Prediction of CH4 adsorption on different activated carbons by developing an optimal multilayer perceptron artificial neural network

A deep understanding of adsorption processes is essential for the design and optimization of industrial units. Storage of methane adsorbed on activated carbon (AC) at low pressure and room temperature (adsorbed natural gas) has been studied in recent years as an alternative model to compressed natural gas and liquefied natural gas technologies. The current study plays a significant role in modeling CH₄ adsorption on different ACs through the optimal multilayer perceptron (MLP) neural network . Therefore, lots of adsorption data points were used for modeling. To optimize the efficiency of a predictive model, two optimization algorithms including LevenbergMarquardt (LM) and Bayesian regularization were utilized to find the optimal models’ parameters during prediction analysis. In order to demonstrate the efficiency of the proposed method, it is compared with several other experimental data points. Results of optimizations indicate the superiority of the proposed method over the other techniques, and forecasting error is remarkably reduced. As a result, it was found that the MLP-LM is the more accurate model for estimating CH₄ adsorption with root-mean-square error and coefficient of determination of 0.00025 and 0.9921, respectively.     

Study of hourly and daily solar irradiation forecast using diagonal recurrent wavelet neural networks

An accurate forecast of solar irradiation is required for various solar energy applications and environmental impact analyses in recent years. Comparatively, various irradiation forecast models based on artificial neural networks (ANN) perform much better in accuracy than many conventional prediction models. However, the forecast precision of most existing ANN based forecast models has not been satisfactory to researchers and engineers so far, and the generalization capability of these networks needs further improving. Combining the prominent dynamic properties of a recurrent neural network (RNN) with the enhanced ability of a wavelet neural network (WNN) in mapping nonlinear functions, a diagonal recurrent wavelet neural network (DRWNN) is newly established in this paper to perform fine forecasting of hourly and daily global solar irradiance. Some additional steps, e.g. applying historical information of cloud cover to sample data sets and the cloud cover from the weather forecast to network input, are adopted to help enhance the forecast precision. Besides, a specially scheduled two phase training algorithm is adopted. As examples, both hourly and daily irradiance forecasts are completed using sample data sets in Shanghai and Macau, and comparisons between irradiation models show that the DRWNN models are definitely more accurate.     

Multiplicative ARMA models to generate hourly series of global irradiation

A methodology to generate hourly series of global irradiation is proposed. The only input parameter which is required is the monthly mean value of daily global irradiation, which is available for most locations. The procedure to obtain new series is based on the use of a multiplicative autoregressive moving-average statistical model for time series with regular and seasonal components. The multiplicative nature of this models enables us to capture the two types of relationships observed in recorded hourly series of global irradiation: on the one hand, the relationship between the value at one hour and the value at the previous hour; and on the other hand, the relationship between the value at one hour in one day and the value at the same hour in the previous day. In this paper the main drawback which arises when using these models to generate new series is solved: namely, the need for available recorded series in order to obtain the three parameters contained in the statistical ARMA model which is proposed (autoregressive coefficient, moving-average coefficient and variance of the error term). Specifically, expressions which enable estimatation of these parameters using only monthly mean values of daily global irradiation are proposed in this paper.     

Convective heat transfer and pressure drop study on nanofluids in double-walled reactor by developing an optimal multilayer perceptron artificial neural network

The effect of nanofluid on the cooling performance and pressure drop of a jacked reactor has experimentally been investigated. Aqueous nanofluids of Al₂O₃ and CuO was used as the cool ant inside the cooling jacket of the reactor. The application of the artificial neural networks (ANNs) to predict the performance of a double-walled reactor has been studied. Different architectures of artificial neural networks were developed to predict the convective heat transfer and pressure drop of nanofluids. The experimental results are used for training and testing the ANNs based on two optimal models via feed-forward back-propagation multilayer perceptron (MLP). The comparison of statistical criteria of different network shows that the optimal structure for predicting the convective heat transfer coefficient is the MLP network with one hidden layer and 10 neurons, which has been trained with Levenberg–Marquardt (LM) algorithm. The predicted pressure drop values by the MLP network with two hidden layers and 6 neurons in the each layer has been used from LM training algorithm, which showed a reasonable agreement with the experimental results.     

