Disclosure of Invention
The invention provides a method for predicting the service life of a leadacid battery, which is developed aiming at the problems and comprises the following steps:
acquiring the battery discharge electric quantity value to be predicted in a work project of the leadacid battery in different environments, and calculating the attenuation ratio of the current battery capacity;
calculating the failure coefficient of the leadacid battery to be predicted in practical application according to the attenuation ratio, and further establishing a failure average rate model of the battery with the failure coefficient:
setting the selflearning battery state of charge to be equal to the initial quantity, and setting the selflearning battery state of charge to be equal to the initial quantity according to a least square vector machine according to a structural risk minimization principle when a battery failure coefficient is constantObtaining a least square vector machine LSSVM decision function:
wherein, K (x)_{i},x_{j}) Solving a regression prediction function by adopting internal operation for a kernel function which is an inverse function of an exponential function, and substituting the experimental numerical value into the formula to obtain a value a and a value b corresponding to a plurality of batteries to be detected;
substituting the erroroptimized a, b values into a battery residual life prediction model:
(t) was obtained as (0.798+ ln)^{Δc}).t^{2}+ΔT93.7；
And (3) wherein t represents the operation period of the storage battery, the general trend is reduced, the experimental numerical value is substituted into the number of the solving detection periods, and the inverse function is solved to obtain the operation life of the storage battery.
The battery failure rate is the speed of the decrease of the battery failure coefficient in the battery operation period; using a bellshaped function yax^{2}+ b, where the value of a is varied and Y represents the remaining life cycle of the battery; because a is that the changed parameter is not a fixed value, a vector machine function is adopted to solve a, and meanwhile, the derivative of the whole bellshaped function is solved to obtain the failure rate in a period.
In a preferred embodiment, the failure coefficient calculation is performed according to the following formula to obtain the ith point failure coefficient β_{mi}；
Wherein, C actual discharge capacity, ambient T temperature; t1i is the first discharge temperature, and the failure coefficient of the mth discharge is calculated as:
β＝(β_{m1}+β_{m2}+…+β_{mN})/N
n is the number of standard points taken on the cell discharge curve.
In a preferred embodiment, the step of establishing the average failure rate model of the battery with the failure coefficient further comprises the following steps:
comprehensively optimizing and disposing test data by using a bellshaped function and a vector machine, wherein the optimization method comprises the following steps: the battery failure rate is the rate at which the battery failure coefficient drops during the battery operating cycle;
the model is a bellshaped function y ═ ax^{2}+ b (wherein the value a is changed), Y represents the remaining life cycle of the battery, and since the parameter a is changed and not a fixed value, a is solved by adopting a vector machine function, and meanwhile, the derivative of the whole bellshaped function is obtained, namely the failure rate in one cycle;
establishing a storage battery actual application life prediction model under different use environments comprises the following steps: the power supply frequency of the battery, the magnitude of load current, the stability of the power supply current and the environmental temperature of the machine room;
calculating the battery failure rate based on the battery failure rate and the active vulcanization model instead of the battery vulcanization mechanism, wherein the established predicted life model expression is as follows:
where Δ β is the failure coefficient change, the period of the Δ T coefficient change,
whereinThe main factor affecting the activity of the battery to cause vulcanization and ultimately leading to the reduction of the service life of the application, a.b is the property coefficient affecting the frequency of the battery and the inherent property of the battery active under different environmental use conditions.
Furthermore, under the condition of not considering physical inherent factors such as battery active substances, design materials and the like, the relationship of the factors influencing the reduction of the battery activity is shown in the following expression of the relationship among power supply time, current, temperature and capacity:
is a factor of the reaction depth between the active and the medium,a factor for supplying energy to participate in chemical reactions, a being the active motor capacity of the battery, whereinThe expression is as follows:
I_{10}standard current; t standard time; k_{T}A temperature correction coefficient; k_{i}＝It/I_{10}A current correction factor;where t is the realtime period;while maleIn the formula T_{t}When greater than 10, set I_{10}.t_{10}Is C_{10},I_{t}.t_{t}And Ct, because of the environmental factors of the actual machine room, the selflearning battery state of charge is equal to the initial quantity, and the relational expression of the application life and the battery residual capacity temperature is shown as follows.
Due to the adoption of the technical scheme, the battery pack service life detection method disclosed by the invention can continuously optimize the battery discharge function according to the historical discharge state of the battery, and finally can be used as a method for evaluating the current battery service life, so that the detection of the battery service life can be completed remotely, and the method has the advantages of high precision, simplicity and convenience in operation and the like which are not possessed by the existing algorithm.
example, as shown in fig. 13, it is assumed in this example that 1 leadacid battery operating performance removes the effects of physical damage factors; the leadacid battery 2 is supposed to be charged in time after the operation and discharge are finished so as to be free from the influence of vulcanization; a400 AH twoclass valveregulated leadacid battery monomer is selected as a research sample.
The method comprises the following steps: analyzing factors influencing the reduction of the running performance of the leadacid battery and the relation between the performance reduction and the change of each running parameter, and determining parameters influencing the performance reduction of the leadacid battery;
the performance parameters of the leadacid battery in actual operation change along with the service time, the reason for causing the activity reduction of the battery is objective factors of manufacturing process and design,
for example in the manufacture of batteries
The cleaning degree of the active matter of the polar plate,
The quality of the lead plaster,
The strength of the grid material,
An isolating material,
The material of the shell body,
Medium, and efficiency of hydrogenoxygen combination.
The decline in performance of lead acid batteries during use refers to the mechanism of change in the rated capacity of the battery in operation, either up or down, but the general trend is irreversible.
The reasons for the reduction of the service life of the leadacid battery are as follows:
