A method for predicting the remaining life of a power battery

Abstract

The present invention relates to a method for predicting the remaining life of a power battery. First, an initial life decay core quantity diagram and a life decay subsidiary quantity correction diagram are constructed based on an existing power battery decay data set and correction relationship. Then, based on the real-time acquired data, a habitual preference-standard operating condition prediction model is combined and calculated to obtain the required data information. Finally, based on the required data information, a search is performed in the initial life decay core quantity diagram and the life decay subsidiary quantity correction diagram to obtain the health state loss of the power battery. The remaining service life of the power battery is calculated based on the health state loss. The present invention establishes a multi-dimensional comprehensive life decay model, which takes into account comprehensive influencing factors, has high model accuracy, and is highly consistent with the actual decay process of the power battery. At the same time, it has low requirements on hardware equipment, occupies less computing resources, and has the ability to quantify the remaining service life of the power battery in real time.

Description

Technical Field

[0001] The present invention relates to the technical field of power batteries, and in particular to a method for predicting the remaining life of a power battery. Background Art

[0002] The life decline of power batteries is a dynamic time-varying process with typical nonlinear behavior and complex internal mechanisms, which is affected by various internal and external factors. Especially under vehicle conditions, the instability of environmental factors and load conditions makes it extremely difficult to predict the remaining life of power batteries.

[0003] At present, there are two main methods for predicting the remaining life of automotive power batteries:

[0004] One is to quantify the life decline of the power battery based on the establishment of the remaining life model, and then predict its remaining service life. This method starts from exploring the factors at the mechanism level that affect the life decline of the power battery, and the reaction mechanism of the power battery is extremely complex. Therefore, this method has the problems of considering the factors that affect the life decline of the automotive power battery relatively one-sidedly, low accuracy in predicting the remaining life of the power battery, and a large difference from the actual remaining life of the power battery.

[0005] The other is to analyze the data related to the life of the power battery under vehicle conditions to explore the changing rules of the remaining life of the vehicle power battery through big data methods, and then predict its remaining life. This method requires the collection of a large amount of real vehicle data for analysis and processing, which occupies a lot of computing resources. The quality of the collected data depends on the accuracy of the sensor, and the hardware equipment is required to be high, so the cost is high. Summary of the invention

[0006] In view of this, the present invention provides a method for predicting the remaining life of a power battery, which takes into account comprehensive influencing factors, has high model accuracy, is highly consistent with the actual degradation of the power battery, occupies less computing resources, has low requirements for hardware equipment, and has extremely high potential for actual vehicle application.

[0007] To achieve the above object, the present invention provides the following solutions:

[0008] A method for predicting the remaining life of a power battery, comprising:

[0009] Acquire a decay data set of a power battery, and perform key decay point mining on the decay data set to obtain a key decay point data set;

[0010] Based on the key point data set, combined with the cycle life and discharge depth related data set, an initial life decay core quantity map is obtained;

[0011] Constructing a cycle life correction relationship, and obtaining a life decay accessory quantity correction diagram based on the cycle life correction relationship and in combination with a life decay accessory quantity threshold range;

[0012] Obtain real-time standard working conditions, and combine the habit preference-standard working condition prediction model to obtain the real-time driver habit preference;

[0013] Based on the real-time driver habit preference, a real-time standard vehicle speed is obtained; based on the real-time standard vehicle speed, a real-time demand data set is obtained;

[0014] Based on the real-time demand data set, in combination with the initial life decline core quantity diagram and the life decline subsidiary quantity correction diagram, a health state loss quantity is obtained;

[0015] The remaining operational life is obtained based on the health state loss amount.

[0016] Preferably, the acquiring of the decay data set of the power battery and performing key decay point mining on the decay data set to obtain the key decay point data set include:

[0017] Acquire an original decay test data set of power batteries of the same batch and the same model, and clean the original decay data set to obtain a cleaned data set;

[0018] Filling the cleaned data set based on the Gaussian process method so that the ratio of the final capacity to the rated capacity reaches a set ratio, thereby obtaining a filled data set;

[0019] Performing first-order derivative and second-order derivative of the filled data set in sequence, and obtaining a key decay point information set based on a critical change rate;

[0020] Based on a Gaussian distribution fitter and in combination with the key decay point information set, data extraction is performed on the filled data set to obtain the key decay point data set.

