A method and device for predicting battery service life and capacity

Abstract

The present invention relates to a method and device for detecting the service life and capacity of a battery, comprising the following detection steps: S1, performing a constant current discharge test on N sample batteries of the same type as a battery to be tested, and obtaining N groups of voltage time series data urr1, urr2...urrN; S2, constructing a time normalization formula independently applicable to each group of voltage time series data; S3, performing time normalization on the N groups of voltage time series data according to the time normalization formula, and drawing a point set diagram according to the N groups of voltage time series data after time normalization, and obtaining a convergence trajectory; S4, performing data fitting on the convergence trajectory to obtain a time normalization standard curve L, and the voltage value range of the time normalization standard curve L is recorded as [UA, UB]; S5, performing constant current discharge on the battery to be tested, and obtaining relevant parameters of the battery to be tested; S6, predicting the service life of the battery to be tested; S7, calculating the capacity of the battery to be tested.

Description

Technical Field

[0001] The present invention relates to the technical field of battery capacity detection, and in particular to a method and device for predicting the service life and capacity of a battery. Background Art

[0002] When the existing battery pack is subjected to capacity verification charge and discharge test, C10 (10-hour discharge rate, battery capacity is 200Ah, C10=200 / 10=20A) constant current discharge is adopted. When the discharged capacity exceeds 80% (such as routine maintenance) or 100% (such as newly installed battery pack) of the rated capacity, if the voltage of each battery is still greater than 1.8V (battery with rated voltage of 2V), it means that the capacity of the battery pack is qualified. After the test is qualified, it is left to stand for 1 to 2 hours, and then the C10 current is used to perform constant current voltage limiting charging, constant voltage charging and floating charging in sequence. It can be put into operation only after the charging is completed. This verification test method has the following main defects:

[0003] Consumes a lot of time, manpower and financial resources

[0004] At present, in theory, the battery pack discharge test requires constant current discharge for 8 hours (such as battery packs that have been put into operation) or 10 hours (such as newly installed battery packs). After the discharge is completed, charging takes about 15 hours. In addition to the preparatory work before charging and discharging and the time spent on the round trip to the substation, it takes 3 days to complete a set of battery pack verification charge and discharge tests. If the full battery verification charge and discharge test is outsourced, it takes 3 days for a single substation, and the project maintenance cost is high. In addition, uncontrollable factors in the test process also increase resource costs. On the one hand, during the battery pack discharge test, when unqualified batteries appear, the discharge detector will sound an alarm and stop discharging. At this time, the test personnel are required to manually remove the unqualified batteries, re-remove and install the signal lines, and then delete the batteries from the discharge detector before restarting the discharge steps. When there are many unqualified batteries, the test time will be greatly increased. On the other hand, the signal collector wire clamp used in the battery discharge test is small and has a small bite force. During the test, it is very easy to loosen due to human operation, equipment vibration and other reasons. Once it falls off, the signal line needs to be reconnected and the test restarted, which requires the test personnel to closely observe and handle the whole process.

[0005] (2) Backup battery packs pose safety risks

[0006] In most cases, when conducting a battery pack verification discharge test, the battery pack to be tested needs to be removed from the DC system, and a set of spare battery packs needs to be connected to the DC system. The spare battery packs need to be carried by maintenance personnel themselves. Due to the complexity and diversity of battery models, it is difficult for the spare battery packs to fully match the DC system of the substation, and there is a possibility of damage to the battery packs during transportation back and forth. If the spare battery pack is used as the backup power supply for the DC system of the substation for a long time, when a serious short circuit fault occurs near the substation, the spare battery pack may not be able to supply power normally due to its own problems. This will cause the AC system voltage in the station to drop seriously, and the DC charger that relies on AC power may stop working, making the protection device in the station unable to operate to eliminate the fault. The fault can only be eliminated by the upper level protection action, which prolongs the time to handle the fault.

[0007] A paper titled "Research on Prediction Model of Remaining Discharge Time of Lead-acid Batteries" in the 19th issue of the 151st issue of Modern Industrial Economy and Informatization in 2017 mentioned a data processing scheme of normalizing time and voltage, which is just a common data preprocessing scheme, that is, normalizing the data and compressing it to the range of 0 to 1. The disadvantage of this processing scheme is that when the difference between the starting voltage and the ending voltage of each discharge curve is large, the normalized curves will not overlap. Summary of the invention

[0008] The object of the present invention is to provide a method and device for predicting the service life and capacity of a battery, which is not limited by battery type and batch, and at the same time improves the applicable scope of the prediction model.

