-friendly dyeing process. In this investigation, a feed-forward neural network (FFNN) model with sigmoid hidden neurons and a linear output neuron was used to predict the color strength dyeing property of merino wool fiber. Experimental results showed that the proposed model achieved a regression value of 0.9 for the color strength dyeing property. As demonstrated, the proposed FFNN model is effective and can be utilized to forecast the color strength dyeing properties of merino wool fiber.

Application of neural networks and neuro-fuzzy models in construction scheduling

Construction scheduling is a complex process that involves a large number of variables, making it difficult to develop accurate and efficient schedules. Traditional scheduling techniques rely on manual analysis and intuition, which are prone to errors and often fail to account for all the variables involved. This results in project delays, cost overruns, and poor project performance. Artificial intelligence models have shown promise in improving construction scheduling accuracy by incorporating historical data, site-specific conditions, and other variables that traditional scheduling methods may not consider. In this research study, application of soft-computing techniques to evaluate construction schedule and control of project activities in order to achieve optimal performance in execution of building projects were carried out. Artificial neural network and neuro-fuzzy models were developed using data extracted from a residential two-storey reinforced concrete framed-structure construction schedule and project execution documents. The evaluation of project performance indicators in earned value analysis from 0 to 100% progress at 5% increment with a total of seventeen tasks were carried out using Microsoft Project software and data obtained from the computation were utilized for model development. Using input–output and curve-fitting (nftool) function in MATLAB, a 6-10-1 two-layer feed-forward network with tansig activation-function (AF) for the hidden neurons and linear AF output neurons was generated with Levenberg–Marquardt (Trainlm) training algorithm. Similarly, with the aid of ANFIS toolbox in MATLAB software, the training, testing and validation of the ANFIS model were carried out using hybrid optimization learning algorithm at 100 epochs and the Gaussian-membership-function (gaussmf). Loss-function parameters namely MAE, RMSE and R-values were taken as the performance evaluation criteria of the developed models. The generated statistical results indicates no significant difference between model-results and experimental values with MAE, RMSE, R2 of 1.9815, 2.256 and 99.9% respectively for ANFIS-model and MAE, RMSE, R2 of 2.146, 2.4095 and 99.998% respectively for the ANN-model. The model performance indicated that the ANFIS-model outclassed the ANN-model with their results satisfactory to deal with complex relationships between the model variables to produce accurate target response. The findings from this research study will improve the accuracy of construction scheduling, resulting in improved project performance and reduced costs.

Neural Network Aided Homogenization Approach for Predicting Effective Thermal Conductivity of Composite Construction Materials.

Thermal conductivity is a fundamental material parameter involved in various infrastructure design guides around the world. This paper developed an innovative neural network (NN) aided homogenization approach for predicting the effective thermal conductivity of various composite construction materials. The 2-D meso-structures of dense graded asphalt mixture, porous asphalt mixture, and cement concrete were generated and divided into 2n × 2n square elements with specific thermal conductivity values. A two-layer feed-forward neural network with sigmoid hidden neurons and linear output neurons was built to predict the effective thermal conductivity of the 2 × 2 block. The Levenberg-Marquardt backpropagation algorithm was used to train the network. By repeatedly using the neural network, the effective thermal conductivities of 2-D meso-structures were calculated. The accuracy of the above NN aided homogenization approach was validated with experiment, and various factors affecting the effective thermal conductivity were analyzed. The analysis results show that the accuracy of the NN aided approach is acceptable with relative errors of 1.92~4.34% for the dense graded asphalt mixture, 1.10~6.85% for the porous asphalt mixture, and 1.13~3.14% for the cement concrete. The relative errors for all the materials are lower than 5% when the heterogeneous structures are divided into 512 × 512 elements. Ignoring the actual material meso-structures may lead to significant errors (134.01%) in predicting the effective thermal conductivity of materials with high heterogeneity such as porous asphalt mixture. While proper simplification is acceptable for dense construction composite materials. The effective thermal conductivity of composite cement-asphalt mixtures increases with higher saturation of grouted material. However, the improvement effect of the high-conductive cement paste on the composite cement-asphalt mixtures could be significantly reduced when the cement paste concentrates at the bottom of the mixture. Cracked aggregates and segregation of material components tend to decrease the effective thermal conductivity of construction materials. The NN aided homogenization approach presented in this paper is useful for selecting the effective thermal conductivity of construction materials.