[0021] Preferably, the obtaining of the original decay test data set of power batteries of the same batch and the same model, and cleaning the original decay data set to obtain the cleaned data set, comprises:

[0022] Acquire a remaining capacity data set and a rated capacity of power batteries of the same batch and the same model, and obtain a health status data set based on the remaining capacity data set and the rated capacity;

[0023] Acquire a cycle life data set, construct a relationship between the health status data set and the cycle life data set, and obtain the original decay data set;

[0024] The original decay data set is double cleaned based on a dynamic filtering algorithm and a rainflow scoring algorithm to obtain the cleaned data set.

[0025] Preferably, the initial life decay core quantity diagram is obtained based on the key point data set and combined with the cycle life and discharge depth related data set, including:

[0026] Obtaining a life decay trajectory based on the cycle life and discharge depth correlation data set;

[0027] Based on the life decay trajectory and in combination with the key point data set, the initial life decay core quantity map is obtained.

[0028] Preferably, the constructing of the cycle life correction relationship, based on the cycle life correction relationship and in combination with the life decay accessory amount threshold range, obtains a life decay accessory amount correction diagram, including:

[0029] Construct the correction relationship between state of charge and cycle life; construct the correction relationship between charge and discharge rate and cycle life;

[0030] Constructing the cycle life correction relationship based on the state of charge versus cycle life correction relationship and the charge and discharge rate versus cycle life correction relationship;

[0031] Determine a state of charge threshold range and a charge and discharge rate threshold range; the life decay accessory amount threshold range includes the state of charge threshold range and the charge and discharge rate threshold range;

[0032] Based on the cycle life correction relationship, the life decay accessory amount correction diagram is obtained in combination with the state of charge threshold range and the charge and discharge rate threshold range.

[0033] Preferably, the real-time standard operating condition is acquired, and the real-time driver habit preference is obtained by combining the habit preference-standard operating condition prediction model, including:

[0034] Obtaining a preference training set and a preference test set, and constructing an initial habit preference-standard working condition prediction model; the preference training set and the preference test set are both habit preference-standard working condition mapping matrices;

[0035] Training the initial habit preference-standard operating condition prediction model based on the preference training set and the preference test set to obtain the habit preference-standard operating condition prediction model;

[0036] Acquire real-time vehicle speed data and historical vehicle speed data, and obtain the real-time standard operating condition by using a fuzzy reasoning method based on the real-time vehicle speed data and the historical vehicle speed data;

[0037] The real-time standard operating condition is input into the habit preference-standard operating condition prediction model to obtain the real-time driver habit preference.

[0038] Preferably, obtaining a real-time standard vehicle speed based on the real-time driver's habitual preference; and obtaining a real-time demand data set based on the real-time standard vehicle speed include:

[0039] Based on the real-time driver habit preference, obtaining the real-time standard vehicle speed;

[0040] Based on the real-time standard vehicle speed, obtaining the real-time required power of the power battery;

[0041] Acquire the real-time voltage of the power battery, and obtain the real-time required current based on the real-time required power and the real-time voltage;

[0042] Based on the real-time demand current and the rated capacity of the power battery, a real-time charge and discharge rate and a real-time state of charge are obtained;

[0043] A real-time discharge depth is obtained based on the real-time state of charge; the required data set includes a real-time charge and discharge ratio, the real-time state of charge and the discharge depth.

[0044] Preferably, the health state loss amount is obtained based on the real-time demand data set, in combination with the initial life decline core quantity diagram and the life decline subsidiary quantity correction diagram, including:

[0045] Based on the real-time discharge depth and the health state at the last moment, searching in the initial life decay core quantity graph to obtain the real-time cycle life;

[0046] Based on the real-time charge and discharge rate and the real-time state of charge, a search is performed in the life degradation accessory amount correction map to obtain a real-time comprehensive correction amount;

[0047] Based on the real-time cycle life and the real-time comprehensive correction amount, a real-time corrected cycle life is obtained;

[0048] Based on the end time of the real-time standard working condition, rolling update is performed to obtain the end corrected cycle life;

[0049] The health state loss amount is obtained based on the real-time corrected cycle life and the end corrected cycle life.

[0050] Preferably, the method further comprises:

[0051] The operating remaining life corresponding to several standard operating conditions existing at the current moment is weightedly summed to obtain a comprehensive remaining life.