[0009] To achieve the above object, the present invention is implemented through the following technical solutions:

[0010] A method for detecting the service life and capacity of a battery comprises the following detection steps:

[0011] S1, perform a constant current discharge test on N sample batteries of the same type as the battery to be tested, and obtain N groups of voltage time series data urr1, urr2...urrN;

[0012] S2, constructing a time normalization formula that is independently applicable to each set of voltage time series data;

[0013] S3, performing time normalization on the N groups of voltage time series data respectively according to the time normalization formula, and drawing a point set diagram according to the N groups of voltage time series data after time normalization to obtain a convergence trajectory;

[0014] S4, performing data fitting on the convergence trajectory to obtain a time-normalized standard curve L, wherein the voltage value range of the time-normalized standard curve L is recorded as [UA, UB];

[0015] S5. Perform constant current discharge on the battery under test to obtain relevant parameters of the battery under test;

[0016] S6. Predict the available usage duration of the battery under test according to the time normalization standard curve L and the relevant parameters of the battery under test;

[0017] S7. Obtain the capacity of the battery under test based on the available usage duration of the battery under test.

[0018] Further, the N groups of voltage time series data urr1, urr2...urrN are the voltage time series data of N sample batteries during the stable voltage drop period.

[0019] Further, each group of the voltage time series data is composed of a series of ut data measured by each sample battery during the constant current discharge test, and the value of N is an integer greater than zero.

[0020] Further, step S2 further includes the following steps:

[0021] Draw a point set diagram based on the obtained N groups of voltage time series data urr1, urr2...urrN. Project the N groups of voltage time series data urr1, urr2...urrN onto the voltage axis respectively to obtain N projection intervals on the voltage axis. Denote the maximum overlapping interval of the N projection intervals on the voltage axis as [U1, U2], where U1 < U2. Arbitrarily take two points u1 and u2 in the interval [U1, U2], and u1 and u2 satisfy U1 ≤ u1 < u2 ≤ U2. Draw parallel lines s1 and s2 to the time axis through points u1 and u2 respectively. Then, the t data in each group of voltage time series data has a maximum t value tmx and a minimum t value tmn between the parallel lines s1 and s2; construct a time normalization formula applicable to each group of voltage time series data independently according to the tmx and tmn applicable to each group of voltage time series data urr1, urr2...urrN respectively. The time normalization formula is:

[0022]

[0023] where tg is the time value on the T axis after time normalization, and t is all time values of each group of voltage time series data.

[0024] Further, step S2 further includes the following steps:

[0025] Use the N groups of voltage time series data urr1, urr2...urrN to draw a line graph composed of N sample battery broken lines. Then, the parallel lines s1 and s2 will intersect each sample battery broken line at 2 points respectively. Among them, the t value in the intersection coordinate with the line s1 is denoted as tmx, and the t value in the intersection coordinate with the line s2 is denoted as tmn.

[0026] Further, the convergence trajectory obtained in S3 has the following characteristics: the voltage time-series data defined within the parallel lines s1 and s2 is normalized to the interval [0, 1], and the voltage time-series data outside the parallel lines s1 and s2 is normalized outside the interval [0, 1].

[0027] Further, the data fitting method adopted in S4 is any one or several of machine learning regression algorithms, perceptrons, multi-layer perceptrons (commonly known as neural networks), least squares method, linear fitting, polynomial fitting, exponential fitting, and Gaussian fitting.

[0028] Further, S5 further includes the following steps: performing a constant-current discharge test on the battery under test, with a total constant-current discharge duration of t3, taking the voltage UC corresponding to a moment t1 as the starting voltage, where the period from t1 to t3 is within the voltage steady-drop period of the battery under test, and UC satisfies UA < UC < UB, recording the voltage time-series data within the period from t1 to t3, denoted as arr, and recording the constant-current discharge current i of the battery under test.

[0029] Further, take t2 ∈ (t1, t3], and denote the time-voltage sequence data within the time range from t1 to t2 as brr.

[0030] Further, the voltage time-series data obtained by the battery under test during the voltage steady-drop period needs to be preprocessed, and the preprocessing method is any one or several of exponential smoothing, Kalman filtering, and moving average.

[0031] Further, S6 further includes the following steps:

[0032] The available duration includes the optimal duration st, where the optimal duration st is the discharge duration of the voltage of the battery under test from UC to UD; intercept the curve segment of the voltage from UC to UD in the normalized curve equation L, denoted as L1, scale and translate the time axis of L1 to [0, 1], and the scaled and translated curve is denoted as L2. The optimal duration st can be predicted through the equation of curve L2.