ANN-based performance prediction of electrical discharge machining of Ti-13Nb-13Zr alloys

PurposeThe purpose of this paper is to predict the machining performance of electrical discharge machining of Ti-13Nb-13Zr (TNZ) alloy, a promising biomedical alloy, using artificial neural networks (ANN) models.Design/methodology/approachIn the research, three major performance characteristics, i.e. the material removal rate (MRR), tool wear rate (TWR) and surface roughness (SR), were chosen for the study. The input parameters for machining were the voltage, current, pulse-on time and pulse-off time. For the ANN model, a two-layer feedforward network with sigmoid hidden neurons and linear output neurons were chosen. Levenberg–Marquardt backpropagation algorithm was used to train the neural networks.FindingsThe optimal ANN structure comprises four neurons in input layer, ten neurons in hidden layer and one neuron in the output layer (4–10-1). In predicting MRR, the 60–20-20 data split provides the lowest MSE (0.0021179) and highest R-value for training (0.99976). On the contrary, the 70–15-15 data split results in the best performance in predicting both TWR and SR. The model achieves the lowest MSE and highest R-value for training in predicting TWR as 1.17E-06 and 0.84488, respectively. Increasing the number of hidden neurons of the network further deteriorates the performance. In predicting SR, the authors find the best MSE and R-value as 0.86748 and 0.94024, respectively.Originality/valueThis is a novel approach in performance prediction of electrical discharge machining in terms of new workpiece material (TNZ alloys).

Trans-media resistance investigation of hybrid aerial underwater vehicle base on hydrodynamic experiments and machine learning

In this paper, the trans-media hydrodynamics of a multi-rotor hybrid aerial underwater vehicle (HAUV) is studied base on experiments and machine learning. First, experiments are conducted to obtain the hydrodynamic forces during water exit/entry. Then, based on the experimental data and phenomena, it is evident that the multi-rotor HAUV's total resistance, particularly the impact of the arms, should not be ignored. The overall resistance is primarily affected by the depth and velocity at constant velocity. The total resistance increases as the velocity increases. The maximum total resistance in water exit/entry can be 46%/64% of gravity. Finally, a two-layer feed-forward network with six sigmoid hidden neurons and linear output neurons is used for machine learning training. The training results show that using experimental data and a machine learning method, the trans-media resistance of HAUV can be accurately predicted within a certain range.

A novel neural network and grey correlation analysis method for computation of the heat transfer limit of a loop heat pipe (LHP)

A loop heat pipe (LHP) has the advantages of larger heat transfer capacity and anti-gravity operational performance. The current prediction models for LHP heat transfer capacity have the difficulties in popularization of data volume and determination of accurate parametrical data, leading to the uncertain and varying outcomes that are inconsistent and away from reality. To address these challenges, this paper developed a first-of-its-kind big-data-driven LHP heat transfer limit prediction model by employing the neural network and grey correlation analysis method, which have advantages of high precision and large data volume. A double-layer feedforward neural network with sigmoid hidden neuron and linear output neuron was constructed to predict the heat transfer limit of the LHP. The grey scale analysis is applied to select the variables with correlation coefficient greater than 0.5, thus giving the clear identification of the both input parameters (e.g. refrigerant temperature, filling liquid quantity, height difference between evaporator and condenser, and number of heat pipe array) and output ones (heat transfer limit). The previously validated LHP heat transfer limit calculation model is used to calculate the heat transfer limit corresponding to the selected parameters, thus formulating 1,010,038 sets of data points. Of those calculated datasets, 707,026 (70% of data) are treated as a training set, 151,506 (15% of data) as a verification set, and 151,506 groups of data (15% of data) as the test sets for training. After several optimization and debugging, the number of hidden layer neurons is determined to be 100. The correlation coefficient (R), mean square error (MSE) and mean relative error (MRE) are 0.9997, 52.7 and 0.32% respectively, all of which are within reasonable accuracy range. The results show that the model has good prediction accuracy and consistence and is an effective tool to characterize and optimize the LHP in various application synergies.

