If any part of this thesis has been previously submitted for a degree or any other qualification at this University or any other institution, this has been clearly stated. Therefore, in a series-connected battery pack, balancing the voltage between the battery cells is essential to protect the cells and extend the life of the battery. Franklin Bien for his constant support of my master's studies and research, for his patience, motivation, enthusiasm and immense knowledge.
His guidance helped me throughout the research and writing of my thesis.
Electric Vehicles and Battery Management System
This data includes cell voltages, battery pack input and output current, and pack environmental temperatures. In addition, the processing unit makes a decision that is applied to the battery pack via the BCU control unit.
Lithium-ion Battery Model
The battery remains in this phase for approx. 3 hours (it can be varied depending on each type of batteries). When the battery reaches the maximum threshold voltage (determined by the treatment of battery and aging level), the battery is fully charged.
Motivation of Battery Equalization
This Figure 1.6 shows the burned auxiliary power battery of a Boeing 787 that caught fire at Boston's Logan International Airport in January 2007. The discharge phase is interrupted when, for example, the battery is discharged to a lower value of 3.0 V/cell, so that the current is stopped. Because if the discharge phase remains below 2.7 V/cell, the copper shunts may have formed in the cells and led to a partial or total electrical short circuit.
Therefore, the cell must be put into sleep mode when overdischarged, to avoid permanently damaging a battery [5–7]. However, the excess energy from the higher voltage cells is transferred to heat which leads to energy waste and shortens battery life. Energy recovery is the solution to overcome the disadvantage of additional losses in the previous methods.
In recent times, integrated individual cell equalizer (ICE) based on DC-DC converter is under development for a series-connected battery pack, so that implementation and individual control of equalization can be easily achieved. However, the complex control of DC-DC converters for a large number of cells is one of the problems. Therefore, in the ICE-based equalization scheme, the control circuit is one important part that contributes efficiency to the performance of battery equalization.
Passive Cell-balancing Methods
Second, the control methods for these topologies are discussed, since the equalization controller is an important part of BMS. The detection mode requires the higher level of control circuitry to determine where the cells energy should be routed. This method is more efficient than the continuous mode, but requires higher and more difficult control circuit than the continuous mode.
Active Cell-balancing Methods
In this way, the energy between the cells of the pack is transferred in order to balance. The main disadvantage of this method is the long equalization time of the energy transfer from the first cell to the last cell in the string battery pack. Therefore, the current from the cell stack passes through the primary transformer and induces currents in the secondary transformers.
The design of multi-wind transformer must be adjusted according to the number of cells, which limits modularization. Most of the induced current will be supplied via the diodes to the cells with the lowest voltage. In forward structure, when the voltage deviation is detected, the switch connected to the cell with the highest voltage is turned on, so that the energy of this cell is transferred to others via the transformer and the anti-parallel diodes of the switch [ 23 ].
The bidirectional Cukb converter [5,12] as shown in 2.7-a), which balances each pair of the neighboring cells. The boost converter used to remove the excess energy from single cell to the total pack, or buck boost converter used for removing excess energy from the highest cells to the DC link, storage element, and retransmission of the weaker cells. In this scheme, the charge moves from the most charged battery to the least charged battery through a DC switching capacitor through flyback operations.
Cell Equalization Control Methods
Consequently, the discharge state of the battery should not drop below its low voltage threshold, normally set in the range of 2.6-2.8V. Due to series connection between battery cells, the imbalance between cells is caused by changes in the internal impedance and cell capacity. Although fuzzy logic has advantages to model nonlinear system, the previous control methods based on fuzzy logic [10–13] require the knowledge and experience of the battery cells.
The following section represents the proposed system of cell equalization for BMS consisting of hundreds of Li-ion battery cells. In the proposed system, the Cbuk DC-DC converter topology is used, because of its simple circuits and small size without transformers. As mentioned before, in the conventional schemes of cell balancing, it is difficult to handle the battery cell model for describing the equalizing property of the Li-ion battery strings due to electrochemical reactions and ambient temperature.