[0033] Further, S6 further includes the following steps:

[0034] Suppose the equation of the curve L2 after scaling and translating the time axis is as follows:

[0035] t = startx + x × (endx - startx)

[0036]

[0037] Among them, startx and endx are the boundary values ​​of the intercepted curve segment L1 on the time axis, and x is the independent variable of the new equation, x∈[0,1].

[0038] Use st to construct a time normalization formula for arr, the normalization formula is:

[0039]

[0040] Wherein, t is the measured value of the time of arr, t1 is the time when the discharge of the battery to be tested reaches the voltage UC, and tg1 is the value after time normalization;

[0041] The normalized data is recorded as arr1. Substitute the time of arr1 into the curve equation L2 to obtain a voltage value sequence, recorded as urr. If the error between the voltages corresponding to the time corresponding to urr and arr1 can be minimized, then st is the optimal time, that is, the predicted value of the discharge time of the battery to be tested from UC to UD. The optimization method can be used to solve st.

[0042] Furthermore, the optimization method is any one or more of the following: gradient descent method, Newton method, quasi-Newton method, conjugate gradient method, Lagrange multiplier method, conjugate gradient method or grid search method.

[0043] Furthermore, the usable time also includes a total usable time qt and a remaining usable time rt. The total usable time qt is the discharge time for the voltage of the battery to be tested to drop from constant current discharge to UD, and the remaining usable time rt is the usable discharge time for the battery to be tested to drop to UD after the constant current test of the battery to be tested is completed. The qt=st+t1, and the rt=st-(t3-t1).

[0044] Furthermore, the capacity of the battery to be tested includes the capacity sc consumed from UC to UD, the total available capacity qc and the remaining available capacity rc, the total available capacity qc is the total capacity that can be discharged from the battery voltage to be tested from the start of constant current discharge to the UD, and the remaining available capacity rc is the capacity that the battery to be tested is predicted to be discharged to UD after the constant current test of the battery to be tested is completed, the sc = i × st, the qc = i × qt = i × (st + t1), and the rc = i × rt = i × [st - (t3 - t1)].

[0045] Furthermore, the battery may be any one or more of a lead-acid battery, a nickel-hydrogen battery, a lithium-ion battery, a nickel-cadmium battery, a sodium-sulfur battery, and a nickel-zinc battery.

[0046] Furthermore, the UA is the minimum voltage value of all voltage time series data of the convergence trajectory, and the UB is the maximum voltage value of all voltage time series data of the convergence trajectory.

[0047] A battery usable time and capacity detection device comprises: a model building module (1), a battery detection module to be tested (2) and a battery usable time and capacity prediction module to be tested (3);

[0048] The model building module (1) is used to collect relevant parameters of the sample battery under constant current discharge conditions, and obtain a time normalized standard curve L and a voltage value range [UA, UB] of the time normalized standard curve L according to steps S1-S4 in the battery service life and capacity detection method;

[0049] The battery detection module (2) is used to collect and record relevant time and voltage parameters of the battery under constant current discharge conditions according to step S5 of the battery usable life and capacity detection method;

[0050] The battery usable time and capacity prediction module (3) is used to obtain the usable time and capacity of the battery to be tested according to steps S6-S7 of the battery usable time and capacity detection method.

[0051] A storage medium stores a program, which, when executed by a processor, enables the processor to execute a method for predicting battery service life and capacity as described in any embodiment of the present invention.

[0052] The present invention has the following beneficial effects:

[0053] 1. The capacity prediction method of the present invention has strong versatility, high accuracy and extremely short time consumption, which solves the technical problem that the existing battery capacity testing technology cannot take into account versatility, accuracy and speed at the same time.

[0054] 2. The present invention is applicable to batteries of different capacities and scenarios with different constant current discharge currents.

[0055] 3. The present invention proposes a time normalization formula. This formula is an extension of the common voltage interval condition. The formula can be used to use data outside the common voltage interval to expand the voltage range of the final normalized curve, that is, the formula can more easily obtain a time normalized curve with a large voltage range.