Neural Network Modeling of the Solar Photovoltaic Thermal Module Performance

The design of photovoltaic thermal modules (PVT modules) features a nonuniform distribution of coolant temperature in the channel, and the solar cells (SC) that are in thermal contact with the PVT module channel are under different temperature conditions. The nonuniform distribution of the SC temperature causes undesirable effects that are complex and nonlinear, such as a drop of generated power and SC damage resulting from the occurrence of hot spots. The aim of the work is to develop a mathematical model for modeling the PVT module thermal and electrical performance based on a feedforward artificial neural network (FNN). A two-layer FNN with sigmoid hidden neurons and linear output neurons has been developed. The input layer is made up of the ambient temperature, coolant flowrate and environmental variables (the total insolation). The output layer represents the PVT module thermal and electrical efficiencies. The developed FNN was trained and adapted on the basis of modeling and an experimental database on the PVT module input and output parameters. The developed FNN was trained using the Levenberg–Marquardt algorithm. The mean absolute error achieved during the training is in the range from –0.319 to 0.448 for electrical efficiency and from –0.129 to 0.198 for thermal efficiency. The r.m.s error is 0.0678 for electrical efficiency and 0.0247 for thermal efficiency; the training time is 15 s or longer. An effective model has been developed that implements a new approach to modeling the performance of PVT modules based on artificial neural network algorithms with fairly close values of thermal and electrical efficiencies.

Lane‐Changing Trajectory Prediction Modeling Using Neural Networks

Concerning autonomous driving, lane‐changing (LC) is essential, particularly within complicated dynamic settings. It is a challenging task to model LC since driving behavior is complicated and uncertain. The present study adopts a dual‐layer feed‐forward backpropagation neural network involving sigmoid hidden neurons and linear output neurons for evaluating intrinsic LC complexity. Furthermore, the estimation and validation of the model were performed by large‐scale trajectory data. Empirical LC data were obtained from the Next Generation Simulation (NGSIM) project for training and testing the neural network‐based LC model. The findings revealed that the introduced model could make precise LC predictions of vehicles under small trajectory errors and satisfactory accuracy. The present work assessed LC beginning/endpoints and velocity estimates by analyzing the vehicles around. It was observed that the neural network model yielded almost the same predictions as the observational LC trajectories as well as following vehicle trajectories on the original and target lanes. Furthermore, for LC behavior characteristic validation, the neural network‐produced LC gap distributions underwent comparisons to real‐life data, demonstrating the characteristics of LC gap distributions not to differ from the real‐life LC behavior substantially. Eventually, the introduced neural network‐based LC model was compared to a support vector regression‐based LC model. It was found that the trajectory predictions of both models were adequately consistent with the observational data and could capture both lateral and longitudinal vehicle movements. In turn, this demonstrates that the neural network and support vector regression models had satisfactory performance. Also, the proposed models were evaluated using new inputs such as speed, gap, and position of the subject vehicle. The analysis findings indicated that the performance of the proposed NN and SVR models was higher than the model with new inputs.

Modeling the Levelized Cost of Energy for Concentrating Solar Thermal Power Systems Based on a Nonlinear AutoRegressive Neural Network With Exogenous Inputs

This article presents the results of the development of a mathematical model for predicting the levelized cost of energy (LCOE) for solar concentrating thermal power systems (CSP systems) based on a nonlinear autoregressive neural network with exogenous inputs (NARX). A two-layer NARX network with sigmoid hidden neurons and linear output neurons has been developed. The input layer is made up of the following variables: the volume of input power of CSP systems in the world, the total world energy consumption, domestic energy consumption, domestic gas consumption, domestic consumption of coal and lignite, domestic energy consumption, the share of renewable energy in electricity generation, the share of wind and solar energy in the production of electricity, carbon dioxide emissions from fuel combustion, the price of Brent oil against the US dollar, and the average price for natural gas auctions. The output layer specifies LCOE values for CSP systems.

