A method for evaluating the stability of power batteries based on projection correlation

Abstract

The present invention relates to the technical field of power battery performance evaluation, and discloses a power battery stability evaluation method based on projection correlation, comprising the following steps: Step 1: Collect basic data; the basic data includes voltage signal data and current signal data of the power battery; Step 2: Preprocess the basic data to obtain target data; Step 3: Extract safety elements from the target data; Step 4: Quantify the safety elements and obtain safety quantification features; Step 5: Identify the safety state based on the safety quantification features. The present invention can be used to solve the technical problem of inaccurate evaluation of the existing power battery state, and can achieve a relatively high accuracy in evaluating the characteristics of the power battery and a relatively high accuracy in evaluating the state of the power battery.

Description

Technical Field

[0001] The present invention relates to the technical field of power battery performance evaluation, and particularly to a method for evaluating the stability of a power battery based on projection correlation. Background Art

[0002] With the continuous acceleration of the process of new and old kinetic energy conversion, new energy has gradually penetrated into all walks of life. In the automotive industry, the development trend of new energy vehicles is excellent, and more and more car owners choose to drive new energy vehicles for travel. At the same time, the safety issues of new energy vehicles have also received extensive attention. Among them, as an important or the only energy source of new energy vehicles, the operation safety of the power battery system is closely related to the safety issues of new energy vehicles, and the state stability problem of the power battery urgently needs attention.

[0003] In response to this, existing methods for evaluating the state of power batteries often analyze the historical data of power batteries to determine the state of power batteries. For example, based on historical data, by comparing thresholds, analyzing numerical fluctuations, or performing conventional mechanism analysis, to determine battery stability and battery operating status, etc. However, with the current diversified development of new energy vehicles, the charging structure of new energy vehicles has been continuously optimized, the fast charging volume has been continuously increased, the charging power has been continuously improved, and the operating scenarios have become more and more variable. This series of improvements and changes have made the operating environmental conditions of power batteries also continuously change, and the data performance of power batteries has become more and more multi-dimensional and complex, bringing great challenges to exploring and mining the battery safety status and vehicle safety status contained in operation big data. It is difficult to achieve the state evaluation of the power battery of a new energy vehicle with a strongly coupled complex system through mechanism analysis, and the existing methods for evaluating the state of power batteries cannot adapt to the actual environment of the current operation of power batteries, resulting in a low accuracy of state evaluation. Summary of the Invention

[0004] The present invention aims to provide a method for evaluating the stability of a power battery based on projection correlation, which is used to solve the technical problem of inaccurate state evaluation of existing power batteries, and can achieve a high accuracy of power battery characteristic evaluation and a high accuracy of power battery state evaluation.

[0005] The basic solution provided by the present invention is as follows: A method for evaluating the stability of a power battery based on projection correlation, comprising the following steps:

[0006] Step 1: Collect basic data; the basic data includes voltage signal data and current signal data of the power battery;

[0007] Step 2: Preprocess the basic data to obtain target data;

[0008] Step 3: Extract safety factors from the target data; when extracting safety factors, calculate and extract according to the following formula:

[0009] s = (V d I) 2 / 2 ;

[0010] wherein, V d is the median voltage difference of the battery cells of the power battery; I is the total current of the power battery; s is the safety factor characterizing abnormality;

[0011] Step 4: Quantify the safety factor and obtain the safety quantification feature; when quantifying the safety factor, the quantification is performed according to the following formula:

[0012] p = 1 - λ;

[0013] wherein, λ is the variance entropy, λ = E 2 (s) / ( 2 ) and 0 ≤ λ ≤ 1; p is the safety quantification feature;

[0014] Step 5: Identify the safety state based on the safety quantification feature.

[0015] The working principle and advantages of the present invention are as follows: First, the basic data collected in this solution includes voltage signal data and current signal data. These two types of data are a kind of characterization quantity of the comprehensive state of the battery, covering rich battery state information, capable of providing a sufficient basis for battery feature analysis, and the data is relatively easy to collect and extract, and is relatively easy to operate.

