Search

Due to the low switching loss, compact size, and low price, the LC series resonant converter (LCSRC) as an energy carrier is promising in battery equalizers. However, the conventional equalizers based on LCSRC only can realize the one-cell-to-one-cell balancing for a high-voltage series-connected battery pack, causing a low equalizing speed and efficiency. Therefore, two direct multicell-to-multicell (MC2MC) balancing topologies based on matrix LC converter and half-bridge LC converter are proposed, respectively, which achieve the direct energy transfer from a consecutive more-charged-cell group to a consecutive less-charged-cell group, significantly improving the equalizing speed and efficiency. An analytical model for the LCSRC is developed to provide a guidance for the optimal design of the proposed equalizers. The prototypes for the proposed two balancing topologies are constructed. The MC2MC balancing performances are verified though the comparative experimental results at different initial voltage distributions. In particular, the proposed equalizers simultaneously achieve the fast balancing speed and high equalization efficiency. The measured peak efficiency is about 87.9% at the four-cell-to-four-cell equalization.

Capacity is one of the most critical parameters of lithium-ion batteries (LIBs). Retired batteries generally also have 70%-80% of the nominal capacity, so it can bring considerable benefits through echelon use. However, the existing methods cannot simultaneously meet the capacity estimation accuracy and efficiency required for large-scale retired batteries. So this may lead to costing excessive time and energy. To solve the above problems, a fast and accurate capacity evaluation method based on support vector machine (SVM) is proposed in this paper. And the parameters (penalty coefficient and kernel function width) of the SVM model are optimized by the particle swarm optimization (PSO). In order to expand the scope of application, capacity evaluation models in three different SOC are established. The inputs of the model are sampling voltages selected from the featured charging curve at three cases, and the output is the battery capacity. Experimental results demonstrate that the maximum error in the first case is less than 3.02%, and the errors in the other two cases are 1.88% and 1.97% respectively..

Battery characterization data is the basis for battery modeling and state estimation. It is generally believed that the higher the sampling frequency, the finer the data, and the higher the model and state estimation accuracy. Several different kinds of battery models have been proposed in literature, which can be generally classified into white-box models, gray-box models, and black-box models. However, scientific selection strategy for sampling frequency is very important but rarely studied. This paper studies the influence of sampling frequency on the accuracy of battery model and state estimation under four different sampling frequencies: 0.2 Hz, 1 Hz, 2 Hz, and 10 Hz. Then, a function is proposed to depict the relationship between accuracy and sampling frequency, which shows an optimal selection principle. The iterative identification algorithm is presented to identify the model parameters, and state-of-charge (SOC) is estimated via extended Kalman filter algorithm. Experimental results with different operating conditions clearly show the relationship between sampling frequency, accuracy, and data quantity, and the proposed selection strategy has high practical value and universality. The tradeoff strategy between the sampling frequency and the model accuracy has high applicability and universality. In future works, the voltage data of battery pack with cells connected in series and parallel at different sampling frequencies will be obtained, and the effect of voltage mismatch on the necessary sampling frequency will be studied.

Accurate power prediction for a battery pack is of great importance for efficient and safe operation of electric vehicles. However, a battery pack typically consists of dozens of cells connected in series, and how to accurately predict its peak power in practical applications has always been a challenging issue. This paper proposes a low-complexity peak power prediction method for a series-connected battery pack, where the peak power of battery pack is predicted depending only on representative cells. Firstly, considering state of charge and voltage limits for power prediction, the representative cells are selected by using two easily available variables, namely characteristic voltage and ohmic resistance. Secondly, the state of charge and voltage of the representative cell are estimated accurately by the dual adaptive extended Kalman filter algorithm. Finally, the multi-parameter limited method for battery pack is developed to predict the peak power under Urban Dynamometer Driving Schedule test at different temperatures. The experimental results verify the feasibility and robustness of the proposed selection method, and the peak power of battery pack is predicted with low complexity and high accuracy.

For the safe operation of electric vehicles, it is of critical importance to quickly detect and accurately identify different types of faults in battery packs. However, the performance characteristics of many faults in the battery system are hidden and similar, therefore a false positive fault detection happens occasionally. This paper presents a multi-fault diagnostic strategy based on an interleaved voltage measurement topology and improved correlation coefficient method, which can diagnose several types of faults (i.e. the internal/external short circuit, sensor faults and connection faults). The proposed voltage measurement method can correlate each battery and contact resistance with two different sensors respectively, so as to accurately identify the location and type of the faults. In order to eliminate the effect of battery inconsistencies and measurement error, the improved correlation coefficient method is utilized to monitor fault signatures. The non-model method proposed in this paper can avoid the aliasing phenomenon and has high sensitivity and robustness. Both theoretical analysis and experimental results validate the feasibility and advantages of the multi-fault diagnostic method.

With a big mass of lithium-ion batteries retired from electric vehicles, researches on battery secondary use are rather urgent and important. However, existing screening researches cannot ensure high accuracy and high efficiency meanwhile. To solve the shortcomings, an accurate and fast screening approach on the basis of Neural Network is proposed in this work. The specific partial charging curves of battery cells are decided by incremental capacity analysis for its ability to detect battery aging state. Then critical features extraction is executed based on partial charging curves. They are used as the input of the screening Neural Network model and help finish the model training. The results on 234 LiFePO4 battery cells proves the validity of the method. The screening accuracy can be high as 97.4%. As batteries count climbed to 5000, the screening efficiency can be 9 times higher than traditional capacity testing approach.

The power battery system is the core for electric vehicles, which is composed by hundreds or thousands of cells in series and parallel generally. Cells have inherent inconsistency because of slight deviations and uncertain factors in their production process. And they work in the complicated charge and discharge environment for a long time, so their inconsistency will be further aggravated to greatly reduce the available capacity and shorten the service life. There is an urgent need to establish an evaluation mechanism for the inconsistency of batteries to provide the gist for the high-efficiency and reliable management. The test platform and scheme are built to obtain accurate data. And a comprehensive evaluation method for battery inconsistency is proposed based on the information entropy. Taking the capacity, internal resistance and the ratio of constant current charge capacity to constant voltage charge capacity as evaluation factors, the inconsistency of the battery pack composed of twelve cells is analyzed comprehensively. Experimental results show that the method can scientifically evaluate the inconsistency of the battery pack in different life and has wide adaptability.