Intelligent prediction of gas consumption in three-hour intervals, using data recorded in specific time periods

Gas consumption, as one of the most consumed energy carriers in the country, is very important in crisis management, production management, allocation and consumption, preventing wastage and reducing environmental pollution. However, due to the oldness of the measuring equipment, it is not possible to record data continuously, and the available information about gas consumption is usually limited to the periodic gas consumption reports of consumers. Therefore, real-time forecasting of gas consumption is associated with high uncertainty, and it is not possible to check the accuracy of the forecast due to the lack of real-time data. The aim of this research is to intelligently predict gas consumption in three-hour intervals based on its periodic consumption information. The method presented in this research includes three steps.In the first step, using the fuzzy clustering algorithm, the gas consumption data of household consumers are divided into three categories: low consumption, balanced and high consumption. In the second step, using a deep neural network, the temperature is predicted in three-hour intervals. In the third step, using a system based on fuzzy logic, the three-hour gas consumption of household consumers is estimated based on the predicted temperature, day of the year and time of day. The implementation of the proposed method on the data of the city of Birjand shows that the total instantaneous consumption predicted for all three low-consumption, balanced and high-consumption clusters follows the average periodic consumption of the consumers of those clusters with an acceptable error. Also, the presence of a strong negative correlation between temperature and gas consumption, especially for balanced consumers in cold seasons, confirms the influence of gas consumption on temperature.