[0016] Second, this solution specifically defines the safety factor and the safety quantification feature, providing a brand-new quantification analysis method and safety state evaluation method. First, this solution discovers the internal guiding reason for the low evaluation accuracy in the existing power battery state evaluation. That is, affected by the current good development trend of new energy vehicles, the functions of new energy vehicles are becoming more and more abundant, and their application scenarios are becoming more and more changeable and complex. Correspondingly, the data performance of the relevant operation data of new energy vehicles, especially the relevant operation data of their key power source - the power battery, is becoming more and more complex and multi-dimensional. In this case, combined with the physical characteristics of the battery system itself and factors such as the design and acquisition accuracy of relevant sensors, there must be information coupling, redundancy and errors between different signal data, and this kind of information coupling, redundancy and errors are often closely related to the signal data and hidden in the conventional signal data, which is more difficult to be discovered and excluded compared with the conventional abnormal data with large or small values or large fluctuations; and this is exactly the internal guiding reason for the low evaluation accuracy ignored by the existing power battery state evaluation; the conventional evaluation scheme does not discover the above data error hidden dangers and cannot handle such hidden dangers.

[0017] Secondly, this scheme provides an effective solution to the above-mentioned intrinsic guiding reasons. This scheme overcomes the technical difficulties in dealing with the above-mentioned intrinsic guiding reasons and proposes a breakthrough quantitative analysis method based on projection correlation. Specifically, this scheme specifically defines safety factors, and extracts safety factors from voltage and current signal data according to specific formulas, wherein the safety factor-related formula is designed based on an equivalent circuit model, which characterizes the projection characteristics of the voltage in the current direction, namely, the projection correlation coefficient. This coefficient (i.e., safety factor) can fully reflect the correlation between the voltage signal and the current signal and express it quantitatively, and this correlation is a pure linear correlation between the voltage signal and the current signal, which is relatively unaffected by external factors such as the battery system, working conditions, and environment. By calculating and extracting this coefficient (i.e., safety factor), it is possible to effectively exclude redundant information and noise information generated by external factors in the power battery characterization signal, and more accurately extract the signal characteristics that characterize the power battery operating status. Signal characteristics obscured by noise redundancy can also be accurately extracted. The obtained safety factor can effectively and accurately describe the changes in voltage and current during vehicle operation, that is, during the operation of the power battery, and quantitatively describe the evolution trend of abnormal battery fault characteristics, that is, the fault formation process, so as to achieve a higher accuracy in power battery characteristic assessment, realize early identification of faults, and effectively improve the accuracy of subsequent power battery safety status assessment.

[0018] Further, in step 2, the preprocessing operation includes: removing abnormal data from the basic data; the abnormal data is data in the basic data that exceeds a first specified threshold.

[0019] Beneficial effect: Screening out excessive signal values ​​can effectively prevent the excessive signal values ​​from affecting the subsequent safety factor extraction process, and ensure the reliability of subsequent safety factor calculations.

[0020] Further, in step 2, the preprocessing operation also includes: identifying and marking special data; when identifying special data, if the difference between the current frame voltage data and the previous frame voltage data exceeds the second specified threshold, the current frame voltage data is determined to be special data and the current frame voltage data is marked; if the difference between the current frame timestamp data and the previous frame timestamp data exceeds the third specified threshold, the current frame timestamp data is determined to be special data and the current frame timestamp data is marked.

[0021] Beneficial effects: The special data types determined and marked are interference pulses (confirmed by the voltage data difference) and time discontinuity points (confirmed by the timestamp data difference). Among them, the interference pulses mark the positions where the data has abnormal jumps, and the time discontinuity points mark the positions where there are long intervals in the vehicle operation data. These two situations are often difficult to accurately evaluate. For these two types of data, after the subsequent safety quantification features are quantified, the risks at the marked abnormal positions can be set to 0, without considering the risks here, and at the same time, it will not introduce artificial errors caused by data cleaning; the data preprocessing operation has a high degree of meticulousness.

[0022] Further, in step 2, the preprocessing operation further includes: performing mean filtering on the basic data.

[0023] Beneficial effects: Mean filtering is a typical linear filtering algorithm. Using mean filtering to process the basic data can effectively weaken the noise in the basic data, and at the same time, the operation steps are simple, which can ensure a high processing efficiency.

[0024] Further, the V d is determined according to the following steps:

[0025] S1: Select the target battery cell;

[0026] S2: Calculate the difference between each frame of voltage data of the target battery cell and the median of the single-cell voltage, and then obtain the median voltage difference of the target battery cell, which is V d .

[0027] Beneficial effects: Selecting a specific battery cell to calculate the median voltage difference, the obtained voltage difference data is more representative, can fully characterize the battery characteristics, and further can make the reliability and accuracy of the safety factor data extracted and calculated based on this voltage difference data higher.

[0028] Further, in step 5, when identifying the safety state, it is identified according to a preset identification strategy; the preset identification strategy includes the following sub-steps:

[0029] Sub-step 1: Based on the safety quantification features, perform risk cumulative calculation according to the following formula and obtain the risk cumulative index:

[0030] Sp = ∑p;

[0031] where Sp is the risk cumulative index; p is the safety quantification feature;

[0032] Sub-step 2: Based on the risk cumulative index, determine the safety state score;

[0033] Sub-step 3: Confirm the battery risk level according to the safety state score.

