A study on fast and stable power conversion control with hybrid model predictive control and machine learning

2021 Fiscal Year Final Research Report

• Project/Area Number: 19K04355
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Review Section: Basic Section 21010:Power engineering-related
• Research Institution: Nagasaki University
• Principal Investigator: Maruta Hidenori 長崎大学, 工学研究科, 准教授 (00363474)
• Project Period (FY): 2019-04-01 – 2022-03-31
• Keywords: DC-DCコンバータ

Outline of Final Research Achievements

In this research, we developed a novel control method using model predictive control and machine learning control for power converters which is treated and modeled as a hybrid system. Model predictive control can realize flexible functions under some constrained conditions, however, it requires heavy computation burden for its optimization problem. We addressed it by developing combinatorial enumeration-based computation and dynamic quantization to obtain pseudo optimal solution. In addition, we also adopted a neural network control to improve transient characteristics combined with model predictive control. The proposed method can obtain superior characteristics both in steady state and transient state simultaneously.

Free Research Field

電力変換器

Academic Significance and Societal Importance of the Research Achievements

電力変換器の制御において、柔軟な制御手法であるモデル予測制御とニューラルネットワーク制御を組み合わせることで、安定性と応答性を両立することが可能な手法を開発した。また、提案手法では、制御における演算量を考慮し、制御器の性能に応じて演算コストと制御精度のトレードオフが可能であるため、比較的容易に提案手法を導入することが可能となることが期待できる。