Research on the current frequency forecasting of a power supply converter for heating the oil pipeline based on gradient boosting decision tree

Abstract

When large-sized oil pipelines are subjected to skin effect heating under the constraint of limited current, the power supply converter must provide a higher current frequency. However, an inappropriate frequency can lead to increased energy losses and consequently reduce heating efficiency. To achieve higher heating efficiency of large-sized pipelines, this paper proposes a temperature control strategy based on optimal frequency forecast. First, a skin effect heating model is established to illustrate the necessity of frequency forecasting. Then, using the small sample data from the oil field, an optimal frequency forecasting method based on Gradient Boosting Decision Tree (GBDT) is proposed. At the forecasted frequency, a dual closed-loop control strategy for current and temperature, based on a designed three-phase converter, is employed to achieve temperature control under limited current conditions. The feasibility and effectiveness of this approach are demonstrated through a case study involving a 159 mm oil pipeline with a heating current of less than 80A. The experiment shows that with the forecasted frequency, the pipeline temperature reached 40 °C within 14 min, and the energy consumption per unit length of the pipeline is 394 W/m, complying with the standard Q/SY 06022-2016.

Keywords

• Frequency forecast
• Gradient Boosting Decision Tree (GBDT)
• Oil pipeline
• Power supply converter
• Skin effect heating
• Small sample data