[0034] Beneficial effects: Discretely integrate the safety quantification features on the time scale to obtain a risk accumulation index, which can fully represent the change and fluctuation of the safety factors of the battery within a certain period of time. Using this index, the safety state score of the battery is evaluated and determined for risk level assessment. Further, the battery risk is quantified, the risk is presented more intuitively, and the safety state of the battery is presented more intuitively.

[0035] Further, in sub-step 2, the safety state score is calculated according to the following formula:

[0036] v = w1 * η + w2 * Q + w3 * R; and 0 ≤ v ≤ 1;

[0037] Wherein, Q is the relative risk, Q = z max , z is the slope value of the Sp curve; R is the risk frequency, and the R value is the number of values in z greater than the fourth specified threshold; η is the absolute risk, η = z a / z b , z a is the mean value of the larger part in z, z b is the mean value of the smaller part in z; w1, w2, and w3 are the weight coefficients of η, Q, and R respectively.

[0038] Beneficial effects: Based on the numerical extraction of the function curve corresponding to the risk accumulation index from multiple aspects, the risk-related reference information contained in the risk accumulation index is fully explored, which can ensure the comprehensiveness and reliability of the safety state score evaluation; and each risk-related reference information has its corresponding weight allocation, which can ensure the meticulousness and accuracy of the safety state score evaluation.

[0039] Further, in sub-step 3, the safety state score is directly proportional to the battery risk level. The larger the safety state score, the higher the battery risk level.

[0040] Beneficial effects: This solution clearly defines the corresponding relationship between the safety state score and the battery risk level, and the risk level assessment is reliable. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 It is a schematic flowchart of the method of an embodiment of a power battery stability evaluation method based on projection correlation of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0042] The following is a more detailed description through specific embodiments:

[0043] The embodiment is basically as shown in the attached Figure 1 : A power battery stability evaluation method based on projection correlation, characterized by including the following steps:

[0044] Step 1: Collect basic data; the basic data includes voltage signal data and current signal data of the power battery.

[0045] Specifically, in this embodiment, the basic data is parsed from the message log of the power battery that complies with the GB32960 protocol; the data obtained in this way is complete, perfect, and compliant with the standard, and has a high reliability.

[0046] Step 2: Preprocess the basic data to obtain target data;

[0047] The preprocessing operations include:

[0048] (1) Remove abnormal data from the basic data; the abnormal data is the data in the basic data that exceeds the first specified threshold. In this embodiment, the first specified threshold is set to [2.5, 4.25], and the specified threshold is accurately set, which can fully clean the abnormal data.

[0049] (2) Identify and mark special data; when identifying special data, if the difference between the current frame voltage data and the previous frame voltage data exceeds the second specified threshold, it is determined that the current frame voltage data is special data and the current frame voltage data is marked; if the difference between the current frame timestamp data and the previous frame timestamp data exceeds the third specified threshold, it is determined that the current frame timestamp data is special data and the current frame timestamp data is marked. In this embodiment, the second specified threshold is set to 3s d ; s d is the standard deviation of the voltage change rate. The third specified threshold is set to 120s. The specified threshold is accurately set, which can fully capture special data.

[0050] In addition, in this embodiment, for the marked special data, after the subsequent safety quantization feature quantization is completed, the risk at the marked abnormal position is set to 0, and the risk here is not considered, so as to avoid introducing artificial errors caused by data cleaning; the data preprocessing operation has a high degree of meticulousness.

[0051] (3) Perform mean filtering processing on the basic data.

[0052] Step 3: Extract safety elements from the target data; when extracting safety elements, calculate and extract according to the following formula:

[0053] s = (V d I) 2 / 2 ;

[0054] where V d is the median voltage difference of the battery cells of the power battery; I is the total current of the power battery; s is the safety element representing abnormality, that is, the eigenvector of the projection part of the voltage of the power battery in the current direction, that is, the projection correlation coefficient.

[0055] Specifically, the V d is determined according to the following steps:

[0056] S1: Select the target battery cell;

[0057] In this embodiment, the target battery cell is selected according to the following method:

[0058] First, calculate the difference value V of the single - cell voltages of each battery cell in the power battery d , and then mark the sampling points in sequence according to the following sign function.

[0059]

[0060] In the above formula, - 1 and 1 respectively represent the abnormality of the voltage being too small and the abnormality of the voltage being too large, and 0 represents normal; α is the threshold coefficient for defining abnormality, which is a constant; V a is the range voltage, and V a is equal to the difference between the maximum value and the minimum value of the single - cell voltage at any moment.

