Cloud-Based Battery Monitoring and State of Charge Estimation Platform for 48V Battery Systems

摘要

With the rapid advances in the energy storage technologies and the drop in price, the battery has emerged as one of the most important energy storage systems in stationary and mobile applications. However, without a proper monitoring and control of the batteries by a Battery Management System (BMS) several problems with regard to safety, reliability, durability and cost will appear. As a result of the fast development and application of Internet of Things (IoT) and cloud computing technologies, BMS can be revolutionized to further improve the computational power and system reliability, implement advanced battery diagnostic algorithms and achieve data visualization. In this work, a Cloud BMS for 48V battery systems was developed and validated. The designed platform consists of an in-house designed BMS-Slave board with battery sensors, a single-board computer (Raspberry Pi 3), a cloud database, a User Interface (UI) and an Application Programming Interface (API). The BMS-Slave board performs the data acquisition by measuring the voltage, current, temperature and pulse resistances of the batteries at given sampling time. It sends the data based on the Controller Area Network (CAN) protocol to the Raspberry Pi, which serves as the IoT communication component in the platform, and can be connected to a display or be accessed remotely. The Raspberry Pi is responsible for translating the CAN-signals into digital data and sending the data to the cloud based on the TCP/IP protocol and Message Queuing Telemetry Transport (MQTT) protocol with assured security. The users of the battery system can monitor and control the states of each battery by using the UI and the API of the cloud database, which are supported technically by the aedifion GmbH. As one of the benefits of the cloud BMS, advanced battery diagnostic algorithms were applied in the cloud, ensuring more accurate monitoring and more intelligent control of the battery systems. The battery model used in this work is an improved Thevenin model considering the polarization characteristics of the battery by adding two RC elements. The model parameters such as the internal resistances and the equivalent capacitances were identified using Particle Swarm Optimization (PSO) algorithm based on the current and voltage measurement data under dynamic load profile. The adaptive H infinity filter based state of charge estimation algorithm was implemented, which is able to adjust its process and measurement noise matrices based on the covariance matching technique. A test bench was set up to validate the Cloud BMS for a 48V battery system, which is composed of four 12V, 7Ah Yuasa AGM lead-acid batteries connected in series. The functionality, accuracy and robustness of the cloud-based battery monitoring system, PSO based parameter estimator and adaptive H infinity filter based SoC estimator were evaluated under a dynamic load profile. This poster will present the structure and technical details of the developed cloud-based battery management system and the conducted validation results. This research is kindly funded by the European Regional Development Fund NRW under the proj ect number EU-1-1-0 81B.