A distributed charging pile energy storage battery intelligent testing method and system

Abstract

The present invention relates to the field of electric power monitoring technology, and specifically to a distributed charging pile energy storage battery intelligent testing method and system. The method first obtains the performance parameter time series data of different time series processes under the same functional type of each energy storage battery; in the performance parameter time series data of two time series processes, according to the difference between the corresponding performance parameter values ​​of the data points, the degree of outliers of the data points, and the adjusted confidence of the data points, the target distance measurement values ​​of the two time series processes are obtained; in all time series processes of all energy storage batteries, according to the target distance measurement values ​​of every two time series processes, anomaly detection is performed to obtain the performance test results of each energy storage battery. In the present invention, by constructing a distance measurement value reflecting performance anomalies, the error of anomaly detection is reduced and the accuracy of performance test results is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power monitoring, and in particular to an intelligent testing method and system for distributed charging pile energy storage batteries. Background Art

[0002] Charging piles are a type of infrastructure that provides power for electric vehicles. Distributed charging piles refer to multiple charging piles scattered in different areas and places in the city. Distributed charging piles are usually decentralized charging equipment built based on parking lots in various functional areas. Distributed charging piles can provide more flexible and convenient charging services. Energy storage batteries are the core components of distributed charging piles. Their main function is to store electrical energy when the power grid is sufficient so that it can be released when needed. The performance of energy storage batteries directly affects the efficiency of energy utilization. If the performance of energy storage batteries is abnormal, it may lead to energy waste and reduce energy utilization efficiency. Intelligent testing of distributed charging pile energy storage batteries is to monitor the timing process of energy storage batteries in real time and evaluate the status and performance of energy storage batteries, which is crucial to ensure the stable operation of the charging pile system.

[0003] Since there is a large difference between the data in the time series process corresponding to the abnormal performance of the energy storage battery and the data in the time series process corresponding to the normal performance of the energy storage battery, the existing LOF (Local Outlier Factor) anomaly detection method measures the distance between different time series processes through the differences of all data in the time series process, performs anomaly detection, and then obtains the performance test results of the energy storage battery. In the process of measuring the distance between different time series processes, the existing LOF anomaly detection method causes a surge phase due to the start of the time series process and weather factors. The data points corresponding to the surge phase cannot accurately reflect the true performance of the energy storage battery, resulting in the distance measurement value being difficult to reflect the difference caused by the performance anomaly, resulting in a large error in the anomaly detection and inaccurate performance test results. Summary of the invention

[0004] In order to solve the technical problem that in the process of measuring the distance between different time series processes in the prior art LOF anomaly detection method, the start of the time series process and weather factors will cause a surge stage, and the corresponding data points in the surge stage cannot accurately reflect the true performance of the energy storage battery, resulting in the distance measurement value being difficult to reflect the difference caused by the performance anomaly, resulting in a large error in the anomaly detection and inaccurate performance test results, the purpose of the present invention is to provide a distributed charging pile energy storage battery intelligent testing method and system, and the technical scheme adopted is as follows:

[0005] A distributed charging pile energy storage battery intelligent testing method, the method comprising:

[0006] Acquire performance parameter time series data of different time series processes of each energy storage battery under the same functional type; the performance parameter time series data includes performance parameter values ​​at each moment;

[0007] In the performance parameter time series data of the time series process, the initial confidence of the data point is obtained according to the time series position corresponding to the data point and the local fluctuation of the performance parameter value corresponding to the data point; each data point at the same time as the data point is used as each reference time data of the data point; according to the local distribution of the data point and the irrelevance between the local distribution of the reference time data, the initial confidence of the data point is adjusted to obtain the adjusted confidence of the data point;

[0008] In the performance parameter time series data of the two time series processes, according to the difference between the performance parameter values ​​corresponding to the data points, the degree of outliers of the data points, and the adjusted confidence of the data points, the target distance measurement values ​​of the two time series processes are obtained;

[0009] In all the time sequence processes of all the energy storage batteries, anomaly detection is performed according to the target distance measurement value between every two time sequence processes to obtain the performance test result of each energy storage battery.

[0010] Furthermore, the method for obtaining the initial confidence level includes:

[0011] Get the surrounding data points of a data point;

[0012] According to the volatility of the corresponding performance parameter values ​​of all surrounding data points of the data point, the local volatility value of the data point is obtained;

[0013] Obtaining a performance confidence value of a data point according to a time series position of the data point in the performance parameter time series data to which it belongs;

[0014] The initial confidence of the data point is obtained according to the local fluctuation degree value and the performance confidence value; the performance confidence value and the initial confidence are positively correlated; the local fluctuation degree value and the initial confidence are negatively correlated.

