CASPIAN JOURNAL
MANAGEMENT AND HIGH TECHNOLOGIES
EVOLUTIONAL CONNECTIONISTS MODELS FOR IDENTIFICATION OF SYSTEMS DYNAMICS: IMPLEMENTING IN AUTOMATION ENERGY FORECASTING
 |
Read | Shcherbakov Maxim V., Kozlov Ilya P., Shcherbakova Natalia L. EVOLUTIONAL CONNECTIONISTS MODELS FOR IDENTIFICATION OF SYSTEMS DYNAMICS: IMPLEMENTING IN AUTOMATION ENERGY FORECASTING // Caspian journal : management and high technologies. — 2011. — №4. — pp. 70-75. |
Shcherbakov Maxim V. - Cand. in Technics, Volgograd State Technical University, 28 Lenin avenue, Volgograd, 400131, Russia, maxim.shcherbakov@gmail.com.
Kozlov Ilya P. - Master, Volgograd State Technical University, 28 Lenin avenue, Volgograd, 400131, Russia, benkyo.nanodesu@gmail.com.
Shcherbakova Natalia L. - Cand. in Technics, Volgograd State Technical University, 28 Lenin avenue, Volgograd, 400131, Russia, snl@gebeus.ru.
The article considers the problem of energy consumption forecasting. Often there is no possibility to create energy consumption forecast in time by expert. The main reason is large number of objects to be analyzed. In this case, the urgent issue develops adequate and accurate predictive techniques able to work in automatic mode. The original method base on classifying of object’s dynamics within short interval of observation has been applied. The method uses the cyclical property of the objects, classifies and predicts their behavior within short interval of observation. There are two main phases: the phase of synthesis and adjusting the models and the prediction or verification phase. At the first phase the behavior classifying and forecasting models and forecasting models for each class are created. Also the ensemble of the models could be created. To demonstrate the effectiveness of the method, the cases of commercial buildings energy consumption forecasting were performed. The results of the different methods were compared. The forecast error decreased in 1,4–3,1 times for the suggested method.
Key words: energy consumption forecasting,evolutional connectionists systems,hybrid models,behavior identification