Investigations on the relationship among the porosity, permeability and pore throat size of transition zone samples in carbonate reservoirs using multiple regression analysis, artificial neural network and adaptive neuro-fuzzy interface system

Finding an accurate method for estimating permeability aside from well logs has been a difficult task for many years. The most commonly used methods targeted towards regression technique to understand the correlation between pore throat radii, porosity and permeability are Winland and Pittman equation approaches. While these methods are very common among petrophysicists, they do not give a good prediction in certain cases. Consequently, this paper investigates the relationship among porosity, permeability, and pore throat radii using three methods such as multiple regression analysis, artificial neural network (ANN), and adaptive neuro-fuzzy inference system (ANFIS) for application in transition zone permeability modeling. Firstly, a comprehensive mercury injection capillary pressure (MICP) test was conducted using 228 transition zone carbonate core samples from a field located in the Middle-East region. Multiple regression analysis was later performed to estimate the permeability using pore throat and porosity measurement. For the ANN, a two-layer feed-forward neural network with sigmoid hidden neurons and a linear output neuron was used. The technique involves training, validation, and testing of input/output data. However, for the ANFIS method, a hybrid optimization consisting of least-square and backpropagation gradient descent methods with a subtractive clustering technique was used. The ANFIS combines both the artificial neural network and fuzzy logic inference system (FIS) for the training, validation, and testing of input/output data. The results show that the best correlation for the multiple regression technique is achieved for pore throat radii with 35% mercury saturation (R35). However, for both the ANN and ANFIS techniques, pore throat radii with 55% mercury saturation (R55) gives the best result. Both the ANN and ANFIS are later found to be more effective and efficient and thus recommended as compared with the multiple regression technique commonly used in the industry.

Нейросетевое прогнозирование полной приведенной стоимости электроэнергии для солнечных фотоэлектрических систем

Elaboration of a new approach to the development of models for predicting the economic indicators of solar photovoltaic systems by using artificial neural network algorithms is becoming of special importance. As is known, the relationships between economic indicators are often difficult to identify. Nonlinear autoregressive models can provide more reliable results than those obtained from predictive linear models based on vector autoregression. The article presents the results from the development of a mathematical model for predicting the levelized cost of energy (LCOE) for solar photovoltaic systems based on a nonlinear autoregressive neural network with exogenous inputs (NARX). A two-layer NARX network with hidden sigmoid neurons and linear output neurons has been developed. The input layer is made up of the following variables: the amount of power consumed from solar photovoltaic systems around the world; the total worldwide energy consumption; domestic consumption of energy, gas, coal, and lignite; the shares of renewable energy, wind and solar energy in electricity generation; carbon dioxide emissions from fuel combustion; the price of Brent oil in US dollars, and the average price for natural gas. The output layer determines the LCOE values for solar photovoltaic systems. The developed NARX network was trained on the basis of retrospective data for 2005-2010 using the Levenberg-Marquardt algorithm. The correlation coefficient value achieved in the course of training made 0.99904, and the mean square error value was in the range from 0.00042 to 0.0029

DEVELOPMENT OF NUMERICAL MODELS FOR THE PREDICTION OF TEMPERATURE AND SURFACE ROUGHNESS DURING THE MACHINING OPERATION OF TITANIUM ALLOY (Ti6Al14V)

Temperature and surface roughness are important factors, which determine the degree of machinability and the performance of both the cutting tool and the work piece material. In this study, numerical models obtained from the Response Surface Methodology (RSM) and Artificial Neural Network (ANN) techniques were used for predicting the magnitude of the temperature and surface roughness during the machining operation of titanium alloy (Ti6Al4V). The design of the numerical experiment was carried out using the Response Surface Methodology (RSM) for the combination of the process parameters while the Artificial Neural Network (ANN) with 3 input layers, 10 sigmoid hidden neurons and 3 linear output neurons were employed for the prediction of the values of temperature. The ANN was iteratively trained using the Levenberg-Marquardt backpropagation algorithm. The physical experiments were carried out using a DMU80monoBLOCK Deckel Maho 5-axis CNC milling machine with a maximum spindle speed of 18 000 rpm. A carbide-cutting insert (RCKT1204MO-PM S40T) was used for the machining operation. A professional infrared video thermometer with an LCD display and camera function (MT 696) with infrared temperature range of −50−1000 °C, was employed for the temperature measurement while the surface roughness of the work pieces were measured using the Mitutoyo SJ – 201, surface roughness machine. The results obtained indicate that there is high degree of agreement between the values of temperature and surface roughness measured from the physical experiments and the predicted values obtained using the ANN and RSM. This signifies that the developed RSM and ANN models are highly suitable for predictive purposes. This work can find application in the production and manufacturing industries especially for the control, optimization and process monitoring of process parameters.