[0015] Furthermore, the method for obtaining the performance confidence value includes:

[0016] A performance confidence value is obtained according to a time interval between a data point and an initial data point in the performance parameter time series data to which the data point belongs; the time interval is positively correlated with the performance confidence value.

[0017] Furthermore, the method for obtaining the adjusted confidence level includes:

[0018] Taking the data point and the reference time data as the distribution data point to be analyzed; determining the preset neighborhood area time series data of the distribution data point to be analyzed;

[0019] Obtaining a reference degree value of a data point according to the irrelevance between the preset neighborhood area time series data corresponding to the data point and the preset neighborhood area time series data corresponding to each reference time point data;

[0020] According to the reference degree value of the data point, the initial confidence degree of the data point is adjusted to obtain the adjusted confidence degree of the data point; the reference degree value, the initial confidence degree and the adjusted confidence degree are all positively correlated.

[0021] Furthermore, the method for obtaining the reference level value includes:

[0022] Based on the Pearson correlation formula, according to the correlation between the preset neighborhood area time series data corresponding to the data point and the preset neighborhood area time series data corresponding to the reference time data, the correlation coefficient value between the data point and the reference time data is obtained;

[0023] According to the correlation coefficient value between the data point and all reference time data, the reference degree value of the data point is obtained; the correlation coefficient value and the reference degree value are negatively correlated.

[0024] Furthermore, the method for obtaining the target distance measurement value includes:

[0025] Based on the DTW matching method, data points in the performance parameter time series data of two time series processes are matched to obtain each matching pair; the matching pair includes two matching data points;

[0026] According to the corresponding outlier degree of two matching data points of the matching pair, the matching outlier value of the matching pair is obtained;

[0027] Obtaining a matching confidence value of the matching pair according to the adjusted confidence corresponding to the two matching data points of the matching pair;

[0028] Obtaining a performance difference value of the matching pair according to a difference between corresponding performance parameter values ​​of two matching data points of the matching pair;

[0029] Obtaining a local distance measurement value of the matching pair according to a performance difference value, a matching outlier value, and a matching confidence value of the matching pair; the performance difference value, the matching outlier value, and the matching confidence value are all positively correlated with the local distance measurement value;

[0030] Among all matching pairs corresponding to the two time series processes, target distance measurement values ​​of the two time series processes are obtained according to local distance measurement values ​​of all the matching pairs; the local distance measurement value and the target distance measurement value are positively correlated.

[0031] Furthermore, the method for obtaining the matching outlier includes:

[0032] According to the outlier degree of the matching data point in the corresponding performance parameter time series data, the outlier degree value of the matching data point is obtained;

[0033] According to the outlier degree values ​​corresponding to the two matching data points of the matching pair, the matching outlier value of the matching pair is obtained.

[0034] Furthermore, the method for obtaining the performance test results includes:

[0035] In all the time sequence processes of all the energy storage batteries, anomaly detection is performed according to the target distance measurement value between every two time sequence processes to obtain each abnormal time sequence process;

[0036] According to the abnormal timing process of each energy storage battery, the performance test result of each energy storage battery is obtained.

[0037] Furthermore, the method for acquiring the abnormal timing process includes:

[0038] Perform anomaly detection according to the target distance measurement value of each two time series processes to obtain an anomaly score for each time series process;

[0039] Each time series process whose abnormality score is greater than a preset abnormality threshold is regarded as each abnormal time series process.

[0040] The present invention proposes a distributed charging pile energy storage battery intelligent testing system, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of any one of the distributed charging pile energy storage battery intelligent testing methods are implemented.

[0041] The present invention has the following beneficial effects:

[0042] Considering that at the beginning of the timing process, there will be a short surge phenomenon in which the performance parameter value surges. The data points corresponding to the surge stage are difficult to reflect the true performance of the energy storage battery. The initial confidence of the data point is obtained. The initial confidence can initially reflect the credibility of the data point. The larger the initial confidence, the less likely the data point is to be in the surge stage generated at the beginning of the timing process, and the more credible the data point is. Considering that the weather in the area where the energy storage battery is located will cause the performance parameter value to change, for example, when there is thunder and lightning weather in the area where the energy storage battery is located, it will be affected by the change of the electromagnetic environment around the charging pile at the same time, resulting in a surge phenomenon in the performance parameter value. This data fluctuation is difficult to reflect the true performance value. The initial confidence of the data point is adjusted to obtain the adjusted confidence of the data point. The larger the adjusted confidence, the less likely the data point is to be in abnormal weather and the surge stage generated at the beginning of the timing process, and the more accurate the data point is, the more credible it is.