A new neural network model for evaluating the performance of various hourly slope irradiation models: Implementation for the region of Athens

The present study is divided into two parts. The first part deals with the comparison of various hourly slope irradiation models, found in the literature, and the selection of the most accurate for the region of Athens. In the second part the prediction of global solar irradiance on inclined surfaces is performed, based on neural network techniques.The models tested are classified as isotropic (Liu and Jordan, Koronakis, Jimenez and Castro, Badescu, Tian) and anisotropic (Bugler, Temps and Coulson, Klucher, Ma and Iqbal, Reindl) based on the treatment of diffuse irradiance. For the aforementioned models, a qualitative comparison, based on diagrams, was carried out, and several statistical indices were calculated (coefficient of determination R², mean bias error MBE, relative mean bias error MBE/A(%), root mean square error RMSE, relative root mean square error RMSE/A(%),statistical index t-stat), in order to select the optimal.The isotropic models of “Tian” and “Badescu” show the best accordance to the recorded values. The anisotropic model of “Ma&Iqbal” and the pseudo-isotropic model of “Jimenez&Castro”, show poor performance compared to other models. Finally, a neural network model is developed, which predicts the global solar irradiance on a tilted surface, using as input data the total solar irradiance on a horizontal surface, the extraterrestrial radiation, the solar zenith angle and the solar incidence angle on a tilted plane. The comparison with the aforementioned models has shown that the neural network model, predicts more realistically the total solar irradiance on a tilted surface, as it performs better in regions where the other models show underestimation or overestimation in their calculations.     

Estimation of hourly in-plane irradiation by using minutely horizontal data

To propose a formula for calculating in-plane irradiation onto a tilt surface from measured data on the tilt angle, detailed examination has been made by using data obtained for every 1 min. A new model to estimate in-plane irradiation from horizontal irradiance has been developed. Especially, it is showed that new model can estimate scattered irradiance correctly rather than the existing model by using scattered component ratio and clearness index. It is considered that measuring 1-min data has favorable influences instead of hourly data. Fluctuation of 1-min data contributes to estimate the scattered component.     

Prediction of hourly and daily diffuse fraction using neural network,as compared to linear regression models

For most of the locations all over Egypt the records of diffuse radiation in whatever scale are non-existent. In case that it exists, the quality of these records is not as good as it should be for most purposes and so an estimate of its values is desirable. To achieve such a task, an artificial neural network (ANN) model has been proposed to predict diffuse fraction (K_D) in hourly and daily scale. A comparison between the performances of the ANN model with that of two linear regression models has been reported. An attempt was also done to describe the ANN outputs in terms of first order polynomials relating K_D with clearness index (K_T) and sunshine fraction (S/S₀). If care is taken in considering the corresponding regional climatic differences, these correlations can be generalized and transferred to other sites. The results hint that the ANN model is more suitable to predict diffuse fraction in hourly and daily scales than the regression models in the plain areas of Egypt.     

Stochastic modeling and generation of synthetic sequences of hourly global solar irradiation at Quetta, Pakistan

Using hourly global radiation data at Quetta, Pakistan for 10 yr, an Autoregressive Moving Average (ARMA) process is fitted. Markov Transition Matrices have also been developed. These models are used for generating synthetic sequences for hourly radiations in MJ/m² and that the generated sequences are compared with the observed data. We found the MTM approach relatively better as a simulator compared to ARMA modeling.     

Prediction of chemical exergy of organic substances using artificial neural network-multi layer perceptron

In chemical processes, it would be much beneficial to develop a technique to precisely predict the standard chemical exergy values of different compounds. In this study, a multi-layer perceptron (MLP) network model is developed by using a set of 134 data points obtained from the literature. 114 and 24 Numbers of data points were allocated to training and testing steps, respectively. In addition, molecular weight, number of atoms, and sum of atomic polarizabilities properties were selected as input parameters to be the representative of the substances under consideration. The R² and AARE values were 0.9976 and 2.786, respectively, and the experimental and predicted exergy values showed a great overlap with each other. In fact, based on the results obtained in this paper, the proposed MLP model could be represented as a novel technique in order to calculate the standard chemical exergy values of different compounds.     

An application of the multilayer perceptron: Solar radiation maps in Spain

In this work an application of a methodology to obtain solar radiation maps is presented. This methodology is based on a neural network system [Lippmann, R.P., 1987. An introduction to computing with neural nets. IEEE ASSP Magazine, 4–22] called Multi-Layer Perceptron (MLP) [Haykin, S., 1994. Neural Networks. A Comprehensive Foundation. Macmillan Publishing Company; Hornik, K., Stinchcombe, M., White, H., 1989. Multilayer feedforward networks are universal approximators. Neural Networks, 2(5), 359–366]. To obtain a solar radiation map it is necessary to know the solar radiation of many points spread wide across the zone of the map where it is going to be drawn. For most of the locations all over the world the records of these data (solar radiation in whatever scale, daily or hourly values) are non-existent. Only very few locations have the privilege of having good meteorological stations where records of solar radiation have being registered. But even in those locations with historical records of solar data, the quality of these solar series is not as good as it should be for most purposes. In addition, to draw solar radiation maps the number of points on the maps (real sites) that it is necessary to work with makes this problem difficult to solve. Nevertheless, with the application of the methodology proposed in this paper, this problem has been solved and solar radiation maps have been obtained for a small region of Spain: Jaén province, a southern province of Spain between parallels 38°25′ N and 37°25′ N, and meridians 4°10′ W and 2°10′ W, and for a larger region: Andalucía, the most southern region of Spain situated between parallels 38°40′ N and 36°00′ N, and meridians 7°30′ W and 1°40′ W.     

