A new optimization technique for artificial neural networks applied to prediction of force constants of large molecules

1995 ◽  
Vol 16 (8) ◽  
pp. 923-936 ◽  
Author(s):  
Thomas H. Fischer ◽  
Wesley P. Petersen ◽  
Hans Peter Lüthi
Keyword(s):  
Neural Networks ◽  
Artificial Neural Networks ◽  
Optimization Technique ◽  
Force Constants ◽  
Large Molecules ◽  
Artificial Neural

scholarly journals Prediction of Sound Insulation of Sandwich Partition Panels by Means of Artificial Neural Networks

2017 ◽  
Vol 42 (4) ◽  
pp. 643-651
Author(s):  
Naveen Garg ◽  
Siddharth Dhruw ◽  
Laghu Gandhi
Keyword(s):  
Neural Networks ◽  
Artificial Neural Networks ◽  
Optimization Technique ◽  
Back Propagation ◽  
National Research Council ◽  
Sound Insulation ◽  
Ann Model ◽  
Back Propagation Algorithm ◽  
Design And Optimization ◽  
Artificial Neural

Abstract The paper presents the application of Artificial Neural Networks (ANN) in predicting sound insulation through multi-layered sandwich gypsum partition panels. The objective of the work is to develop an Artificial Neural Network (ANN) model to estimate the Rw and STC value of sandwich gypsum constructions. The experimental results reported by National Research Council, Canada for Gypsum board walls (Halliwell et al., 1998) were utilized to develop the model. A multilayer feed-forward approach comprising of 13 input parameters was developed for predicting the Rw and STC value of sandwich gypsum constructions. The Levenberg-Marquardt optimization technique has been used to update the weights in back-propagation algorithm. The presented approach could be very useful for design and optimization of acoustic performance of new sandwich partition panels providing higher sound insulation. The developed ANN model shows a prediction error of ±3 dB or points with a confidence level higher than 95%.

scholarly journals Short Term Load Forecasting System Using Artificial Neural Networks For Improved Demand Scheduling By Power Utilities.

2021 ◽  
Author(s):  
Jubilee Mzingaye Sibanda ◽  
Jephias Gwamuri
Keyword(s):  
Neural Networks ◽  
Artificial Neural Networks ◽  
Optimization Technique ◽  
Load Forecasting ◽  
Back Propagation Algorithm ◽  
Short Term ◽  
Short Term Load Forecasting ◽  
Forecasting System ◽  
Artificial Neural ◽  
Power Utilities

Abstract Load forecasting is a technique used by power utilities to predict electricity demand to maintain the balance between supply and demand. The problem comes when the power utilities draw more than the inadvertent power from the power pool. This necessitates the need for more accurate forecasting models. In this study, a short-term load forecasting system using artificial neural networks in MatLab was performed. The Levenberg-Marquardt optimization technique which has one of the best learning rates was used as a back-propagation algorithm for the Multilayer Feed Forward ANN model using MatLab® R2018a ANN Toolbox. Historical electricity load data obtained from a feeder line at the Zimbabwe Electricity Transmission and Distribution Company (ZETDC) Marvel 420 kV substation in Bulawayo Zimbabwe was used for the training, testing, and validation of the model. Results indicate that ANNs can forecast load with an accuracy of 6.71%. The results indicate that the proposed technique is robust in forecasting future load demands for the daily operational planning of power system distribution

scholarly journals A Hybrid Optimization Approach for Autonomy Enhancement of Nearly-Zero-Energy Buildings Based on Battery Performance and Artificial Neural Networks

Energies ◽  
2020 ◽  
Vol 13 (14) ◽  
pp. 3680
Author(s):  
Giorgos S. Georgiou ◽  
Pavlos Nikolaidis ◽  
Soteris A. Kalogirou ◽  
Paul Christodoulides
Keyword(s):  
Neural Networks ◽  
Artificial Neural Networks ◽  
Renewable Energy ◽  
Real Time ◽  
Optimization Technique ◽  
Zero Energy ◽  
Eu Directive ◽  
Import And Export ◽  
Artificial Neural ◽  
Zero Energy Buildings

Reducing the primary energy consumption in buildings and simultaneously increasing self-consumption from renewable energy sources in nearly-zero-energy buildings, as per the 2010/31/EU directive, is crucial nowadays. This work solved the problem of nearly zeroing the net grid electrical energy in buildings in real time. This target was achieved using linear programming (LP)—a convex optimization technique leading to global solutions—to optimally decide the daily charging or discharging (dispatch) of the energy storage in an adaptive manner, in real time, and hence control and minimize both the import and export grid energies. LP was assisted by equally powerful methods, such as artificial neural networks (ANN) for forecasting the building’s load demand and photovoltaic (PV) on a 24 hour basis, and genetic algorithm (GA)—a heuristic optimization technique—for driving the optimum dispatch. Moreover, to address the non-linear nature of the battery and model the energy dispatch in a more realistic manner, the proven freeware system advisor model (SAM) of National Renewable Energy Laboratory (NREL) was integrated with the proposed approach to give the final dispatch. Assessing the case of a building, the results showed that the annual hourly profile of the import and export energies was smoothed and flattened, as compared to the cases without storage and/or using a conventional controller. With the proposed approach, the annual aggregated grid usage was reduced by 53% and the building’s annual energy needs were covered by the renewable energy system at a rate of 60%. It was therefore concluded that the proposed hybrid methodology can provide a tool to maximize the autonomy of nearly-zero-energy buildings and bring them a step closer to implementation.

Sign in / Sign up

Export Citation Format

Share Document