[0043] Considering that the performance difference between two time series processes is mainly measured based on the difference between all data points, abnormal weather and the surge phase generated at the beginning of the time series process will cause data points to be unreliable, and unreliable data points will make the distance measurement between the two time series processes inaccurate; since the greater the degree of outliers of data points, the more likely it is that the data points are outliers caused by performance anomalies, and the difference between the corresponding performance parameter values ​​of the data points can better reflect the performance difference; obtain the target distance measurement value of the two time series processes. The target distance measurement value can more accurately reflect the performance difference between the two time series processes, improve the reliability of anomaly detection, and make the performance test results more reliable. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are 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 work.

[0045] Figure 1 A flow chart of a distributed charging pile energy storage battery intelligent testing method provided by one embodiment of the present invention;

[0046] Figure 2 A structural diagram of a distributed charging pile energy storage battery intelligent testing system provided by an embodiment of the present invention;

[0047] Figure 3 A schematic diagram of a computer device for intelligent testing of energy storage batteries in a distributed charging pile provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0048] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the distributed charging pile energy storage battery intelligent testing method and system proposed by the present invention, its specific implementation method, structure, features and effects, in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0049] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0050] The specific scheme of a distributed charging pile energy storage battery intelligent testing method and system provided by the present invention is described in detail below with reference to the accompanying drawings.

[0051] See also Figure 1 , which shows a flow chart of a distributed charging pile energy storage battery intelligent testing method provided by an embodiment of the present invention, the method comprising the following steps:

[0052] Step S1: Acquire the performance parameter time series data of different time series processes of each energy storage battery under the same functional type; the performance parameter time series data includes the performance parameter values ​​at each moment.

[0053] Since the main purpose of the present invention is to perform intelligent testing on distributed charging pile energy storage batteries, the test is to evaluate the state and performance of the energy storage battery by real-time monitoring of the timing process of the energy storage battery. In the parking lot area, there are often multiple distributed charging piles, and the distributed charging piles include multiple energy storage batteries. The energy storage batteries include multiple timing processes corresponding to the same functional type. Current is the basic performance parameter value of the energy storage battery. Monitoring it can timely detect performance abnormalities. The discharge operation state is an important functional type of the energy storage battery, which is crucial to monitoring the battery status and performance. In addition, the performance parameter values ​​include voltage, power, etc.; the functional type also includes charging operation status.

[0054] In the embodiment of the present invention, the current is used as the performance parameter value and the discharge operation state is used as the function type. By deploying a current sensor on each energy storage battery, each discharge process is regarded as each time series process under the same function type; in each time series process, the current is sampled according to the preset sampling frequency to obtain the performance parameter value at each moment, and the performance parameter values ​​at all moments are counted in sequence to obtain the performance parameter time series data of each time series process of each energy storage battery. For example, if a certain energy storage battery is in the discharge process at 8:00-9:00 and 11:30-12:00, 8:00-9:00 and 11:30-12:00 are regarded as a time series process respectively.

[0055] The research object of the present invention is all time series processes of the energy storage battery under the same functional type in the parking lot area; that is, the performance parameter time series data of all time series processes of the same functional type of all distributed charging piles corresponding to all energy storage batteries in the same parking lot area are counted as the performance parameter time series data of all time series processes of all energy storage batteries, so as to facilitate the subsequent analysis of the influence of weather factors in the area where the energy storage battery is located on the performance parameter values.

[0056] In one embodiment of the present invention, the preset sampling frequency is 10ms / time, and the implementer can set it according to the implementation scenario. It should be noted that in other embodiments of the present invention, voltage can also be selected as the performance parameter value, and the charging operation state can be selected as the function type, and each charging process can be regarded as each timing process under the same function type, which will not be elaborated or limited here. It should be noted that in order to facilitate calculations, all indicator data involved in the calculations in the embodiments of the present invention are subjected to data preprocessing to eliminate the dimension effect. The specific means of removing the dimension effect are technical means well known to those skilled in the art and are not limited here.

[0057] Step S2: In the performance parameter timing data of the timing process, the initial confidence of the data point is obtained according to the timing position corresponding to the data point and the local fluctuation of the performance parameter value corresponding to the data point; each data point at the same time as the data point is used as the reference time data of the data point; according to the local distribution of the data point and the irrelevance between the local distribution of the reference time data, the initial confidence of the data point is adjusted to obtain the adjusted confidence of the data point.

