BATTERY BALANCING SYSTEM FOR ELECTRIC VEHICLE WITH MULTI-LAYER INDUCTOR TOPOLOGY USING SUPERVISED LEARNING AND TYPE-2 FUZZY LOGIC SYSTEM ALGORITHM

• Main Author: Hartono
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/79309
• Institution: Institut Teknologi Bandung

Lithium-ion batteries are a type of battery widely used as energy storage devices in electric vehicles due to their high energy density, long lifespan, low self-discharge, and environmental friendliness. Because the energy storage capacity of each battery cell is limited, in order to fulfill the voltage and power requirements for electric vehicles, the battery cells must be connected in series and parallel. Using batteries in series in large numbers can potentially lead to problems due to differences in cell capacity, discharge rate, internal resistance, and lifespan caused by variations during the manufacturing process and chemical reactions in each battery cell, resulting in excessive battery cell charge imbalances. Cell charge imbalances can cause the charging and discharging processes to stop immediately, reducing total capacity, decreasing performance, and affecting the battery pack's lifespan. Various efforts are being made through the development of battery balancing systems to overcome this issue. In general, battery balancing systems can be categorized into passive balancing systems and active balancing systems. Based on the components used, active balancing systems are divided into three types: capacitor-based, inductor-based, and transformer-based balancing systems. Inductor-based balancing systems offer several advantages, such as good balancing speed and efficiency, compact dimensions and weight, and affordable pricing. Previous research has developed inductor-based battery balancing systems using conventional control with fixed PWM parameters. Meanwhile, others have conducted development by adjusting the duty cycle based on cell voltage conditions using fuzzy logic control, which is an intuitive but non-robust intelligent controller that cannot accommodate uncertainty. In this study, a battery balancing system was developed with a control strategy that combines a directional differentiation algorithm and a type-2 fuzzy logic system to set the values of the PWM signal parameters used to turn on and off MOSFET on a multi-layer inductor-based balancing series for eight lithium-ion battery cells assembled in series. The input used by the controller is the difference between the maximum cell voltage and the average battery package voltage, while the output from the directed learning algorithm is the switching frequency, and the fuzzy logic output type-2 is the duty cycle value of the PWM signal. At the experimental test, initially the initial voltage of each battery cell was set to different values in the top-down scenario. The test is carried out for 9000 seconds under static conditions for each control method, assuming the battery pack reaches equal conditions when the cell voltage difference is not greater than 50 mV. Data from the voltage measurement during the intercell, intergroup, and intermodular balancing processes are presented in the balancing chart. Then, a comparison was made between the performance of the system using the proposed control method and the conventional control method. Tests on dynamic conditions are also carried out to ensure that the developed system can balance the battery under charging and operating conditions. Based on the experimental results, the designed control method can increase the balancing speed by 17.3% as well as reduce the voltage gap between cells, groups, and modules, which is larger than conventional control methods using fixed PWM parameters. The balancing current with the conventional method was 221.10 mA, while the balancing current with the proposed method reached 257.21 mA. Furthermore, using the designed method, the number of switching can be reduced by 6.2%, thereby reducing the potential degradation effects and prolonging the life of the components. The efficiency of the balancing system with the developed method is 87.05%; there is no significant difference in efficiency with conventional methods. ?