[0056] 4. The present invention has good adaptability; first, the voltage of the battery to be tested at the beginning of the discharge stage may be greater than the maximum voltage of the normalized curve, but as the discharge proceeds, the battery to be tested will eventually enter a voltage steady decline interval, and the voltage will eventually decrease. As long as the voltage falls into the voltage interval of the normalized curve, the subsequent discharge of the battery to be tested can be predicted; second, the present technology can be applied to the prediction of the usable time and capacity of batteries including but not limited to lead-acid batteries, nickel-hydrogen batteries, lithium-ion batteries, nickel-cadmium batteries, sodium-sulfur batteries, nickel-zinc batteries, and the like. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Attached Figure 1 is a flow chart of a method for predicting battery capacity and usable time of the present invention;

[0058] Attached Figure 2 is a schematic diagram of the point set in step S2 of the present invention;

[0059] Attached Figure 3 is a schematic diagram of a broken line in step S2 of the present invention;

[0060] Attached Figure 4 is a schematic diagram of the convergence trajectory of step S3 of the present invention;

[0061] Attached Figure 5 is a schematic diagram of a time-normalized standard curve L obtained by data fitting in step S4 of the present invention;

[0062] Attached Figure 6 It is a schematic diagram of the prediction process and effect verification of the battery to be tested of the present invention;

[0063] Attached Figure 7 is a schematic diagram of the curve L2 in step S6 of the present invention;

[0064] Attached Figure 8 It is a structural schematic diagram of a device for predicting battery capacity and usable time of the present invention.

[0065] The reference numerals in the figure are as follows:

[0066] 1. Model building module; 2. Battery detection module; 3. Battery usage time and capacity prediction module. DETAILED DESCRIPTION

[0067] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0068] According to an embodiment of the present invention, an embodiment of a method for predicting battery service life and capacity is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0069] Figure 1is a flowchart of a method for predicting the available duration and capacity of a battery according to an embodiment of the present invention. As Figure 1 shown, the method includes two stages. The first stage is the establishment stage of the time normalization curve model, which is generally carried out in a laboratory. The second stage is the prediction stage of the capacity of the battery to be measured, which is generally carried out at the prediction site of the battery to be measured.

[0070] The first stage includes the following steps:

[0071] S1. Perform a constant current discharge test on N sample batteries of the same type as the battery to be measured to obtain N sets of voltage time series data urr1, urr2...urrN.

[0072] Specifically, based on the battery discharge principle, it can be known that during the constant current discharge process, the battery will first experience a period of rapid voltage change and then enter a period of steady voltage decline. Therefore, in step S1, it is necessary to剔除 the data of the period of rapid voltage change in the initial stage of discharge for each sample battery. Therefore, the N sets of voltage time series data urr1, urr2...urrN are the voltage time series data of N sample batteries during the period of steady voltage decline.

[0073] Furthermore, each set of voltage time series data is composed of a series of ut (i.e., voltage time, abbreviated as ut) data measured for each sample battery during the constant current discharge.

[0074] Furthermore, the value of N is an integer greater than zero.

[0075] S2. Construct a time normalization formula independently applicable to each set of voltage time series data.

[0076] Specifically, as Figure 2 shown, draw a point set diagram according to the obtained N sets of voltage time series data urr1, urr2...urrN, project the N sets of voltage time series data urr1, urr2...urrN onto the voltage axis respectively to obtain N projection intervals on the voltage axis, and record the maximum overlapping interval of the N projection intervals on the voltage axis as [U1, U2], where U1 < U2. Arbitrarily take two points u1 and u2 in the interval [U1, U2], and the u1 and u2 satisfy U1 ≤ u1 < u2 ≤ U2. Draw parallel lines s1 and s2 to the time axis through the points u1 and u2 respectively. Then, the t data in each set of voltage time series data has a maximum t value tmx and a minimum t value tmn between the parallel lines s1 and s2;

[0077] The maximum t value tmx and the minimum t value tmn can be obtained by the following method. As Figure 3As shown, a line graph consisting of N sample battery fold lines is drawn using the N groups of voltage timing data urr1, urr2...urrN, then the parallel lines s1 and s2 will intersect with each sample battery fold line at 2 points respectively, wherein the t value in the coordinates of the intersection with the straight line s1 is recorded as tmx, and the t value in the coordinates of the intersection with the straight line s2 is recorded as tmn. Therefore, each group of voltage timing data urr1, urr2...urrN has its own independently applicable tmx and tmn. Figure 3 It also demonstrates how to take the values ​​of tmx and tmn corresponding to two sample battery fold lines among N sample battery fold lines, where tmx1 and tmn1 are the maximum t value and minimum t value of the voltage timing data corresponding to the sample battery fold line close to the bottom, and tmx2 and tmn2 are the maximum t value and minimum t value of the voltage timing data corresponding to the sample battery fold line close to the top.