Modeling and forecasting the mean hourly wind speed time series using GMDH-based abductive networks

Wind speed forecasts are important for the operation and maintenance of wind farms and their profitable integration into power grids, as well as many important applications in shipping, aviation, and the environment. Modern machine learning techniques including neural networks have been used for this purpose, but it has proved hard to make significant improvements on the performance of the simple persistence model. As an alternative approach, we propose here the use of abductive networks, which offer the advantages of simplified and more automated model synthesis and transparent analytical input–output models. Various abductive models for predicting the mean hourly wind speed 1 h ahead have been developed using wind speed data at Dhahran, Saudi Arabia during the month of May over the years 1994–2005. The models were evaluated on the data for May 2006. Models described include a single generic model to forecast next-hour speed from the previous 24 hourly measurements and an hour index, which give an overall mean absolute error (MAE) of 0.85 m/s and a correlation coefficient of 0.83 between actual and predicted values. The model achieves an improvement of 8.2% reduction in MAE compared to hourly persistence. The above model was used iteratively to forecast the hourly wind speed 6 h and 24 h ahead at the end of a given day, with MAEs of 1.20 m/s and 1.42 m/s which are lower than forecasting errors based on day-to-day persistence by 14.6% and 13.7%. Relative improvements on persistence exceed those reported for several machine learning approaches reported in the literature.     

基于小波神经网络的太阳辐照强度预测

提出了基于小波神经网络的太阳辐照强度预测方法。利用皮尔逊相关系数分析法和曲线估计筛选出影响太阳辐照强度的重要因素;采用小波理论和神经网络理论相结合的小波神经网络分别建立春、夏、秋、冬4个预测模型;采用最小均方误差能量函数法自动优化网络结构,把历史太阳辐照强度、经度、纬度、海拔高度、天气类型、日照时数、最高温度、最低温度、相对湿度、大气压强作为模型的最优输入;采用L-M训练方法对太阳辐照强度进行了min级预测。通过对4个季节特殊天气类型的太阳辐照强度预测,并与BP神经网络进行对比,验证了该方法的可行性和准确性。     

Generation of daily solar irradiation by means of artificial neural net works

The present study proposes the utilization of Artificial Neural Networks (ANN) as an alternative for generating synthetic series of daily solar irradiation. The sequences were generated from the use of daily temporal series of a group of meteorological variables that were measured simultaneously. The data used were measured between the years of 1998 and 2006 in two temperate climate localities of Brazil, Ilha Solteira (São Paulo) and Pelotas (Rio Grande do Sul). The estimates were taken for the months of January, April, July and October, through two models which are distinguished regarding the use or nonuse of measured bright sunshine hours as an input variable. An evaluation of the performance of the 56 months of solar irradiation generated by way of ANN showed that by using the measured bright sunshine hours as an input variable (model 1), the RMSE obtained were less or equal to 23.2% being that of those, although 43 of those months presented RMSE less or equal to 12.3%. In the case of the model that did not use the measured bright sunshine hours but used a daylight length (model 2), RMSE were obtained that varied from 8.5% to 37.5%, although 38 of those months presented RMSE less or equal to 20.0%.A comparison of the monthly series for all of the years, achieved by means of the Kolmogorov–Smirnov test (to a confidence level of 99%), demonstrated that of the 16 series generated by ANN model only two, obtained by model 2 for the months of April and July in Pelotas, presented significant difference in relation to the distributions of the measured series and that all mean deviations obtained were inferior to 0.39 MJ/m².It was also verified that the two ANN models were able to reproduce the principal statistical characteristics of the frequency distributions of the measured series such as: mean, mode, asymmetry and Kurtosis.     