[0052] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0053] The present invention relates to a method for predicting the remaining life of a power battery. First, an initial life decline core quantity diagram and a life decline subsidiary quantity correction diagram are constructed based on an existing power battery decline data set and a correction relationship; then, based on the real-time acquired data, a habitual preference-standard operating condition prediction model is combined and calculated to obtain the required data information; finally, based on the required data information, a search is performed in the initial life decline core quantity diagram and the life decline subsidiary quantity correction diagram to obtain the health state loss of the power battery, and the remaining service life of the power battery is calculated based on the health state loss. The present invention establishes a multi-dimensional comprehensive life decline model, which considers comprehensive influencing factors, has high model accuracy, and is highly consistent with the actual decline process of the power battery. At the same time, it has low requirements on hardware equipment, occupies less computing resources, and has the ability to quantify the remaining service life of the power battery in real time. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0055] Figure 1 This is a flow chart of a method for predicting the remaining life of a power battery according to the present invention;

[0056] Figure 2 This is a schematic diagram of the original decay test data set of the present invention;

[0057] Figure 3 This is a schematic diagram of cleaning the data set of the present invention;

[0058] Figure 4 A schematic diagram of filling a data set for the present invention;

[0059] Figure 5 This is the initial life decay core quantity diagram of the present invention;

[0060] Figure 6 This is a life-degradation accessory correction diagram of the present invention;

[0061] Figure 7 It is a schematic diagram of the standard vehicle speed simulation of the present invention;

[0062] Figure 8 This is a schematic diagram of power demand simulation of the present invention;

[0063] Fig. 9 This is a schematic diagram of the demand current simulation of the present invention;

[0064] Fig.10 This is a schematic diagram of the charge and discharge rate simulation of the present invention;

[0065] Fig.11 It is a schematic diagram of the state of charge simulation of the present invention;

[0066] Fig.12 It is a schematic diagram of the discharge depth simulation of the present invention;

[0067] Fig.13 This is a schematic diagram of the health status simulation of the present invention. DETAILED DESCRIPTION

[0068] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions 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 are within the scope of protection of the present invention.

[0069] The purpose of the present invention is to provide a method for predicting the remaining life of a power battery, which takes into account comprehensive influencing factors, has high model accuracy, is highly consistent with the actual degradation of the power battery, occupies less computing resources, has low requirements for hardware equipment, and has extremely high potential for real vehicle application.

[0070] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0071] Figure 1 The flowchart of the method for predicting the remaining life of a power battery of the present invention is shown in the figure. As shown in the figure, the present invention provides a method for predicting the remaining life of a power battery, comprising:

[0072] Step S1, obtaining a decay data set of a power battery, and mining key decay points on the decay data set to obtain a key decay point data set. Specifically, step S1 includes:

[0073] Step S11, obtaining the original decay test data set of the same batch and the same model of power batteries, and cleaning the original decay test data set to obtain a cleaned data set. In this embodiment, the power battery cycle life test data of Stanford University is selected to construct the original decay test data, and specifically 42 groups of test data are selected to constitute the original decay test data. The original decay test data is as follows: Figure 2 Further, the step S11 includes:

[0074] Step S111, obtaining a remaining capacity data set and a rated capacity of power batteries of the same batch and model, and obtaining a health status data set based on the remaining capacity data set and the rated capacity, specifically, dividing the remaining capacity data set by the rated capacity to obtain the health status data set.

[0075] Step S112, acquiring a cycle life data set, constructing a relationship between the health status data set and the cycle life data set, and obtaining the original decay data set.

[0076] Step S113, double cleaning the original decay data set based on a dynamic filtering algorithm and a rainflow scoring algorithm to obtain the cleaned data set.

[0077] Depend on Figure 2 It can be seen that some group data have capacity recovery effect, and the curve has abnormal burrs. Therefore, in this embodiment, double cleaning is performed based on the dynamic filtering algorithm and the rain flow scoring algorithm. The specific process is as follows:

[0078] First, the original decay data set is cleaned based on a dynamic filtering algorithm to obtain a first cleaned data set, filtering out drastic fluctuations in the decay trajectory of the power battery life, cleaning out abnormal data caused by the capacity recovery effect as much as possible, and retaining the main features of the original data to the greatest extent.

