Short-term load forecasting of urban gas using a hybrid model based on improved fruit fly optimization algorithm and support vector machine. (November 2019)

• Record Type: Journal Article
• Authors: Lu, Hongfang; Azimi, Mohammadamin; Iseley, Tom
• Is Part Of: Energy reports. Volume 5(2019)
• Page Start: 666
• Page End: 677
• Publication Date: 2019-11
• Subjects: Short-term -- Gas load -- Forecasting -- Urban gas -- Fruit fly optimization algorithm -- Support vector machine; Power resources -- Periodicals; Energy industries -- Periodicals; Power resources; Periodicals; Electronic journals; 621.04205
• DOI: 10.1016/j.egyr.2019.06.003
• Languages: English
• ISSNs: 2352-4847

Abstract

Abstract: The accurate forecasting of short-term load for urban gas is the premise of gas supply sales, pipe network planning, and energy optimization scheduling. This paper adopts a hybrid model that integrates the fruit fly optimization algorithm (FFOA), simulated annealing algorithm (SA), cross factor (CF) and support vector machine (SVM) to forecast the short-term gas load of urban gas. In the model, SA and CF are used to optimize the FFOA algorithm. This paper takes the urban gas system in Kunming, China as a case study and uses the CF-SA-FFOA-SVM algorithm to predict the gas consumption and compares the results with the other four forecasting methods such as back-propagation neural network (BPNN) and autoregressive integrated moving average model (ARIMA). Besides, this paper analyzes the influence of temperature types (daily maximum temperature and daily mean temperature) in models on forecasting results, and the applicability of the algorithm for forecasting weekly and monthly gas load is analyzed. Moreover, the impact of grouping raw data by the feature on forecasting result is discussed. The following conclusions are drawn: (1) compared with other forecasting models, CF-SA-FFOA-SVM model has higher gas load forecasting accuracy. (2) for Kunming city, if the daily maximum temperature is used as the input variable in the gas load forecasting model, the forecasting accuracy is higher. (3) grouping raw data according to holiday attributes or gas types can effectively improve the accuracy of load forecasting. Highlights: A hybrid algorithm CF-SA-FFOA-SVM is used to forecast the short-term urban gas load. The influence of temperature types in forecasting models on forecasting results is discussed. The applicability of the algorithm for forecasting daily, weekly and monthly gas load is analyzed. The effect of data grouping-based forecasting method is analyzed.