[0078] According to the tmx and tmn that are independently applicable to each set of voltage time series data urr1, urr2, ..., urrN, a time normalization formula that is independently applicable to each set of voltage time series data is constructed. The time normalization formula is:

[0079]

[0080] Wherein, tg is the T-axis time value after time normalization, and t is all the time values ​​of each set of voltage time series data;

[0081] S3, performing time normalization on the N groups of voltage time series data respectively according to the time normalization formula, and drawing a point set diagram according to the N groups of voltage time series data after time normalization to obtain a convergence trajectory.

[0082] The point set diagram of N groups of voltage time series data after time normalization is as follows: Figure 4 As shown, the voltage time series data defined within the parallel lines s1 and s2 are normalized to the interval [0,1], and the voltage time series data outside the parallel lines s1 and s2 are normalized to outside the interval [0,1].

[0083] S4, performing data fitting on the convergence trajectory to obtain a time-normalized standard curve L, wherein the voltage value range of the time-normalized standard curve L is recorded as [UA, UB].

[0084] Specifically, if the data fitting is the least squares method, the specific steps are as follows:

[0085] Suppose there is a univariate polynomial curve L, the sum of the differences between the voltage time series data points on the convergence trajectory and the curve L is the smallest, and the equation of the curve L is assumed to be:

[0086]

[0087] Wherein, k is an integer greater than 0, and the order of the polynomial is k-1; β is the coefficient of the polynomial, the value range of u is recorded as u∈[UA,UB], UA is the minimum voltage value of all voltage time series data of the convergence trajectory, and UB is the maximum voltage value of all voltage time series data of the convergence trajectory;

[0088] The coefficient matrix of curve L can be obtained by the following formula:

[0089] β=(X T X) -1 X T Y

[0090] Among them, X is the polynomial matrix of the voltage time series data at time t, and Y is the column vector composed of the voltage U corresponding to t. The form of X is as follows:

[0091]

[0092] The time-normalized standard curve L obtained by data fitting is as follows: Figure 5 shown.

[0093] Specifically, the data fitting may also adopt various regression algorithms of machine learning, perceptron, multi-layer perceptron (commonly known as: neural network), least squares method, linear fitting, polynomial fitting, exponential fitting, Gaussian fitting and other data fitting methods.

[0094] The value range [UA, UB] of the time-normalized standard curve L determines the voltage range that can be predicted by the time-normalized standard curve L, and the electrochemical properties of the battery determine that the discharge voltage of the battery has a rough range; therefore, as long as enough constant current discharge data of the same type of batteries with a large discharge voltage range are collected, the voltage range that can be predicted by the time-normalized standard curve L can be made wider, greatly improving the versatility of this prediction method.

[0095] The second stage combines the time normalization standard curve L, and by performing a short-term constant current discharge on the battery to be tested, a small amount of voltage time series data is obtained to obtain the usable time and capacity of the battery to be tested, which specifically includes the following steps:

[0096] S5, performing a constant current discharge test on the battery to be tested to obtain relevant parameters of the battery to be tested.

[0097] Specifically, Figure 6As shown, a constant current discharge test is performed on the battery to be measured. The total duration of the constant current discharge is t3. The voltage UC corresponding to a moment t1 is taken as the starting voltage. The period from t1 to t3 is within the period of the stable voltage drop of the battery to be measured. The UC satisfies UA < UC < UB. Record the voltage time series data during the period from t1 to t3, denoted as arr, and record the constant current discharge current i of the battery to be measured.

[0098] Further, since non-constant current data or poor-quality data may appear in the final stage of the constant current discharge test of the battery to be measured, the data in this final stage can be discarded, and the data from t1 to t2 with better data quality in the period from t1 to t3 is selected. Specifically, take t2 ∈ (t1, t3], and record the time-voltage sequence data within the time range from t1 to t2 as brr.

[0099] Further, due to issues such as the accuracy of the instrument and equipment, there will be certain errors in the measured voltage time series data. Therefore, data preprocessing needs to be performed on the measured voltage time series data arr and brr. The methods of the data preprocessing include, but are not limited to, exponential smoothing, Kalman filtering, and moving average, etc.

[0100] S6. Predict the available duration of the battery to be measured according to the time normalization standard curve L and the relevant parameters of the battery to be measured.

[0101] Specifically, as Figure 6 shown, arbitrarily take a voltage UD, where the UD satisfies UC < UD < UB. The available duration includes the optimal duration st. The optimal duration st is the discharge duration when the voltage of the battery to be measured drops from UC to UD; intercept the curve segment of the voltage from UC to UD in the normalization curve equation L, denoted as L1. For the convenience of subsequent calculations, scale and translate the time axis of L1 to [0, 1], and the scaled and translated curve is denoted as L2, as Figure 7 shown. The optimal duration st can be predicted through the equation of the curve L2.