[0061] Calculate the number of abnormal - marked points for each battery cell according to the above formula, and then select the battery cell with the most abnormal - marked points as the target battery cell.

[0062] The target battery cell selected according to this method has strong representativeness, can fully characterize the battery characteristics, and helps to obtain a more reliable V value that can be used for safety factor calculation. d value.

[0063] S2: Calculate the difference between each frame of voltage data of the target battery cell and the median of the single - cell voltages, and then obtain the median voltage difference of the target battery cell, which is V d .

[0064] Step 4: Quantify the safety factor and obtain the safety quantification feature; when quantifying the safety factor, it is quantified according to the following formula:

[0065] p = 1 - λ;

[0066] where λ is the variance entropy, λ = E 2 (s) / ( 2 ) and 0 ≤ λ ≤ 1, and the closer λ is to 1, it means that on the time scale, the fluctuation degree of the safety factor of the specified battery cell is smaller, and the battery state is safer. p is the safety quantification feature.

[0067] Step 5: Identify the safety state based on the safety quantification feature.

[0068] When identifying the safety state, it is identified according to the preset identification strategy. The preset identification strategy includes the following sub - steps:

[0069] Sub-step 1: Based on the safety quantification features, perform risk cumulative calculation according to the following formula to obtain the risk cumulative index:

[0070] The risk cumulative index Sp = ∑p;

[0071] where Sp is the risk cumulative index, and Sp corresponds to a monotonically increasing function curve; p is the safety quantification feature.

[0072] Sub-step 2: Determine the safety status score based on the risk cumulative index;

[0073] The safety status score is calculated according to the following formula:

[0074] v = w1*η + w2*Q + w3*R; and 0 ≤ v ≤ 1; and the larger v is, the worse the safety status of the power battery.

[0075] where Q is the relative risk, Q = z max , z is the slope value of the Sp curve; R is the risk frequency, and the value of R is the number of values in z that are greater than the fourth specified threshold, and the fourth specified threshold is set based on the actual risk assessment requirements.

[0076] η is the absolute risk, η = z a / z b , z a is the mean of the larger part of z, and z b is the mean of the smaller part of z. Specifically, in this embodiment, all the z values of the Sp curve are arranged in descending order of numerical size, and the z values in the first 50% are z a ; the z values in the last 50% are z b .

[0077] w1, w2, and w3 are the weight coefficients of η, Q, and R respectively. In this embodiment, w1, w2, and w3 are the weight coefficients after normalization of η, Q, and R respectively, and w1 + w2 + w3 = 1. With this setting of the scheme, each risk-related reference information has its corresponding weight allocation, which can ensure the meticulous and accurate evaluation of the safety status score.

[0078] Sub-step 3: Confirm the battery risk level according to the safety status score.

[0079] Specifically, the safety status score is proportional to the battery risk level. The larger the safety status score, the higher the battery risk level.

[0080] In specific applications, the battery risk level can be divided into L1, L2, and L3, corresponding to low risk level, medium risk level, and high risk level. Among them, the safety status scores corresponding to each battery risk level are set as follows:

[0081]

[0082] Among them, AlarmLevel is the battery risk level; in practical applications, the specific values of v 1 and v 2 can be set according to actual determination requirements.

[0083] A power battery stability evaluation method based on projection correlation provided by this embodiment provides a brand-new quantitative analysis method and a safety state evaluation method for power battery state evaluation, and can achieve a relatively high accuracy in power battery feature extraction evaluation and a relatively high accuracy in power battery state evaluation.

[0084] Based on current signal data and voltage signal data, this solution specially designs arithmetic expressions related to safety elements. The calculated safety elements characterize the projection part characteristics of voltage in the current direction, can fully reflect the correlation between voltage signals and current signals and represent them quantitatively, can effectively and accurately describe the changes of voltage and current of power batteries during vehicle operation, and can quantitatively describe the evolution trend of sampling abnormal fault characteristics with rich battery state information. Then, combined with the quantitative analysis and state recognition methods in Step 4 and Step 5, it can further perform integrated quantitative analysis on safety elements. The determination of the safety state is obtained based on the confirmation of the safety state score and can be represented by the battery risk level, and the evaluation result is intuitive and detailed.

