MANAGDMENT OF A LOCAL URBAN HEAT SUPPLY SYSTEM BASED ON NEURAL NETWORK MODELING TAKING INTO ACCOUNT THE WEATHER FORECAST
DOI:
https://doi.org/10.14529/ctcr200303Keywords:
heating network, forecast management, intelligent system, neural network modeling, deep leaning neural networks, temperature control of boiler, inertia of the heating networkAbstract
The article deals with the problem of managing the heat supply of consumers, taking into account the weather forecast using a neural network. The monitoring subsystem was used to obtain statistical data that was used to train the neural network. Optimal control of the temperature of the coolant allows you to save fuel especially effectively, in case of rapid changes in weather conditions. The expected fuel economy reaches 5–15 % depending on the air temperature in the season and the state of the heating network.
Aim. The purpose of the research is to develop an intelligent module for the automated control system “Aurora Heat balance in GCH”. The intelligent module allows you to automatically adjust the water temperature in network, taking into account the weather forecast and when the mandatory temperature limits for consumers are met.
Materials and methods. Artificial neural network is considered as the main tool that minimizes errors in manual control of the boiler room temperature. A neural network in the form of a multi-layer perceptron and a deep learning LSTM procedure were used. This made it possible to predict the temperature of the coolant taking into account the inertia of the network and the forecast of air temperature. To protect the model from overtraining, the Dropout method was used with a probability of 0.2.
Results. The possibilities of neural networks to predict the optimal heating temperature of the boiler are investigated. This temperature is calculated taking into account restrictions for heat consumers and taking into account the forecast air temperature. The application of a neural network model in form of a multilayer perceptron is justified. A demonstration example of using an intelligent information system for heat supply management is presented.
Conclusion. The proposed methods and models are tested on real data. This confirms the possibility of their use in the development of intelligent information systems for heat supply management.References
Белоусов, О.А. Интеллектуальная система управления и мониторинга газовой котельной / О.А. Белоусов, С.В. Иванов // Программные продукты и системы. – 2012. – № 1. – https://cyberleninka.ru/article/n/intellektualnaya-sistema-upravleniya-i-monitoringa-gazovoy-kotelnoy.
Регулирование температуры теплоносителя. Энергетика простыми словами. – https://energoworld.ru/blog/regulirovanie-temperatury-teplonositelya.
Панферов, В.И. Об оптимальном управлении температурой теплоносителя в тепловых сетях / В.И. Панферов, О.Ф. Гавей // Вестник ЮУрГУ. Серия «Компьютерные технологии, управление, радиоэлектроника». – 2014. – № 4. – https://cyberleninka.ru/article/n/ob-optimalnom-upravlenii-temperaturoy-teplonositelya-v-teplovyh-setyah.
Ясницкий, Л.Н. Интеллектуальные системы / Л.Н. Ясницкий. – М.: Лаборатория знаний, 2016. – 221 с.
Истомин, Д.А. Экспертная система оценки технического состояния узлов электроцентробежного насоса на основе продукционного представления знаний и нечеткой логики / Д.А. Истомин, В.Ю. Столбов // Вестник ЮУрГУ. Серия «Компьютерные технологии, управление, радиоэлектроника». – 2020. – Т .20, № 1. – С. 133–143.
Комаров, Н.М. Управление инженерными системами интеллектуального здания с использованием информационного и инфографического моделирования / Н.М. Комаров, В.Г. Жаров // СЕРВИС Plus. – 2013. – № 2. – С. 74–81.
https://www.link-labs.com/blog/smart-hvac.
Ki Uhn Ahn, Cheol Soo Park. Application of deep Q-networks for model-free optimal control balancing between different HVAC systems // Science and Technology for the Built Environment. – 2020, Vol. 26, iss. 1, P. 61–74.
A deep reinforcement learning-based autonomous ventilation control system for smart indoor air quality management in a subway station / S. Heo, K.-J. Nam, J. Loy-Benitez et al. // Energy & Buildings. – 2019. – Vol. 202.
Sepp Hochreiter, Jürgen Schmidhuber. Long short-term memory // Neural Computation: journal. – 1997. – Vol. 9, no. 8. – P. 1735–1780.
Переобучение в статистике и машинном обучении. – https://ru.wikipedia.org/wi-ki/%D0%9F%D0%B5%D1%80%D0%B5%D0%BE%D0%B1%D1%83%D1%87%D0%B5%D0%BD%D0%B8%D0%B5 (дата обращения: 22.05.2020).
Dropout – метод решения проблемы переобучения в нейронных сетях. – https://habr.com/ru/company/wunderfund/blog/330814 (дата обращения: 22.05.2020).
Kingma, D.P. Adam: A Method for Stochastic Optimization / D.P. Kingma, J.L. Ba. arXiv:1412.6980 2014.
Некоторые оконные функции и их параметры. – http://www.dsplib.ru/content/win-add/win.html (дата обращения: 02.06.2020).