[0102] Further, the steps to predict the optimal duration st through the equation of the curve L2 are as follows:

[0103] Assume the equation of the curve L2 is as follows:

[0104] t = startx + x × (endx - startx)

[0105]

[0106] where startx and endx are the boundary values of the time axis of the intercepted curve segment L1, x is the independent variable of the new equation, and x ∈ [0, 1];

[0107] A time normalization formula for arr is constructed based on st, and the normalization formula is:

[0108]

[0109] Wherein, t is the measured value of the time of arr, t1 is the time when the discharge of the battery to be tested reaches the voltage UC, and tg1 is the time normalized value.

[0110] The normalized data is recorded as arr1, and the time of arr1 is substituted into the curve equation L2. The obtained voltage time series data is recorded as urr. If the error between the voltage corresponding to the time corresponding to urr and arr1 can be minimized, then st is the optimal time, that is, the predicted value of the discharge time of the battery under test from UC to UD. The optimization method can be used to solve st.

[0111] Furthermore, if the voltage time series data of the battery to be tested in S5 adopts brr, then st can be obtained by replacing the arr data in S6 with brr data.

[0112] Furthermore, the voltage error corresponding to the time when urr and arr1 correspond to each other can be expressed by a variety of statistical methods, for example, it can be expressed by square error, and the error expression is as follows:

[0113]

[0114] Where:

[0115] M: logarithm of time-voltage data in arr;

[0116] βi: coefficient of the time normalized curve equation that has been obtained;

[0117] tj: the jth time value in arr;

[0118] Uj: the jth voltage value in arr, corresponding one to one with tj;

[0119] t1: the time when the battery under test discharges to the voltage UC, which is also the start time of arr;

[0120] st: The time it takes to discharge from voltage UC to voltage UD.

[0121] The st that minimizes the loss value is the predicted value of the time from voltage UC to voltage UD.

[0122] Furthermore, the optimization method can be a gradient descent method, Newton's method, quasi-Newton's method, conjugate gradient method, Lagrange multiplier method, conjugate gradient method or grid search method, among which the gradient descent method, Newton's method, quasi-Newton's method, conjugate gradient method, Lagrange multiplier method, and conjugate gradient method are commonly used methods for solving the optimal value. The grid search method used here means that since the capacity of the battery is known and the discharge current is known, the approximate range of the discharge time (for example, less than 24 hours and greater than 0 hours) is known. Continuous time points are taken within this time range and substituted into the loss formula. The time point that minimizes the loss value is the predicted value of st to be solved.

[0123] Furthermore, the available time also includes a total available time qt and a remaining available time rt. The total available time qt is the discharge time of the battery under test from the start of constant current discharge to UD, and the remaining available time rt is the available discharge time of the battery under test predicted to drop to UD after the constant current test of the battery under test is completed. Figure 6 It can be seen that qt=st+t1, and rt=st-(t3-t1).

[0124] S7, obtaining the capacity of the battery to be tested according to the usable time of the battery to be tested.

[0125] Specifically, the capacity of the battery to be tested includes the capacity sc consumed by discharging from UC to UD, the total available capacity qc and the remaining available capacity rc, wherein the total available capacity qc is the total capacity that can be discharged from the voltage of the battery to be tested from the start of constant current discharge to the UD, and the remaining available capacity rc is the capacity that is predicted to be discharged by the battery to be tested after the constant current test of the battery to be tested is completed when it drops to UD. According to the constant current discharge characteristics of the battery, discharge capacity = discharge current × discharge time, therefore, sc = i × st, qc = i × qt = i × (st + t1), and rc = i × rt = i × [st - (t3 - t1)].

[0126] Furthermore, the battery can be a lead-acid battery, a nickel-hydrogen battery, a lithium-ion battery, a nickel-cadmium battery, a sodium-sulfur battery, a nickel-zinc battery or other types of batteries.

[0127] According to another embodiment of the present invention, an embodiment of a battery service life and capacity prediction device is provided, such as Figure 8 As shown, the battery usable time and capacity prediction device comprises: a model building module 1, a battery detection module 2 and a battery usable time and capacity prediction module 3. The battery usable time and capacity prediction device is described in detail below.

[0128] Model building module 1 is used to collect relevant parameters of the sample battery under constant current discharge conditions, and obtain a time normalized standard curve L and a voltage value range [UA, UB] of the time normalized standard curve L according to steps S1-S4 in the battery service life and capacity detection method.