1. the normal working temperature of the environment temperature in the actual operation of the leadacid battery is 25 ℃, the activity of the battery is reduced below zero, and the physical damage of the battery is serious when the temperature is higher than 35 ℃;
2. the current of the leadacid battery is too large in the charging and discharging processes, so that the reaction efficiency of active substances of a battery plate is reduced, and the battery with insufficient capacity fails;
3. the leadacid battery discharges deeply in the discharging process, and the deep discharging causes irreversible vulcanization of active substances of a battery pole plate, so that the failure of the battery is accelerated;
the prediction of the battery application life refers to that the capacity discharged each time is gradually attenuated along with the increase of the service time of the battery under the condition that the battery is used in different application environments, the battery capacity attenuation ratio is calculated, and the failure coefficient of the leadacid battery is obtained and is used as an important parameter and basis in the prediction of the battery application life.
Step two: designing an accelerated leadacid storage battery application life test, and periodically adopting different parameters of reference points related to capacity attenuation ratio to obtain test data;
in order to accelerate the life test, a class II valveregulated leadacid battery is used as a research sample, the nominal capacity is 400AH, the working temperature is 25 ℃, and the working chargedischarge current is C_{10}The rated capacity discharge cutoff voltage of the battery is 1.800V.
The test conditions are carried out according to actual use, the secondary lead acid battery supplies power for the three types of commercial power communication machine room equipment, the actual use environment supplies power for 4 times per month, 80% of the actual capacity is provided each time, and the cutoff voltage is 1.800V. I.e., discharge testing was performed monthly and data was recorded, with a test current of 30A. The same group was selected and 4 monomers with different deep activity were used, the frequency of collection was monthly and 5 data were recorded, and the test data are shown in table 1.
TABLE 1 leadacid Battery operational Life test raw data
Step three: and calculating the practical application failure coefficient of the leadacid battery, and establishing a failure average rate model.
The method is characterized in that the service life of the storage battery is ended under actual conditions, namely the residual rated capacity of the storage battery is lower than a fixed numerical value, the reduction of the use capacity in different periods is closely related to current, temperature and depth, but the average failure rate of the leadacid battery is close to the same in different periods of the same environment, so that the establishment of a storage battery average failure rate model is an important premise for predicting the actual application life of the battery.
The failure coefficient β of the ith point is calculated according to the following formula_{mi}；
Wherein, C actual discharge capacity, ambient T temperature; t1i is the first discharge temperature. The mth discharge failure coefficient is calculated to be
β＝(β_{m1}+β_{m2}+…+β_{mN})/N (2)
The average rate of failure model of the storage battery calculated according to the formula is shown in the figure I, wherein N is the number of standard points of a battery discharge curve.
Step four: and comprehensively optimizing and disposing the test data by using a bellshaped function and a vector machine under the condition of calculating the failure rate of the battery.
The optimization method comprises the following steps: the battery failure rate is the rate at which the battery failure coefficient drops during the battery operating cycle.
The model is a bellshaped function y ═ ax^{2}+ b (wherein the value of a is changed), Y represents the remaining life cycle of the battery, and since the parameter that a is changed is not a fixed value, a vector machine function is adopted to solve a, and meanwhile, the derivative of the whole bellshaped function is obtained, and the failure rate in one cycle is obtained.
Establishing storage battery actual application life prediction models in different use environments; the method mainly comprises the following steps of battery power supply frequency, load current, power supply current stability and machine room environment temperature.
Calculating the failure rate of the battery based on the battery vulcanization mechanism, and establishing a predicted service life model expression as follows:
answering: the conversion relation between the battery failure rate and the failure coefficient is as follows:
where Δ β is the failure coefficient change, the period of the Δ T coefficient change,
whereinInfluence on the activity of the battery leading to vulcanization and ultimately to a reduction in the service life of the applicationA.b is the inherent property coefficient of the battery active matter and the frequency of the battery under different environmental use conditions;
under the condition that physical inherent factors such as battery active substances, design materials and the like are not considered, the relationship of the battery activity reducing factors is influenced, and the relationship among power supply time, current, temperature and capacity is as follows:
is a factor of the reaction depth between the active and the medium,a factor for supplying energy to participate in chemical reactions, a being the active motor capacity of the battery, whereinThe expression is as follows:
I_{10}standard current; t standard time; k_{T}A temperature correction coefficient; k_{i}＝It/I_{10}A current correction factor;
t real time period
When T is in the formula_{t}When greater than 10, set I_{10}.t_{10}Is C_{10},I_{t}.t_{t}Ct, the air conditioning regulation cannot be negative due to the environmental factors of the actual machine room, if the temperature is negativeThe activity of the battery is limited under zero, the performance is particularly poor, so that the selflearning battery charge state is equal to the initial quantity, and the relational expression of the application life and the residual capacity temperature of the battery is as follows:
when the battery failure coefficient is constant, according to the structure risk minimization principle and the least square vector machine theory:
the core of the optimization is the battery failure rate omega;
wherein ξ is an error variable,  ω! y^{2}Controlling the complexity of the model, C being a penalty factor of constant, b being a deviation, x_{i}A battery state of charge decay variable.
The Lagrange function of least square support vector machine model conversion represents that:
α therein_{i}(i ═ 1, 2.. times, l) is a lagrange multiplier;
the four variables ω, b, ξ, a are separately subjected to partial derivation by an optimization condition, namely, a lagrange function, so as to obtain:
the following can be obtained:
order toThe optimization problem is then transformed to solve the following system of linear equations:
wherein α ═ (α)_{1},α_{2},…,α_{1})^{T}，y＝(y_{1},y_{2},…,y_{1})^{T}。
B and a are calculated by a least square method, and the decision function of the LSSVM is obtained as follows:
due to K (x)_{i},x_{j}) And solving the regression prediction function by adopting internal operation for the inverse function of the kernel function which is an exponential function.
Substituting the experimental values into the vector machine formula yields the following values in table 2:
battery numbering

