We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
Synthesis of intelligent power management system of food manufacturing processes with power consumption prediction.
- Authors
Baliuta, Serhii; Kopylova, Liudmyla; Kuevda, Valerii; Kuievda, Iuliia; Lytvyn, Iryna
- Abstract
Introduction. The study is done to justify the methods of intelligent power management system of food manufacturing processes (FMP) devoted to increase the effectiveness of electric power use. Materials and methods. The study is based on intelligent control methods, power consumption prediction algorithm using artificial neural networks and active identification method of load static load characteristics. Results and discussion. Based on the system analysis of FMP control it is determined control criteria and functions. To implement the control functions in intelligent power management system of FMP it is used: model of predicting power consumption, making decision algorithms of power management, procedure of forming the optimal list of consumers-regulators and calculating the rational modes of power consumption. To predict the power consumption of food manufacturing process the multilayer perceptron is chosen from the set of artificial neural network architectures. It is shown that the optimal configuration of the neural network in the task is three layered perceptron with hidden layer where the number of elements equals to the half-sum of the elements of the input and output layers. Perceptron training is carried out by combined backpropagation/Cauchy machine method. The computational experiment with prediction of power consumption of food manufacturing process on the next year has the learning error in the range of 0.05-0.06. The algorithm of determining the optimal voltage, which provides energy efficient operating modes of power system, uses static load characteristics. To make the algorithm more accurate and effective it is used procedure of static load characteristics identification in the interactive mode for the main modes of the technological process, taking into account the state of consumers-regulators and the degree of compensation of reactive power with the help of transformer equipped with the electronic switch. Conclusions. The algorithms of intelligent power management system of FMP become more efficient with the use of power consumption prediction and static load characteristics identification algorithms.
- Subjects
MANUFACTURING processes; MULTILAYER perceptrons; FORECASTING; COMPUTER performance; DEAD loads (Mechanics); FOOD industry; RADIAL distribution function; LOAD forecasting (Electric power systems)
- Publication
Ukrainian Journal of Food Science, 2020, Vol 8, Issue 1, p105
- ISSN
2310-1008
- Publication type
Article
- DOI
10.24263/2310-1008-2020-8-1-11