[0079] Then, based on the rain flow counting algorithm, the number of capacity recovery effects existing in the first cleaning data set and the severity of the capacity recovery effect are counted. Based on the rain flow counting algorithm, the health state regeneration phenomenon of each group of data is counted, and based on the depth of data abnormal events per unit health state and the number of data abnormal events per unit cycle, all data are scored, and a cleaning threshold is set. The data above the threshold is cleaned to obtain the second cleaning data set. For any group of first cleaning data in the first cleaning data set, the scoring rules are defined as follows:

[0080]

[0081] Where: Data i Cycle represents the number of healthy state regeneration phenomena and the corresponding depth of the first cleaning data of the i-th group based on the rain flow counting algorithm. i Indicates the number of health state regeneration phenomena, Depth i Indicates the corresponding depth, Score i It represents the score of the first cleaning data of the i-th group after statistics based on the rainflow counting algorithm, Test cycles represents the total number of test cycles of the first cleaning data of the i-th group, represents the maximum remaining capacity during the first cleaning data test of the i-th group, represents the minimum remaining capacity during the test of the i-th group of first cleaning data, ∑ represents a summation symbol, and N represents the total number of first cleaning data in the first cleaning data set.

[0082] The second cleaning data set is standardized to obtain the cleaning data set. Figure 3 As shown, the standardization processing formula is as follows:

[0083]

[0084] Where: Score represents the standardized score of the second cleaned data set of the i-th group. i It represents the score of the second cleaning data of the i-th group after statistics based on the rain flow counting algorithm, min(Score i ) represents the minimum score in the second cleaning data of the i-th group, max(Score i ) represents the maximum score in the second cleaning data of the i-th group.

[0085] Step S12, filling the cleaned data set based on the Gaussian process method so that the ratio of the final capacity to the rated capacity reaches a set ratio, thereby obtaining a filled data set.

[0086] The cleaned data set is divided into a filling training set and a filling test set. The filling training set is trained by Gaussian process to form a power battery life decline predictor. The filling test set is used to verify the effect of the power battery life decline predictor and the error is fed back to the power battery life decline predictor. The parameters are adjusted repeatedly until the required prediction accuracy is achieved, thereby completing the data filling. In addition, in order to achieve the purpose of training and testing at the same time, the filling training set is further divided into an effect training set and an effect test set in a ratio of 9:1, and the training effect of the power battery life decline predictor is observed repeatedly to generate the final power battery life decline predictor, thereby completing the data filling.

[0087] In this embodiment, the set ratio is 0.8, and the set ratio is defined as the end of the life of the power battery. Figure 4 shown.

[0088] Step S13, performing first-order derivative and second-order derivative on the filled data set in sequence, and obtaining a key decay point information set based on a critical change rate.

[0089] Step S14, based on a Gaussian distribution fitter and in combination with the key decay point information set, data extraction is performed on the filled data set to obtain the key decay point data set.

[0090] For each set of key decay point information in the key decay point information set, the relationship between each key decay point and the end of life is defined as follows:

[0091]

[0092] Where P ti Indicates the relationship between the historical cycle life corresponding to the i-th key decay point in the k-th group of key decay point information and the total cycle life of the power battery; CyclesSOH=k represents the cycle life of the k-th key decay point, Dead Cycles It represents the cycle life corresponding to the decay of the power battery to the end of its life, and k∈M represents the position label of the key decay point.

[0093] Then the weight vector with the health status of the power battery as the state variable is obtained:

[0094]

[0095] Where V k represents the weight vector of the kth key decay point, and SOH represents the battery health state of the power battery.

[0096] Based on the weight vector and the key decay point information set, a Gaussian distribution fitter is used to extract data from the filled data set to obtain a key decay point data set.

[0097] Step S2, based on the key point data set, combined with the cycle life and discharge depth related data set, obtain an initial life decay core quantity map.

[0098] Preferably, step S2 comprises:

[0099] Step S21, obtaining a life decay trajectory based on the cycle life and discharge depth correlation data set.

[0100] Step S22, based on the life decay trajectory and in combination with the key point data set, obtain the initial life decay core quantity map.

[0101] The initial life decay core quantity diagram can quantify the relationship between the health status of the power battery and the discharge depth and cycle life, and can locate the nonlinear decay trajectory of the power battery life. The initial life decay core quantity diagram is as follows: Figure 5 As shown, Figure 5 middle, Figure 5 (a) is a three-dimensional form of the initial life decay core quantity map, Figure 5 (b) is a two-dimensional form of the initial life decay core quantity diagram, which represents the relationship between the health state and cycle life of the power battery, Figure 5(c) is a two-dimensional form of the initial life decay core quantity diagram, which represents the relationship between the health state of the power battery and the discharge depth, Figure 5 (d) is a two-dimensional form of the initial life decay core quantity diagram, which represents the relationship between the discharge depth and cycle life of the power battery.

