# Forecasting of natural gas consumption in China’s logistics industry based on semi-hierarchical control

## TL;DR

This paper studies how China's logistics industry might shift from oil to natural gas, using forecasting methods to predict future consumption trends.

## Contribution

The novel contribution is the application of semi-hierarchical control and combined forecasting methods to improve natural gas consumption predictions in the logistics industry.

## Key findings

- Natural gas is expected to replace oil in China's logistics industry due to policy support and lower costs.
- Combined forecasting methods show higher stability and lower error rates compared to individual forecasting models.
- Forecasting results help government departments estimate natural gas consumption trends under different market conditions.

## Abstract

This paper proposes a research framework based on semi-hierarchical control, analyzes the mechanism of gas instead of oil in China’s logistics industry, and uses several forecasting methods to forecast. The research findings include that: (1) the driving mechanism of substitution of natural gas for gasoline and diesel indicates that natural gas is encouraged by China’s policies and the cost of use is lower, China’s logistics industry will reduce its dependence on gasoline and diesel. (2) By using grey forecasting method, regression trend method and Bass model to forecast natural gas consumption in logistics industry, they show that the forecasting results under different circumstances are helpful for China’s government departments to estimate the consumption trend of natural gas in logistics industry according to different market environments. (3) Based on the reverse feedback mechanism of semi -hierarchical control, combined forecasting methods are established, the hard problem that the combined forecasting coefficients are also solved. Combined forecasting methods are useful complements to meet the forecasting demands of logistics industry’s natural gas consumption, and further improve the forecasting accuracy. (4) According to mean relative error, the error percentages of grey forecasting, regression trend method and Bass model are respectively in 5.373%, 2.9%, and 5.94%, the error percentages of combined forecasting methods are within 2.9%−3.1%, the combined forecasting methods have better forecasting stability.