[0129] The battery detection module 2 is used to collect and record the time and voltage parameters related to the battery under constant current discharge conditions according to step S5 of the battery usable life and capacity detection method.

[0130] The battery usable time and capacity prediction module 3 is used to obtain the usable time and capacity of the battery to be tested according to steps S6-S7 of the battery usable time and capacity detection method.

[0131] According to another embodiment of the present invention, a storage medium is provided, which stores a program, wherein when the program is executed by a processor, the processor executes any one of the above-mentioned methods for predicting the battery service life and capacity.

[0132] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0133] In the above embodiments of the present invention, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0134] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units can be a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0135] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0136] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0137] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk and other media that can store program codes.

[0138] The above description is only an embodiment of the present invention, and does not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made by using the content of the present invention specification, or directly or indirectly applied in other related technical fields, is also included in the patent protection scope of the present invention.

Claims

1. A method for detecting the usable life and capacity of a battery, characterized in that: The detection steps include the following: S1, perform a constant current discharge test on N sample batteries of the same type as the battery to be tested, and obtain N groups of voltage time series data urr1, urr2...urrN; S2, constructing a time normalization formula that is independently applicable to each set of voltage time series data; S3, performing time normalization on the N groups of voltage time series data respectively according to the time normalization formula, and drawing a point set diagram according to the N groups of voltage time series data after time normalization to obtain a convergence trajectory; S4, performing data fitting on the convergence trajectory to obtain a time-normalized standard curve L, wherein the voltage value range of the time-normalized standard curve L is denoted as [U A ,U B ]; S5, performing constant current discharge on the battery to be tested to obtain relevant parameters of the battery to be tested; S6, predicting the usable time of the battery to be tested according to the time normalized standard curve L and relevant parameters of the battery to be tested; S7, obtaining the capacity of the battery to be tested according to the usable time of the battery to be tested; Wherein, the S2 further includes the following steps: Draw a point set graph according to the obtained N sets of voltage time series data urr1, urr2...urrN. Project the N sets of voltage time series data urr1, urr2...urrN onto the voltage axis respectively to obtain N projection intervals on the voltage axis. Denote the maximum overlapping interval of the N projection intervals on the voltage axis on the voltage axis as [U1, U2], where U1 < U2. Arbitrarily take two points u1 and u2 in the interval [U1, U2], and u1 and u2 satisfy U1 ≤ u1 < u2 ≤ U2. Draw parallel lines s1 and s2 to the time axis through points u1 and u2 respectively. Then the t data in each set of voltage time series data has the maximum t value t mx and the minimum t value t mn ; According to the t mx and t mn independently applicable to each set of voltage time series data urr1, urr2...urrN, construct a time normalization formula independently applicable to each set of voltage time series data. The time normalization formula is: Among them, t g is the T-axis time value after time normalization, and t is all the time values ​​of each set of voltage time series data; Wherein, the S2 further includes the following steps: Using the N groups of voltage time series data urr1, urr2...urrN, a line graph consisting of N sample battery lines is drawn. Then, the parallel lines s1 and s2 intersect with each sample battery line at two points respectively, wherein the t value in the coordinates of the intersection with the straight line s1 is recorded as t mx , the t value of the intersection coordinates with line s2 is recorded as t mn ; Wherein, the step S5 further includes the following steps: performing a constant current discharge test on the battery to be tested, wherein the total constant current discharge time is t3, and taking a voltage U corresponding to a time t1 C is the starting voltage, the period from t1 to t3 is the period during which the voltage of the battery to be tested steadily decreases, and the U C Meet U A C B , record the voltage time series data in the period from t1 to t3, recorded as arr, and record the constant current discharge current i of the battery to be tested;​​ Wherein, the S6 further comprises the following steps: The usable time includes an optimal time st, wherein the optimal time st is the time when the voltage of the battery to be tested is changed by the voltage U C To the U D The discharge time is: In the normalized curve equation L, the intercept voltage is U C To U D The curve segment is denoted as L1. The time axis of L1 is scaled and translated to [0,1]. The scaled and translated curve is denoted as L2. The optimal duration st can be predicted through the equation of the curve L2. Wherein, the S6 further comprises the following steps: Assume that the equation of the curve L2 after scaling and translation of the time axis is as follows: t=startx+x×(endx-startx) Among them, startx, endx are the boundary values ​​of the intercepted curve segment L1 time axis, x is the independent variable of the new equation, x∈[0,1]; Use st to construct a time normalization formula for arr, the normalization formula is: Wherein, t is the measured value of the time arr, t1 is the time when the battery to be tested discharges to the voltage U C Time, t g1 is the time-normalized value; The normalized data is recorded as arr1. Substitute the time of arr1 into the curve equation L2 to obtain the voltage value sequence, recorded as urr. If the error between the voltage corresponding to the time of urr and arr1 can be minimized, then st is the optimal time, that is, the battery to be tested is U C Down to U D The predicted value of the discharge duration can be solved by the optimization method.

