Improvement of online detection method of battery remaining capacity under high noise conditions

0 Introduction

The remaining capacity of the battery is a problem that users are most concerned about. It is closely related to the reliability of the entire power supply system. The higher the remaining capacity of the battery, the higher the system reliability. Therefore, it is of great significance to detect the remaining capacity of the battery in real time online without consuming the battery energy and affecting the normal operation of the electrical equipment.

The battery is a complex electrochemical system. When it is running under different load conditions or different ambient temperatures, the actual remaining capacity available for release is different; and as the battery is used for a longer time, its capacity will also decrease. The remaining capacity is usually estimated based on the electrolyte density of the battery. This method has great limitations: in the later period of battery use, with the corrosion and fracture of the positive and negative plates, it is difficult to accurately calculate the remaining capacity; at the same time, this method is also difficult to adapt to the online detection of VRLA batteries that are currently widely used. Among the several commonly used battery remaining capacity detection methods in recent years, for batteries used online, the internal resistance method has the least impact on the system and can be accurately measured throughout the battery's entire service life. Therefore, the internal resistance method is considered to be a relatively ideal method. However, in the case of high noise, it is found that the actual measured accuracy of the battery remaining capacity is not satisfactory. Therefore, it is imperative to improve the online detection method of the battery remaining capacity in the case of high noise.

1. Implementation plan for predicting remaining capacity using the internal resistance method

A large number of research results show that there is a good correlation between the internal resistance of the battery and the degree of charge. The American GNB company has tested nearly 500 VRLA batteries with a capacity of 200 to 1000A·h and a battery pack voltage of 18 to 360V. The experimental results show that the correlation between the internal resistance of the battery and the capacity is very good, and the correlation coefficient can reach 88%. As the battery charges, the internal resistance gradually decreases; as the discharge process proceeds, the internal resistance gradually increases. In addition, as the battery ages, its residual capacity decreases and the internal resistance gradually increases. The typical relationship curve between the internal resistance of the battery and the residual capacity is shown in Figure 1.

Figure 1 Relationship curve between battery internal resistance and remaining capacity

When the battery is fully charged (full) and fully discharged (discharged), its internal resistance differs by 2 to 4 times, and the rate of change is much greater than the rate of change of the battery terminal voltage (about 30% to 40%). Therefore, by measuring the internal resistance of the battery, its remaining capacity can be predicted more accurately. In addition, for batteries used online, the internal resistance method has another outstanding advantage, which is that it has the least impact on the system and can be accurately measured throughout the battery's entire service life. Therefore, it is not difficult to see that the internal resistance method is most suitable for online measurement of the remaining capacity of VRLA batteries.

The specific implementation plan of the internal resistance method to predict the remaining capacity is: first, fully charge the battery (for example, a 2V battery is charged to 2.23V and the floating charge current is 10mA), then discharge the battery at a discharge rate of 0.1 C , and record the internal resistance and the remaining capacity during the discharge process. When the battery is discharged (a 2V battery is discharged to 1.80V), a complete discharge curve can be obtained, that is, the corresponding relationship between the remaining capacity and the internal resistance of the battery. This curve is stored in the FLASHROM of the battery monitoring system. When testing batteries of the same model and specification in the future, the processor calculates the remaining capacity based on the internal resistance value obtained by the online test through a table lookup. Therefore, the key to this method is how to measure the internal resistance of the battery online.

The principle of measuring the internal resistance of a battery is as follows: a constant AC audio current source _Is is applied to both _ends of the battery , and then the terminal voltage _Vo and the angle θ between Is and Vo _are detected . Obviously, _the AC impedance of the battery is Z = Vo / Is , and R = Z ×cos θ is the internal resistance _of the battery we want to obtain. The specific implementation scheme is shown in Figure 2.

Figure 2 Implementation of the internal resistance method to predict remaining power

The 300Hz signal generating circuit in the figure is composed of a 14-bit binary serial counter/divider CD4060 and a bandpass filter circuit. The specific circuit is shown in Figure 3.

Figure 3 300Hz signal generation circuit

2 Improvement of online measurement method under high noise conditions

The above method is used to measure the remaining capacity of the battery with high accuracy, and the error is better than 7% in offline measurement. However, in real-time online measurement, it is found that the accuracy of the remaining capacity of the battery measured is not satisfactory, and sometimes the error is even more than 10%. What causes this error? To this end, we tested several working states of the battery, namely the floating charge state under normal power supply, the discharge state under mains failure, and the state under inverter failure. It was found that in the case of inverter failure, the accuracy can be similar to that of offline testing. So far, it can be concluded that the reason for the reduced accuracy is caused by the feedback noise of the inverter. For this reason, we tried to use a low-pass filter (cut-off frequency of 1kHz) to improve the test accuracy, but the effect was not ideal. It can be seen that the feedback noise of the inverter is mainly concentrated in the range below 1kHz, so it is difficult to achieve in hardware.

Considering that n standard internal resistance values ​​have been tested in advance, the least square fitting method can be used in the software to correct the data. The so-called least square problem is to find an undetermined function f ( x ) so that the sum of the squares of the differences between f ( x ) and the standard value y is minimized, that is,

s = [ y _i － f ( x _i )] ² （1）

The solution process of f ( x ) is as follows: First, assume that f ( x ) is an n- degree polynomial, that is,

f ( x )= a ₀ ＋ a ₁ x ＋…＋ a _n x ⁿ = a _i x ⁱ （2）

Then take out the n standard internal resistance values ​​that have been tested and set them as y ₁ , y ₂ , …, y _n ; thus equation (1) can be simplified to [page]

s =[ a ₀ ＋ a ₁ x _i ＋…＋ a _n x ⁿ _i － y _i ] ² （3）

According to the extreme value principle in calculus, to minimize equation (3), its partial derivative with respect to each coefficient must be 0, that is:

（4）

There are n + 1 equations in this formula , so n + 1 unknowns can be solved . Substituting equation (3) into equation (4) and simplifying it, we can get

（5）

Finally, by substituting the n resistance values ​​x ₁ , … , x _n and the n standard internal resistance values ​​y ₁ , y ₂ , … , yn into formula (5), we can determine n + ₁ coefficients a ₀ , a _{1 ,} … , an , and thus obtain f ( x ) .

3 Experimental Results

The system test results are given by taking battery parameters and AC and DC voltage as examples. The standard meter used for the test is ESCORT3155A; the battery used for the test is GFM200 of Nandu Company, and the capacity obtained by discharging it at a constant current of 0.1 C discharge rate under rated load is used as the standard capacity; the test environment temperature is 18℃. The test results of the battery remaining capacity are listed in Table 1.

Table 1 Battery remaining capacity test results

It can be seen from the experimental results that after the least squares method fitting, the measurement results are basically close to the offline measurement results. Its accuracy can fully meet the requirements of the "Technical Requirements for Centralized Monitoring Systems of Communication Power Supply and Air Conditioning".

4 Conclusion

The online detection of the remaining capacity of the battery has a profound impact on the reliability of the entire power supply system. It has also been a hot and difficult issue for research at home and abroad. This paper gives a specific implementation plan for measuring the remaining capacity of the battery by the internal resistance method, and proposes an improved solution for online detection of the remaining capacity of the battery under high noise conditions. The experimental results prove that this improved method can obtain satisfactory measurement accuracy under high noise conditions.