a value

b value

#1

0.975

95.2

#2

0.971

96.1

#3

0.982

95.5

#4

0.953

87.8 
And performing error optimization according to the data obtained in the table 2, and substituting the data into a life prediction mathematical model formula 8:
f(t)＝(0.798+ln^{Δc}).t^{2}+ΔT93.7 (17)
wherein t represents the operating period of the storage battery, since the service life of the operating battery is dynamically changed but the general trend is reduced, the number of detection periods is solved by substituting the experimental condition numerical value into the formula 17, and the inverse function is solved to obtain:
battery numbering

Operating life (period)

#1

24.5

#2

24

#3

25

#4

23.5 
The method for predicting the residual operation life of the battery is to calculate the failure rate of the battery according to a battery life model formula in the last detection period of the battery and predict the residual application period number.
Calculation of prediction error
The result predicted by any method has a certain difference with the actual value, and the difference between the predicted service life of the storage battery and the actual service life of the storage battery is the prediction error. The prediction error should reflect the accuracy of the prediction result, and the error value is inversely related to the accuracy. There are of course many different indicators of the calculation of the prediction error, and one is given below to evaluate the error predicted herein: relative Error Relative Percentage Error, RPE
Wherein Q is_{i}Is the actual measured value, f_{i}Is a predicted value.
Prediction result and error
The predicted values and the measured values are visually expressed as shown in fig. 2, and it can be seen from the figure that the predicted values obtained by using the principle of the least squares support vector machine and the measured values obtained by experiments have high consistency. The fitness function is calculated by the formula as:
according to the calculation results, the errors of the prediction result and the actual measurement result of the model do not exceed 10%, and the calculation result of the fitness is 0.0865, which also shows that the least square support vector machine can obtain a very accurate result in the application of the battery life prediction.
The above description is only for the preferred embodiment of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art should be considered to be within the technical scope of the present invention, and the technical solutions and the inventive concepts thereof according to the present invention should be equivalent or changed within the scope of the present invention.