Dynamic battery SOC estimation method?

　　Author ：Iflowpower – Portable Power Station Supplier

Since the development of battery technology, many kinds of methods used to estimate SOC have already occurred. There are only traditional current integrated methods, battery internal resistance, discharge test methods, open circuit voltage methods, load voltages, and more innovative Kalman filtering methods. Fuzzy logical theory and neural networking, etc.

It is currently one of the more common SOC estimation methods in the field of battery management system, and the essence is to estimate the SOC of the battery by accumulating or discharging electricity by accumulating or discharging by accumulating or discharging. At the same time, according to the discharge rate and battery temperature. A certain compensation for the estimated SOC.

If the battery is defined as SOCT0 when the battery is initial in the charge and discharge initial state, then the battery remaining capacity SOC after T is: q, Q is the battery rated capacity, and N is charge and discharge efficiency, also called coulomb efficiency, its value The battery charge and discharge rate is determined, I is the current of T. The current integrated method is relatively simple and reliable than other SOC estimation methods, and the SOC value of the battery can be dynamically estimate, so it is widely used. However, this method also has two limitations: one, current integral method requires an initial SOC value of the battery in advance, and accurately collects the current flowing into or out of the battery, in order to make the estimation error be as small as possible; second, This method is only based on the external feature of the battery, and the battery self-discharge rate, the degree of aging, and the charge and discharge ratio of the battery SOC are ignored to a certain extent.

Long-term use can also cause the measurement error to expand, so it is necessary to introduce Related correction coefficients Correct accumulation errors. (2) Discharge Test Method The discharge test method is to continuous constant current discharge discharge until the cutoff voltage of the battery, multiply the time used by this discharge process by the size value of the discharge current, that is, the remaining capacity of the battery. The method generally uses this method as a calibration method of battery SOC or in the late maintenance of the battery, and is relatively simple, reliable, and the result is relatively accurate without knowing the battery SOC value.

All effectively. However, there are two shortcomings in the discharge test method: First, the test process of this method needs a lot of time; second, when using this method, it is necessary to remove the target battery from the electric vehicle, so the method cannot be used to calculate Power battery in working state. (3) The open circuit voltage method is based on the change relationship between the opening voltage of the battery and the OCVOTAGE, OCV) and the internal lithium ion concentration of the battery, and indirectly fits the corresponding relationship between it and the battery SOC.

When performing actual operation, it is necessary to discharge the battery after the battery is filled with a fixed discharge ratio (generally 1c) until the discharge is stopped, and the relationship between OCV and SOC is obtained according to the discharge process. When the battery is in an actual operating state, the current battery SOC can be obtained by finding the OCV-SoC relational table according to the voltage value at both ends of the battery. Although the method is effective for various batteries, it also exists self-defects: First, the target battery must be allowed to stand more than 1 h before measuring OCV, thereby uniformly distributing the internal electrolyte in the battery to obtain a stable end voltage; second, the battery is at different temperatures Or during different life, although the open circuit is the same, the actually SOC may differ, and the measurement result is not guaranteed to be completely accurate long-term use of this method.

Therefore, the open circuit voltage method is the same as the discharge test method, does not apply to the running battery SOC estimation. (4) Kalman filtering method KALMAN filtering method is a new type of optimized self-regression data filtered in the "New Achievements of Linear Filtering and Forecasting Theory" in the 1960s. algorithm.

The essence of the algorithm is that the state of the complex dynamic system can be optimized for the state of the complex dynamic system according to the principle of minimum meanowns. Non-linear dynamic systems will be linear into a state space model of the system in the Kalman filtering method. When the actual application, the system is updated with the observed value of the current time, followed by the observed value of the current time.

"Forecast - Measurement - Corrected" mode, eliminating the deviation and interference of the system random. When the SOC of the powertrain is estimated using the Kalman filtering method, the battery is converted into a state space model in the form of a power system, and the SOC becomes a state variable inside the model. The established system is a linear discrete system.

Since the Kalman filtering method not only corrects the initial error of the system, it can effectively suppress system noise, so there is a significant application value in the SOC estimation of electric vehicle power batteries in operation conditions. However, the method also exists two-point defects: one, Kalman filtering method estimates the accuracy of the SOC depends largely on the accuracy of the battery model, the working characteristics itself is highly non-linear power battery, in the Kalman filtering method After linearization, it is inevitable that there is no error, and if the model is established, the estimated result is not necessarily reliable; the second, the method involved is very complicated, the amount of calculation is extremely large, and the calculated calculation period is longer, and Hardware performance requirements. (5) Neural network method neural network method is analog human brain and its neuron used to deal with a new type of algorithm for nonlinear systems.

It does not require in-depth research of the internal structure of the battery, just extracting a large number of working characteristics from the target battery in advance. Enter the SOC value in the run from the output sample and input it into the system established by using the method. The method is relatively simple in the later processing, that is, it can effectively avoid the error of the Kalman filtering method to make the battery model as linearization, and can obtain the dynamic parameters of the battery in real time.

However, the pre-working volume of the neural network method is relatively large, and a large number of more and comprehensive target sample data is required to train the system. The method of training data and training is largely affecting the estimation accuracy of SOC. In addition, under the complex action of battery temperature, self-discharge ratio and battery aging, the method is used to estimate the SOC value of the same set of batteries for a long time, and its accuracy will also be big discount.

Therefore, this method is not very common in the SOC estimation work of the power battery.