2. A method for detecting battery service life and capacity as claimed in claim 1, characterized in that: The N groups of voltage time series data urr1, urr2, ..., urrN are voltage time series data of N sample batteries during a period of steady voltage decline.

3. A method for detecting battery service life and capacity as claimed in claim 1 or 2, characterized in that: Each group of voltage time series data is composed of a series of ut data measured by each sample battery in a constant current discharge test, and the value of N is an integer greater than zero.

4. A method for detecting battery service life and capacity as claimed in claim 1, characterized in that: The convergence trajectory obtained by S3 has the following characteristics: the voltage time series data defined within the parallel lines s1 and s2 are normalized to the interval [0,1], and the voltage time series data outside the parallel lines s1 and s2 are normalized to outside the interval [0,1].

5. A method for detecting battery service life and capacity as claimed in claim 1, characterized in that: The data fitting method used in S4 is any one or more of a machine learning regression algorithm, a perceptron, a multi-layer perceptron, a least squares method, a linear fitting, a polynomial fitting, an exponential fitting, and a Gaussian fitting.

6. A method for detecting battery service life and capacity as claimed in claim 1, characterized in that: Take t2∈(t1,t3], and record the time voltage series data in the time range from t1 to t2 as brr.

7. A method for detecting battery service life and capacity as claimed in claim 1, characterized in that: The voltage time series data obtained during the stable voltage drop period of the battery to be tested needs to be preprocessed, and the data preprocessing method is any one or more of exponential smoothing, Kalman filtering and moving average.

8. A method for detecting battery service life and capacity as claimed in claim 1, characterized in that: The optimization method is any one or more of the following: gradient descent method, Newton method, quasi-Newton method, conjugate gradient method, Lagrange multiplier method, conjugate gradient method or grid search method.

9. A method for detecting battery service life and capacity as claimed in claim 8, characterized in that: The usable time also includes the total usable time qt and the remaining usable time rt. The total usable time qt is the time when the voltage of the battery to be tested decreases from the constant current discharge to U D The remaining available time rt is the predicted discharge time of the battery under test after the constant current test of the battery under test is completed. D The available discharge time is: qt=st+t1, rt=st-(t3-t1).

10. A method for detecting battery service life and capacity as claimed in claim 1, characterized in that: The capacity of the battery to be tested includes C Discharge to U D The consumed capacity sc, the total available capacity qc and the remaining available capacity rc, wherein the total available capacity qc is the voltage of the battery to be tested from the start of constant current discharge to the U D The remaining available capacity rc is the total capacity that can be discharged after the constant current test of the battery under test is completed. D The capacity that can be discharged is: sc=i×st, qc=i×qt=i×(st+t1), and rc=i×rt=i×[st-(t3-t1)].

11. A method for detecting battery service life and capacity as claimed in claim 1, characterized in that: The battery is any one or more of a lead-acid battery, a nickel-hydrogen battery, a lithium-ion battery, a nickel-cadmium battery, a sodium-sulfur battery, and a nickel-zinc battery.

12. A method for detecting battery service life and capacity as claimed in claim 1, characterized in that: The U A is the minimum voltage value among all voltage time series data of the convergence trajectory, and the U B is the maximum voltage value of all voltage time series data of the convergence trajectory.

13. A battery life and capacity detection device, characterized in that: include: A model building module (1), a battery detection module (2) and a battery usage time and capacity prediction module (3); The model building module (1) is used to collect relevant parameters of the sample battery under constant current discharge conditions, and obtain a time normalized standard curve L and a voltage value range [U] of the time normalized standard curve L according to steps S1-S4 of the battery service life and capacity detection method according to any one of claims 1 to 12. A ,U B ]; The battery detection module (2) is used to collect and record relevant time and voltage parameters of the battery under constant current discharge conditions according to step S5 of the battery usable life and capacity detection method; The battery usable time and capacity prediction module (3) is used to obtain the usable time and capacity of the battery to be tested according to steps S6-S7 of the battery usable time and capacity detection method.

14. A storage medium storing a program, characterized in that: When the program is executed by a processor, the processor is enabled to execute the battery service life and capacity prediction method according to any one of claims 1 to 12.

Images