[0102] Step S3, constructing a cycle life correction relationship, and obtaining a life decay accessory amount correction diagram based on the cycle life correction relationship and in combination with a life decay accessory amount threshold range.

[0103] Specifically, step S3 includes:

[0104] Step S31, constructing a correction relationship between state of charge and cycle life; constructing a correction relationship between charge and discharge rate and cycle life.

[0105] The state of charge to cycle life correction relationship is as follows:

[0106]

[0107] Where: SOC represents the state of charge of the power battery, A i represents the polynomial coefficient, ∑ represents the summation symbol, f() represents the correction relationship between the state of charge of the power battery and the cycle life, Q=3, A3=8.0401, A2=-4.28741, A1=-1.3087, A0=1.7263.

[0108] The charge-discharge rate and cycle life correction relationship is as follows:

[0109]

[0110] Where: Crate represents the charge and discharge rate of the power battery, g() represents the correction relationship between the charge and discharge rate of the power battery and the cycle life. e represents an exponential function with a constant e as the base.

[0111] Step S32, constructing the cycle life correction relationship based on the state of charge versus cycle life correction relationship and the charge and discharge rate versus cycle life correction relationship. The cycle life correction relationship is as follows:

[0112] σ(SOC,Crate)=f(SOC)·g(Crate);

[0113] Where: σ() represents the correction relationship between the state of charge and charge / discharge rate of the power battery and the cycle life.

[0114] Step S33, determining a state of charge threshold range and a charge and discharge rate threshold range; the life degradation accessory amount threshold range includes the state of charge threshold range and the charge and discharge rate threshold range.

[0115] In this embodiment, the state of charge SOC threshold range is: 0≤SOC≤1, and the charge and discharge rate Crate threshold range is: 0≤Crate≤10.

[0116] Step S34, based on the cycle life correction relationship, combined with the state of charge threshold range and the charge and discharge rate threshold range, obtain the life decline accessory correction map. The life decline accessory correction map is as follows: Figure 6 The charge / discharge rate in the figure is the charge / discharge rate.

[0117] Step S4, obtaining the real-time standard operating condition, combining the habit preference-standard operating condition prediction model, and obtaining the real-time driver habit preference.

[0118] Preferably, the step S4 comprises:

[0119] Step S41, obtaining a preference training set and a preference test set, and constructing an initial habit preference-standard operating condition prediction model; the preference training set and the preference test set are both habit preference-standard operating condition mapping matrices. The initial habit preference-standard operating condition prediction model uses a GoogleNet neural network model.

[0120] The habitual preference-standard working condition mapping matrix is ​​as follows:

[0121] Ξ=[Ξ 1 Ξ 2 …Ξ j …Ξ n ]

[0122] Where: j = 1, 2, ..., n represents the total number of standard operating conditions, x = 1, 2, ...X represents the total number of times the driver steps on the accelerator pedal, y = 1, 2, ...Y represents the total number of times the driver steps on the brake pedal, represents the moment when the driver steps on the accelerator pedal under the jth standard operating condition, It represents the intensity of the driver's action when pressing the accelerator pedal under the jth standard working condition. represents the moment when the driver steps on the brake pedal under the jth standard working condition, represents the intensity of the driver's braking action under the jth standard working condition, j represents the driver habit preference-jth standard operating condition mapping matrix, and Ξ represents the driver habit preference-standard operating condition mapping matrix under typical standard operating conditions with a total number of n.

[0123] Step S42: training the initial habit preference-standard operating condition prediction model based on the preference training set and the preference test set to obtain the habit preference-standard operating condition prediction model.

[0124] Step S43, acquiring real-time vehicle speed data and historical vehicle speed data, and based on the real-time vehicle speed data and the historical vehicle speed data, using a fuzzy reasoning method to obtain the real-time standard operating condition.

[0125] Step S44, inputting the real-time standard operating condition into the habit preference-standard operating condition prediction model to obtain the real-time driver habit preference.