Modelling of the mechanical properties of concrete with cement ratio partially replaced by aluminium waste and sawdust ash using artificial neural network

The use of aluminium waste (AW) and sawdust ash (SDA) in concrete was evaluated in this study where the cement ratio was partially replaced by fractions of AW and SDA introduced as a supplementary cementitious material. Artificial neural network (ANN) was adapted as the modelling tool for this study and was developed with a two-layer feed-forward network, hidden neurons with sigmoid activation function and linear output neurons for the simulation of the network. The setting time and concrete compressive strength at varying curing days were predicted using the neural network model with variations of constituents of the cement content consisting of OPC, SDA and AW as the input of the network. Three input and seven output data set were used for the model development using the following algorithms; Data Division: Random, Training: Levenberg–Marquardt and Calculation: MATLAB. The data sets are set aside for validation, training and testing; 70% of the samples are used for training, 15% for validation and 15% are also used for testing. The performance of the networks was evaluated using linear regression, RMSE and R-values. The model performance scored 0.91 and 0.07 for R2 and RMSE, respectively, and performed better than the linear regression model, the results indicate the efficiency, reliability and usefulness of ANN for predicting concrete mechanical properties where AW and SDA are used to replace cement ratio accurately.

A Smart High Accuracy Calibration Algorithm for 3-D Piezoresistive Stress Sensor

Numerical and experimental sensitivity analyses in this paper indicate that the accuracy of a silicon multi-element piezoresistive (PR) stress sensor can be dramatically influenced by the microfabrication non-uniformity and the uncertainties in the values of the PR coefficients and the thermal coefficient of resistance (TCR). The results showed that errors as large as 70% FS or more, in the extracted stress values, may be obtained due to uncertainty of about 2.5% in the values of PR coefficients. This paper aims to evaluate the capabilities of the artificial neural network (ANN) to eliminate the error in stress measurement, due to the fabrication non-uniformity within wafer, wafer-to-wafer, and batch-to-batch, for multi-element PR sensing rosettes. In this paper, sensing chips from two different batches were integrated in building the ANN and testing its performance. The proposed calibration technique employs the neural network fitting Toolbox in MATLAB to generate a two-layer feed-forward network, with sigmoid hidden neurons and linear output neurons. Three different configurations of calibration were designed to test the generalization abilities of the ANN in capturing the in-plane stress components exerted on the silicon chip. The results showed that ANN is capable of accurately predicting the stresses applied to the sensing chip with maximum stress error of 1.5% FS with no need for individual, expensive, and time-consuming calibration process for each sensor.

An Artificial Neural Network and Genetic Algorithm Optimized Model for Biogas Production from Co-digestion of Seed Cake of Karanja and Cattle Dung

In this study, experiments were conducted with four different proportions of seed cake of Karanja (SCK) and cattle dung (CD) mixture, for biogas production. 75, 50 and 25 % of the SCK on a mass basis were mixed with 25, 50 and 75 % of the CD and, named as S1, S2 and S3. For comparison, biogas obtained from 100 % CD (S4) was considered. The samples were kept in four different reactors, for 30 days of observation, and the yield of biogas from the samples S1, S2 and S3 was evaluated. Modeling was carried out for prediction and optimization of biogas production using ANN (artificial neural network) and the GA (genetic algorithm). A multi-layered feed-forward network with hidden neurons and linear output neurons was used for training the network using the input parameters pH, digestion time and C/N ratio for the yield of biogas. The performance of the neural network model was verified, and the correlation coefficients were found to be close to 1, for the samples. The experimental results on the biogas production were validated with the results of the neural network and optimized with the GA. The GA optimized values for pH, digestion time, and the C/N ratio of sample S3 were found to be 6.68, 14.22 days and 24.1:1, respectively. These optimized data can be used to monitor a large scale anaerobic plant. Among all samples, S3 gave a better result with respect to the pH, C/N (carbon/nitrogen) ratio, digestion time and biogas yield.