[0085] Moreover, through the extraction and calculation of safety elements, this solution can extract the pure linear correlation between voltage signals and current signals, can effectively exclude redundant information and noise information brought by factors such as battery systems, working conditions, and environments, and can more accurately extract signal characteristics representing the operating state of power batteries, solving the problem that hidden data (i.e., the above-mentioned redundant information and noise information) cannot be discovered and processed in existing battery state evaluations, resulting in low evaluation accuracy. The extraction of safety elements is more reasonable, the safety quantitative characteristics are more effective, and the determination of the safety state is more accurate. In addition, precisely because this solution can exclude the influence of noise, for subtle fault information that is conventionally masked by noise information, this solution can also capture it through projection correlation, can detect abnormal operation data at the fault evolution stage, and realize early hidden trouble investigation of faults, which helps to eliminate fault risks in advance.

[0086] The above are only embodiments of the present invention. Specific structures and common knowledge such as characteristics that are well-known in the art are not described in detail herein. Those of ordinary skill in the art know all the common general technical knowledge in the technical field to which the invention pertains before the filing date or the priority date, can learn all the prior art in this field, and have the ability to apply the conventional experimental means before this date. Those of ordinary skill in the art can, under the inspiration given in this application, perfect and implement this solution in combination with their own abilities. Some typical well-known structures or well-known methods should not become obstacles for those of ordinary skill in the art to implement this application. It should be noted that for those skilled in the art, without departing from the structure of the present invention, several deformations and improvements can also be made, and these should also be regarded as the protection scope of the present invention, and these will not affect the implementation effect of the present invention and the practicality of the patent. The protection scope claimed in this application shall be subject to the content of its claims, and the specific implementation manners and the like recorded in the specification can be used to interpret the content of the claims.

Claims

1. A method for evaluating the stability of power batteries based on projection correlation, characterized in that, it includes the following steps: Step 1: Collect basic data; the basic data includes voltage signal data and current signal data of the power battery; Step 2: Preprocess the basic data to obtain target data; Step 3: Extract safety elements from the target data; when extracting safety elements, calculate and extract according to the following formula: ； Among them, is the median voltage difference of the battery cells of the power battery; I is the total current of the power battery; s is a safety factor characterizing abnormality, referring to the eigenvector of the projection part of the voltage of the power battery in the current direction; Step 4: Quantify the safety elements and obtain safety quantification features; when quantifying the safety elements, quantify according to the following formula: ； Among them, is the variance entropy, and ; p is the safety quantization feature; Step 5: Identify the safety state based on the safety quantification features; When identifying the safety state, identify according to a preset identification strategy; the preset identification strategy includes the following sub-steps: Sub-step 1: Based on the safety quantification features, perform risk accumulation calculation according to the following formula and obtain a risk accumulation index: ； Among them, is a risk accumulation index, and is a monotonically increasing function curve; Sub-step 2: Determine the safety state score based on the risk accumulation index; The safety state score is calculated according to the following formula: ; and ; Among them, Q is the relative risk, , z is the curve slope value; R is the risk frequency, and the value of R is the number of values in z that are greater than the fourth specified threshold; is the absolute risk, , sort all the z values of the curve in descending order of numerical magnitude. The z values in the top 50% are ; the z values in the bottom 50% are ; , , are respectively the weight coefficients of , Q, and R; Sub-step 3: Confirm the battery risk level according to the safety state score.

2. A method for evaluating the stability of power batteries based on projection correlation according to claim 1, characterized in that, in Step 2, the preprocessing operation includes: removing abnormal data from the basic data; the abnormal data is the data in the basic data that exceeds the first specified threshold.

3. A method for evaluating the stability of power batteries based on projection correlation according to claim 2, characterized in that, in Step 2, the preprocessing operation further includes: identifying and marking special data; when identifying special data, if the difference between the current frame voltage data and the previous frame voltage data exceeds the second specified threshold, it is determined that the current frame voltage data is special data and the current frame voltage data is marked; if the difference between the current frame timestamp data and the previous frame timestamp data exceeds the third specified threshold, it is determined that the current frame timestamp data is special data and the current frame timestamp data is marked.

4. A method for evaluating the stability of power batteries based on projection correlation according to claim 3, characterized in that, in Step 2, the preprocessing operation further includes: performing mean filtering on the basic data.

5. A method for evaluating the stability of power batteries based on projection correlation according to claim 1, characterized in that, The said is determined according to the following steps: S1: Select target battery cells; S2: Calculate the difference between each frame of voltage data of the target battery cell and the median cell voltage, and then obtain the median voltage difference of the target battery cell, which is .

6. A method for evaluating the stability of power batteries based on projection correlation according to claim 1, characterized in that, in Sub-step 3, the safety state score is proportional to the battery risk level, and the larger the safety state score, the higher the battery risk level.

Images