[0126] Step S5, obtaining a real-time standard vehicle speed based on the real-time driver's habitual preference; and obtaining a real-time demand data set based on the real-time standard vehicle speed.

[0127] Specifically, step S5 includes:

[0128] Step S51, based on the real-time driver's habitual preference, obtain the real-time standard vehicle speed. The simulation diagram of the real-time standard vehicle speed is as follows: Figure 7 shown.

[0129] Step S52, based on the real-time standard vehicle speed, obtain the real-time power demand of the power battery. The simulation diagram of the real-time current demand is as follows: Figure 8 shown.

[0130] The calculation formula of the real-time required power is as follows:

[0131]

[0132] Where: P Battery (t) represents the power demand of the power battery at time t, in kW, P Demand (t) represents the vehicle power requirement in kW, v(t) represents the standard vehicle speed at time t in km / h, and F a (t) represents the air resistance of the vehicle at time t, in N, F r (t) represents the rolling resistance of the vehicle at time t, in N, F g (t) represents the slope resistance of the vehicle at time t, in N, η DC / AC represents the efficiency of the DC-AC inverter, η Motor represents the efficiency of the motor, m(t) represents the total mass of the vehicle at time t, in kg, C D Represents the vehicle's drag coefficient, A represents the windward area, and the unit is m 2 , f represents the rolling resistance coefficient, θ represents the slope resistance coefficient, and g represents the acceleration of gravity, the unit is m / s 2 .

[0133] During the operation of the vehicle, the mass change in a short period of time is mainly due to the consumption of fuel, so the mass change is negligible, that is, m(t) is regarded as a constant, so the above formula changes to:

[0134]

[0135] Step S53, obtaining the real-time voltage of the power battery, and obtaining the real-time demand current based on the real-time demand power and the real-time voltage. The simulation diagram of the real-time demand current is as follows: Fig. 9 The calculation formula of the real-time demand current is as follows:

[0136]

[0137] Where: I Battery (t) represents the required current of the power battery at time t, U Battery (t) represents the voltage of the power battery at time t.

[0138] Step S54, based on the real-time demand current and the rated capacity of the power battery, obtain the real-time charge and discharge rate and the real-time state of charge. The simulation diagram of the real-time charge and discharge rate is as follows: Fig.10 As shown, Fig.10 The charge / discharge rate in the above is the charge / discharge rate. The simulation diagram of the real-time state of charge is as follows: Fig.11 shown.

[0139] The calculation formula of the real-time charge and discharge rate is as follows:

[0140]

[0141] Where: Crate(t) represents the charge and discharge rate of the power battery at time t, and Q0 represents the rated capacity of the power battery.

[0142] The calculation formula of the real-time state of charge is as follows:

[0143]

[0144] Where: SOC(t) represents the state of charge of the power battery at time t, It indicates that the demand current for the power battery is integrated during the time period from 0 to t.

[0145] Step S55, obtaining a real-time discharge depth based on the real-time state of charge; the required data set includes a real-time charge and discharge rate, the real-time state of charge and the discharge depth. The simulation diagram of the real-time discharge depth is as follows: Fig.12 shown.

[0146] The calculation formula of the real-time discharge depth is as follows:

[0147] DOD(t)=1-SOC(t);

[0148] Where: DOD(t) represents the discharge depth of the power battery at time t.

[0149] Step S6, based on the real-time demand data set, combined with the initial life decline core quantity diagram and the life decline subsidiary quantity correction diagram, obtain the health status loss amount.

[0150] Further, the step S6 comprises:

[0151] Step S61, based on the real-time discharge depth and the health state at the last moment, search in the initial life decay core quantity diagram to obtain the real-time cycle life. The simulation diagram of the health state is as follows Fig.13 as shown.

[0152] Step S62, based on the real-time charge and discharge rate and the real-time state of charge, searching in the life degradation accessory amount correction map to obtain a real-time comprehensive correction amount.

[0153] Step S63, obtaining the real-time corrected cycle life based on the real-time cycle life and the real-time comprehensive correction amount.

[0154] Step S64, based on the end time of the real-time standard working condition, rolling update is performed to obtain the end modified cycle life. Specifically, based on the end time of the real-time standard working condition, the method of "steps S61-S63" is used to perform rolling update to obtain the end modified cycle life.

[0155] Step S65, based on the real-time modified cycle life and the end modified cycle life, the health state loss is obtained. The real-time modified cycle life and the end modified cycle life are subtracted to obtain the health state loss.