In the proposed system, a feedback loop is provided to enable updating and tracking the changes of battery strings in online measurements. In the proposed system, the balancing currents are controlled by driving a pulse width modulation (PWM) signal corresponding to the control signals constructed in the analog adaptive neuron fuzzy control parts. ANFIS learns the features in the dataset and adjusts the system parameters according to a certain error criterion [25,26].
Equalizing Circuit
Off-line Learning with Adaptive Neuro Fuzzy Con- trollertroller
There is a combination of the least squares method and the backpropagation gradient descent method for training the adaptive parameters of the network [25]. This structure can adjust its predecessor and successor parameters to improve system performance by observing the results. On the other hand, the shapes of activation functions can also be shaped and adjusted in the trianization process [27].
The idea of the back-propagation algorithm is to reduce this error, until the ANN learns the training data. This function describes the error the neural network makes in approximating or classifying the training data, as a function of the network's weights. The backpropagation weight update is equal to the slope of the energy function further scaled by .
We have taken the squared error energy so that the errors of opposite signs do not cancel each other; also the size of the error and not in the error sign could be used. The goal of the learning process is to adjust the ANN parameters (synaptic weights) to minimize the total error energy. Weights are updated pattern by pattern until complete set of training data is used (one epoch) for training the network.
Analog Adaptive Neuron-Fuzzy Control Implementa- tiontion
First, these input training samples pass through layers 1-3 of the ANFIS algorithm, and then LSE optimizes the error square of the output samples, compared to the desired output values. The parameter set of {api, dpi} is updated from the learning process in the microprocessor as presented in the previous section. These output current singleton values are determined after the ANFIS learning process.
The relationships between current drain and gate-source voltage in weak and strong inversion are expressed in the following. The approach to implement the current state multiplier/divider is a combination of a geometric mean circuit and a quadrature/divider circuit as described in 3.13. The concept in 3.13 could be designed by forcing a proper transition from triode region to saturation region of the MOSFET [28].
In the half circuit on the left (including a-b MOSFET devices), the drain-source voltage of transistors M2A and M2B is calculated as follows: 3.23c) For the larger, when IX < IY, M2A will operate in the triode region while M2B is in the saturation region. The figure 3.14 shows the result of the current mode divider implementation with IW = 5µA, IX = 10µWhile IY varies from (0∼20µA). In the proposed system, the voltage sensor block is composed of voltage sensors for battery cells, ANFIS learning algorithm is composed with a microprocessor, and the AANF block is designed in a 0.13−µm CMOS technology with power supply of 2.5V.
Environment Setup and Analysis
Experimental Results
Finally, it causes battery system malfunctions or slows down the equalization time due to lack of tracking and learning ability. The experimental results for the cell voltage trajectories during equalization for the proposed equalization scheme are shown in the following figure. The experimental results show that the proposed equalization scheme, implemented in a low-power microcontroller, reaches the equalization state after about 2500 seconds, which corresponds to the derivation of more than 1V.
Discussion
REPLACE TH
HIS LINE WITH
H YOUR PAP
PER IDENTIFI
ICATION NUM
5) out
ANFIS that is
MBER (DOUB
CD T play
BLE-CLICK H
C ONCLUS
HERE TO EDIT
The DC-DC converter controller for the energy system is a promising area not only in building management systems. The next section presents the circuit design and operation of the proposed dc-dc converter and circuit controller based on fuzzy logic theory. The DC-DC converter is used to transfer the original low DC voltage to a higher usable level (3V).
However, most recent RF power harvesters require a voltage monitor controller and dc-dc converter controller, which consume much of the harvested power. The digital controller must control the input resistance of the dc-dc converter to give a maximum output power. Therefore, the power manager including dc-dc converter and control circuit is one of the limits of current practices in RF power harvester design.
Therefore, the solution could be a soft control circuit for the DC converter, which consumes little power and has a high response speed compared to conventional digital controllers, as shown in 5.1. Since the input voltage collected from many sources can be in a wide range, the soft control circuit is designed to drive the DC-DC conversion to achieve the highest energy efficiency. A soft control single cell equalizer using a pulse converter with discontinuous induction current mode for Li-ion chemistries.
