가변속 냉동사이클 기반 배터리 열관리 시스템의 모델예측 제어기 설계

Abstract

The introduction and technological advancements of eco-friendly vehicles are increasingly recognized as effective alternatives to mitigate climate change. Electric vehicles (EVs), powered by electric motors, boast minimal or virtually zero exhaust emissions during operation. To drive the electric motors, batteries are essential for storing and supplying power. However, these batteries must be carefully managed to prevent overheating-related hazards such as explosions or fires. Additionally, battery performance is closely tied to operating temperatures, and deviation from the optimal temperature range can lead to reduced lifespan and decreased energy efficiency due to a decrease in charging capacity. To maintain the safety and performance of batteries, extensive research is currently being conducted on Battery Thermal Management Systems (BTMS) employing various cooling methods. Particularly, the variable-speed refrigeration cycle has proven effective in responding to a wide range of partial loads, offering high energy efficiency and rapid, precise temperature control during load fluctuations. This technology finds applications in diverse fields such as building HVAC systems, semiconductor processes, and commercial-industrial refrigeration. The variable-speed refrigeration cycle controls the temperature by regulating the compressor's rotation speed and the Electronic Expansion Valve (EEV), concurrently managing refrigerant superheat. In cases of indirect cooling, the cycle involves the circulation of secondary refrigerant, such as Brine, for cooling purposes. To facilitate this, a cooling water pump circulates coolant, absorbing heat from the coolant material, and maintains the temperature within an appropriate range. The Electric Water Pump (EWP) plays a vital role in this process, acting as a crucial control element alongside core components like the compressor, condenser, EEV, and evaporator. While existing research has investigated the control of coolant flow using rule-based approaches based on system characteristic analyses, which set the operating rotation speed of devices like EWP, these methods may face challenges in achieving energy optimization when multiple control variables coexist. Therefore, this study introduces the Model Predictive Control (MPC) technique to comprehensively regulate the coolant flow of EWP in conjunction with a variable-speed refrigeration cycle composed of a variable-speed compressor and EEV. MPC utilizes state feedback and system models to predict future states, concurrently configuring predicted control variables into the objective function to derive optimal control inputs. Consequently, MPC, designed as a multi-variable controller for Multi-Input Multi-Output (MIMO) systems, facilitates integrated control of various parameters. In this process, MPC assigns individual weights to each control variable, enabling swift and flexible adaptation to diverse performance requirements. Furthermore, MPC allows for the anticipation of control device constraints during the optimization process, ensuring control inputs stay within the permissible maximum output range of actuators and guaranteeing optimal control performance. This study applies MPC to design and assess its control performance for an Oil Cooler System (OCS) and BTMS configured with a variable-speed refrigeration cycle. Initially, the feasibility of MPC design for OCS is verified, followed by the application of MPC to BTMS with EWP control. The performance of MPC in regulating battery module temperature, coolant temperature, and refrigerant superheat is thoroughly examined. The models for OCS and BTMS, essential for MPC design, are derived from continuous-time state-space models converted from transfer functions obtained through dynamic experiments. Subsequently, the continuous-time BTMS model is discretized, leading to an augmented model expressed in discrete-time state-space form. Based on this augmented model, MPC is designed in the MATLAB/Simulink environment, and simulations are conducted to validate the proposed design. The impact of key MPC design parameters, namely prediction horizon and control horizon values, is extensively analyzed through simulations.