[0156] Step S7, obtaining the remaining operating life based on the health state loss. The remaining operating life is calculated as follows:

[0157]

[0158] Where: T residual,j Indicates the remaining service life of the power battery corresponding to the current standard operating conditions, t end,j Indicates the start time of the current standard working condition, t end,j Indicates the end time of the current standard working condition, ΔSOH j Indicates the amount of health status loss after running the current standard operating condition.

[0159] Optionally, based on the fact that the actual operating condition of the vehicle during actual operation is composed of a plurality of standard operating conditions, the method of the present invention further includes:

[0160] The operating remaining life corresponding to several standard operating conditions existing at the current moment is weightedly summed to obtain a comprehensive remaining life.

[0161] The health status SOH(t end )=0.999983889449993, and the power battery health status loss after a complete standard working condition of typical suburban conditions in China is ΔSOH=1.611055000672046×10 -5 .

[0162] Combined with the end point of the power battery's life decline in the case of vehicle use, when its health state declines to 80% of the initial value, the remaining service life of the power battery under this standard condition is quantified as:

[0163] T residual =1.63×10 7 Seconds = 4530.82 Hours. Since only a single typical standard operating condition is considered in the example, there is no weight distribution operation. Under the standard operating condition of typical suburban operating conditions in China, the vehicle can still run for 4530.82 hours. If the vehicle works 10 times in this operating condition per day, the vehicle can still run for 1241.32 days. So far, the remaining life of the vehicle's power battery has been predicted.

[0164] The present invention can not only quantify the influence of discharge depth on the life degradation of power batteries, but also take into account the correction of the charge / discharge rate and charge state of the power battery on the quantification of the life degradation of the power battery. The influencing factors considered are comprehensive, the model is highly accurate, and the degree of consistency with the actual degradation process of the power battery is high. In addition, the present invention takes into account the prediction of the remaining service life of automotive power batteries under different operating conditions, which is more in line with actual vehicle usage.

[0165] The present invention can quantify the influence of various factors on the life decline of the power battery based on a table lookup method, has low requirements on hardware equipment, occupies less computing resources, has the ability to quantify the remaining service life of the power battery in real time online, and has the potential for real vehicle application.

[0166] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0167] The principles and implementation methods of the present invention are described in this article using specific examples. The description of the above embodiments is only used to help understand the method and core idea of ​​the present invention. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A method for predicting the remaining life of a power battery, characterized in that: include: Acquire a decay data set of a power battery, and perform key decay point mining on the decay data set to obtain a key decay point data set; Based on the key decay point data set, combined with the cycle life and discharge depth related data set, an initial life decay core quantity map is obtained; Constructing a cycle life correction relationship, based on the cycle life correction relationship, combined with the life decay accessory amount threshold range, obtaining a life decay accessory amount correction diagram; constructing a state of charge versus cycle life correction relationship; constructing a charge and discharge rate versus cycle life correction relationship; Constructing the cycle life correction relationship based on the state of charge versus cycle life correction relationship and the charge and discharge rate versus cycle life correction relationship; Obtain real-time standard working conditions, and combine the habit preference-standard working condition prediction model to obtain the real-time driver habit preference; Based on the real-time driver habit preference, obtaining a real-time standard vehicle speed; Obtaining a real-time demand data set based on the real-time standard vehicle speed; Based on the real-time demand data set, in combination with the initial life decline core quantity diagram and the life decline subsidiary quantity correction diagram, a health state loss quantity is obtained; The remaining operational life is obtained based on the health state loss amount.

2. The method for predicting the remaining life of a power battery according to claim 1, characterized in that: The step of acquiring a decay data set of a power battery and mining key decay points of the decay data set to obtain a key decay point data set includes: Acquire an original decay test data set of power batteries of the same batch and the same model, and clean the original decay test data set to obtain a cleaned data set; Filling the cleaned data set based on the Gaussian process method so that the ratio of the final capacity to the rated capacity reaches a set ratio, thereby obtaining a filled data set; Performing first-order derivative and second-order derivative of the filled data set in sequence, and obtaining a key decay point information set based on a critical change rate; Based on a Gaussian distribution fitter and in combination with the key decay point information set, data extraction is performed on the filled data set to obtain the key decay point data set.