Time series analysis of hourly global horizontal solar radiation

Accurate design and optimization of short response time solar energy systems with storage are sensitive to the stationary and sequential characteristics of hourly solar radiation. We perform monthly time series analyses of hourly global horizontal solar radiation for a wide range of climatic stations that span temperate and tropical conditions. The stationary statistics for individual hours are found to be very similar to the corresponding results for daily total global horizontal radiation, in keeping with a related fundamental observation of Liu & Jordan. Investigation of sequential properties shows that autocorrelation coefficients are, to a good approximation, independent of time of day and that persistence times are nearly as long as the entire daylight period, mainly due to the effect of very strong correlations at one-hour lag times. The isolated effect of two-hour and longer lag times, via the partial autocorrelation coefficients, is found to be negligible in most, but by no means all, instances. Finally, we find no universal correlation between hourly autocorrelation coefficients and monthly average radiation figures.     

Predicting total solar irradiation values using artificial neural networks

This study explores the possibility of developing an artificial neural networks model that could be used to predict monthly average daily total solar irradiation on a horizontal surface for locations in Uganda based on geographical and meteorological data: latitude, longitude, altitude, sunshine duration, relative humidity and maximum temperature. Results have shown good agreement between the predicted and measured values of total solar irradiation. A correlation coefficient of 0.997 was obtained with mean bias error of 0.018 MJ/m² and root mean square error of 0.131 MJ/m². Overall, the artificial neural networks model predicted with an accuracy of 0.1% of the mean absolute percentage error.     

Characterization and simulation of hourly exposure series of global radiation

A statistical model which captures the main features of hourly exposure series of global radiation is proposed. This model is used to obtain a procedure to generate radiation series without imposing, a priori, any restriction on the form of the probability distribution function of the series. The statistical model was taken from the stationary stochastic processes theory. Data were obtained from ten different Spanish locations. As monthly hourly exposure series of global radiation are not stationary, they are modified in order to remove the observed trends. A multiplicative autoregressive moving average model with regular and seasonal components was used. It is statistically accepted that this is the true model which generates most of the analyzed sequences. However, the underlying parameters of the model vary from one location to another and from one month to another. Therefore, it is necessary to examine further the relationship between the parameters of the model and the available data from most locations.     

Cost-optimal Calculation for Ice-Storage Systems Using Neural Network

Based on the eminent characteristics of the ice-storage systems, which can shift cooling electrical demand from peak time to off peak time, this paper describes the ice storage air-conditioning system that is now used much frequently. The authors develop the operating cost model by simplification and introduce a neural network model and try to solve the optimal cost problem of operation by using this neural network model. In calculation, any trajectory of the neural network converges to its solution in finite time, which is consistent with result by simplex method. Comparing with different methods, the neural network is more effective, which can be alternative to simplex method in calculating the optimal cost model for ice storage air-conditioning systems.     

Electric load forecasting by using dynamic neural network

Electrical energy is fundamental for the wellbeing and for the economic development of any country. However, all countries must ensure access to essential resources and ensure the continuity of its supply. Due to the non-storable nature of electrical energy, the amount of consumed active power should always be equal the produced active power just to avoid power system frequency deviation problem. In order to keep the relationship production–consumption relation in compliance with different standards and to secure profitable operations of power system, electric load consumption must be predicted and controlled instantaneously. Several statistical and classical techniques are proposed in the literature but unfortunately all these methods are not accurate in a satisfactory manner. In this paper, a dynamic neural network is used for the prediction of daily power consumption. The suitability and the performance of the proposed approach is illustrated and verified with simulations on load data collected from French Transmission System Operator (RTE) website. The obtained results show that the accuracy and the efficiency are improved comparatively to conventional methods widely used in this field of research.     

Electric load analysis using an artificial neural network

Load forecasting in the current, increasingly liberalized, electricity power market is of crucial importance as a means for producers to optimize and rationalize energy supply. A number of electric power companies are equipped to make forecasts with the aid of traditional statistical methods. This paper presents the use of an artificial neural net to an hourly based load forecasting application for a small electric grid on an Italian island (Lipari) not connected to the mainland. The aim was to examine the forecasting ability of a neural net in a situation where the electric load was subject to considerable seasonal variations over the year. The variations are affected by energy demand related to the tourism season as well as by climatic conditions, especially temperature. The network developed was a multi‐layer perceptron type built on three layers trained with a back‐propagation algorithm. The input layer receives all the most relevant information regarding: the class of day type, the hour in the daytime, the load and background temperature recorded at the indicated time for the months of March, August and October whilst the output layer provides the forecast value at the indicated time in December. The results obtained are encouraging; in the training phase the RMS error rate was around 2% and this rate settled at about 2.6% during testing. As both the margins of error recorded are acceptable, the use of a neural network for electric load forecasting applications can be considered an attractive option. Copyright © 2005 John Wiley & Sons, Ltd.