PREDICTION OF TIGRIS RIVER DISCHARGE IN BAGHDAD CITY USING ARTIFICAL NEURAL NETWORKS

 Artificial Neural Networks (ANNs), with three layers feed- forward network of sigmoid hidden neurons and linear output neurons are performed for predicting Tigris River flow in Baghdad City, middle of Iraq. The network is trained with Levenberg-Marquradt back-propagation algorithm. The number of hidden neurons is estimated according to trial and error procedure. The best model is selected according to trial and error procedure based on root mean square error and coefficient of correlation. The selected model is used to predicate the river discharge for one, two, and three months ahead. Results indicate that the ANNs with Levenberg-Marquradt back-propagation algorithm are a powerful tool for forecasting the river discharge for short term duration. But this ability begins to decrease when increasing the period of forecasting.

The association between cell assemblies and transient dynamics

Neural systems show a wide variety of complex dynamics on different time scales. Specifically, on the short time scale of neuronal activations (milliseconds to seconds), several theoretical models (e.g., state-dependent networks or liquid state machines [1]) demonstrate that complex (transient) dynamics enables neural circuits to process a broad range of nonlinear problems. In other words, neural circuits process inputs to transform them into desired outputs. This approach, amongst others, is based on the inherent stochasticity (randomness) of neural systems. On the other hand, the slower mechanisms of synaptic plasticity (minutes to hours) adapt synaptic efficacies to form ordered structures as neural cell assemblies [2]. Each cell assembly consists of a group of strongly interconnected neurons and its dynamics serves as an associative memory of the previously experienced input. This type of dynamics is in stark contrast to the needed randomness of connectivity for transient dynamics. Intriguingly, several experiments show that both concepts coexist in the same neural circuits [3,4]. In this study we analyze how this coexistence of transient dynamics and cell assemblies can emerge in one neural circuit. The neural network, we investigate, consists of rate- based neurons with connections adapted by a generic combination of Hebbian plasticity and synaptic scaling [5]. A subset of neurons receives repeatedly a time-varying input and forms a cell assembly (Figure ​(Figure1).1). Further repetitions of the input lead to a gradual growth of the cell assembly by incorporating more neurons. In parallel, the network is required to compute a desired nonlinear version of the input. To test whether this computation is successful, the connections between the neurons of the network and a linear output neuron are adapted by a supervised learning algorithm [6]. Remarkably, the performance of the computation increases with the growth in size of the cell assembly. Thus, the slow, ordering dynamics of synaptic plasticity supports the computational capability of the fast transient dynamics. Furthermore, as the cell assemblies are topologically compact, the network is able to perform several computations in parallel (green and red cell assemblies in Figure ​Figure1).1). However, if the cell assemblies become neighbors, dependent on the input, one cell assembly can capture neurons from the other assembly (stripped area in Figure ​Figure1).1). Thus, the cell assemblies compete for computational resources, namely neurons and synapses. Therefore, our work shows a new functional role of cell assemblies in neuronal systems, whereby both stable and transient dynamics are byproduct of the same neural circuit. Thus, both concepts are not mutually exclusive but that they interact with each other in a positive way. Figure 1 Basic setting of the system. CAi: Cell Assembly i; Ii: Input to CAi; Oi: Output of CAi; Fi: Function processed by CAi

Neural Network Based Model of an Industrial Oil-Fired Boiler System

In this study, an oil-fired boiler system is modeled as a multivariable plant with two inputs (feed water rate and oil-fired flow rate) and two outputs (steam temperature and pressure). The plant parameters are modeled using artificial neural network, based on experimental data collected directly from the physical plant. A two-layer feed-forward neural network with Hyperbolic tangent sigmoid transfer function (Tansigmoid) and linear output neurons (Purelin) are used at the hidden and output layer respectively to fit the neural network model. The neural network model is then trained off-line with Levenberg-Marquardt Back Propagation Algorithm (trainlm). The neural network model when subjected to test, using the validation input data; shows that the simulated model outputs for both temperature and pressure agree closely with the actual plant outputs, with regression value of 0.97. Furthermore, Proportional Integral Derivative (PID) Controller is used to control the neural network model. Simulation studies results indicate the effectiveness of the developed technique.