3. The method for predicting the remaining life of a power battery according to claim 2, characterized in that: The obtaining of the original decay test data set of power batteries of the same batch and the same model, and cleaning the original decay test data set to obtain a cleaned data set, includes: Acquire a remaining capacity data set and a rated capacity of power batteries of the same batch and the same model, and obtain a health status data set based on the remaining capacity data set and the rated capacity; Acquire a cycle life data set, construct a relationship between the health status data set and the cycle life data set, and obtain the original decay test data set; The original decay test data set is double cleaned based on a dynamic filtering algorithm and a rainflow scoring algorithm to obtain the cleaned data set.

4. The method for predicting the remaining life of a power battery according to claim 1, characterized in that: Based on the key decay point data set, combined with the cycle life and discharge depth related data set, the initial life decay core quantity map is obtained, including: Obtaining a life decay trajectory based on the cycle life and discharge depth correlation data set; Based on the life decay trajectory and in combination with the key decay point data set, the initial life decay core quantity map is obtained.

5. The method for predicting the remaining life of a power battery according to claim 1, characterized in that: The constructing of the cycle life correction relationship, based on the cycle life correction relationship and in combination with the life decay accessory amount threshold range, obtains a life decay accessory amount correction diagram, including: Construct the correction relationship between state of charge and cycle life; construct the correction relationship between charge and discharge rate and cycle life; Constructing the cycle life correction relationship based on the state of charge versus cycle life correction relationship and the charge and discharge rate versus cycle life correction relationship; Determine a state of charge threshold range and a charge and discharge rate threshold range; the life decay accessory amount threshold range includes the state of charge threshold range and the charge and discharge rate threshold range; Based on the cycle life correction relationship, the life decay accessory amount correction diagram is obtained in combination with the state of charge threshold range and the charge and discharge rate threshold range.

6. The method for predicting the remaining life of a power battery according to claim 1, characterized in that: The real-time standard operating condition is obtained, and the real-time driver habit preference is obtained by combining the habit preference-standard operating condition prediction model, including: Obtaining a preference training set and a preference test set, and constructing an initial habit preference-standard working condition prediction model; the preference training set and the preference test set are both habit preference-standard working condition mapping matrices; Training the initial habit preference-standard operating condition prediction model based on the preference training set and the preference test set to obtain the habit preference-standard operating condition prediction model; Acquire real-time vehicle speed data and historical vehicle speed data, and obtain the real-time standard operating condition by using a fuzzy reasoning method based on the real-time vehicle speed data and the historical vehicle speed data; The real-time standard operating condition is input into the habit preference-standard operating condition prediction model to obtain the real-time driver habit preference.

7. The method for predicting the remaining life of a power battery according to claim 1, characterized in that: The real-time standard vehicle speed is obtained based on the real-time driver's habit preference; A real-time demand data set is obtained based on the real-time standard vehicle speed, including: Based on the real-time driver habit preference, obtaining the real-time standard vehicle speed; Based on the real-time standard vehicle speed, obtaining the real-time required power of the power battery; Acquire the real-time voltage of the power battery, and obtain the real-time required current based on the real-time required power and the real-time voltage; Based on the real-time demand current and the rated capacity of the power battery, a real-time charge and discharge rate and a real-time state of charge are obtained; A real-time discharge depth is obtained based on the real-time state of charge; the required data set includes a real-time charge and discharge ratio, the real-time state of charge and the discharge depth.

8. The method for predicting the remaining life of a power battery according to claim 7, characterized in that: The health state loss amount is obtained based on the real-time demand data set, in combination with the initial life decline core quantity diagram and the life decline subsidiary quantity correction diagram, including: Based on the real-time discharge depth and the health state at the last moment, searching in the initial life decay core quantity graph to obtain the real-time cycle life; Based on the real-time charge and discharge rate and the real-time state of charge, a search is performed in the life degradation accessory amount correction map to obtain a real-time comprehensive correction amount; Based on the real-time cycle life and the real-time comprehensive correction amount, a real-time corrected cycle life is obtained; Based on the end time of the real-time standard working condition, rolling update is performed to obtain the end corrected cycle life; The health state loss amount is obtained based on the real-time corrected cycle life and the end corrected cycle life.

9. The method for predicting the remaining life of a power battery according to claim 1, characterized in that: The method further comprises: The operating remaining life corresponding to several standard operating conditions existing at the current moment is weightedly summed to obtain a comprehensive remaining life.

Images