[0058] Considering that at the beginning of the timing process, there will be a short surge phenomenon in which the performance parameter value surges. The data points corresponding to the surge stage are difficult to reflect the true performance of the energy storage battery. The initial confidence of the data point is obtained. The initial confidence can initially reflect the credibility of the data point. The larger the initial confidence, the less likely the data point is to be in the surge stage generated at the beginning of the timing process, and the more credible the data point is. Considering that the weather in the area where the energy storage battery is located will cause the performance parameter value to change, for example, when there is thunder and lightning weather in the area where the energy storage battery is located, it will be affected by the change of the electromagnetic environment around the charging pile at the same time, resulting in a surge phenomenon in the performance parameter value. This data fluctuation is difficult to reflect the true performance value. The initial confidence of the data point is adjusted to obtain the adjusted confidence of the data point. The larger the adjusted confidence, the less likely the data point is to be in abnormal weather and the surge stage generated at the beginning of the timing process, and the more accurate the data point is, the more credible it is.

[0059] Preferably, in one embodiment of the present invention, considering that the performance difference between two timing processes is measured based on the difference between all data points, the data points corresponding to the surge phase generated at the beginning of the timing process are difficult to reflect the true performance of the energy storage battery. Such data points are unreliable, and it is necessary to construct an initial confidence level to preliminarily reflect the credibility of the data points. The method for obtaining the initial confidence level includes:

[0060] In order to subsequently analyze the local fluctuation degree of the data point, it is necessary to determine the data in the local area where the data point is located, and determine the surrounding data points of the data point in the performance parameter time series data of the time series process.

[0061] In one embodiment of the present invention, a method for acquiring surrounding data points of a data point includes:

[0062] The data point with the preset surrounding data amount closest to the data point in time sequence is used as the surrounding data point of the data point; in one embodiment of the present invention, the preset surrounding data amount is 41, which can be set by the implementer according to the implementation scenario and is not limited here. It should be noted that the surrounding data points of the data point include the data point.

[0063] In another embodiment of the present invention, a method for obtaining surrounding data points of a data point includes: a preset surrounding range is an area constructed with the data point as the center and with a preset surrounding specification as the size; the preset surrounding specification is the minimum circumscribed rectangle that includes a preset amount of surrounding data, and each data point in the preset surrounding range of the data point is used as a surrounding data point of the data point.

[0064] According to the volatility of the corresponding performance parameter values ​​of all surrounding data points of the data point, the local fluctuation degree value of the data point is obtained; the local fluctuation degree value reflects the surrounding instability of the data point. The larger the local fluctuation degree value, the more likely the data point is in the surge stage generated at the beginning of the timing process, and the smaller the initial confidence.

[0065] In one embodiment of the present invention, the variance of the performance parameter values ​​corresponding to all surrounding data points is calculated as the local fluctuation value of the data point, which is not limited here. In other embodiments of the present invention, the standard deviation of the performance parameter values ​​corresponding to all surrounding data points is calculated as the local fluctuation value of the data point, which is not limited here. In one embodiment of the present invention, the data point The local fluctuation value.

[0066] The performance confidence value of the data point is obtained according to the timing position of the data point in the performance parameter timing data to which the data point belongs; the smaller the performance confidence value is, the more likely the data point is in the surge stage generated at the beginning of the timing process, and the smaller the initial confidence is.

[0067] In one embodiment of the present invention, a method for obtaining a performance confidence value of a data point includes:

[0068] The performance confidence value is obtained according to the time interval between the data point and the initial data point in the performance parameter time series data to which it belongs; the time interval is positively correlated with the performance confidence value.

[0069] Obtain the time corresponding to the initial data point in the performance parameter time series data to which the data point belongs; calculate the difference between the time corresponding to the data point and the time corresponding to the initial data point to obtain the lag value of the data point; obtain the performance confidence value of the data point based on the lag value of the data point; the lag value and the performance confidence value are positively correlated.

[0070] According to the local fluctuation degree value and the performance confidence value, the initial confidence of the data point is obtained; the performance confidence value and the initial confidence are positively correlated; the local fluctuation degree value and the initial confidence are negatively correlated.

[0071] Among them, a positive correlation means that the dependent variable will increase as the independent variable increases, and the dependent variable will decrease as the independent variable decreases. The specific relationship can be a multiplication relationship, an addition relationship, or the power of an exponential function, which is determined by actual application; a negative correlation means that the dependent variable will decrease as the independent variable increases, and the dependent variable will increase as the independent variable decreases. It can be a subtraction relationship, a division relationship, etc., which is determined by actual application.

[0072] The initial confidence is obtained according to the initial confidence formula. The initial confidence formula may be specifically, for example, as follows:

[0073] ;in, For data points The initial confidence level of For the In the performance parameter timing data of the timing process, The time corresponding to the data point; For the The time corresponding to the initial data point in the performance parameter time series data of a time series process; is the surge hyperparameter; For data points The standard deviation of the performance parameter values ​​corresponding to all surrounding data points in the preset surrounding range; is the denominator adjustment factor; is an exponential function with e as base; For data points In one embodiment of the present invention, the denominator adjustment factor is 0.001, and the denominator adjustment factor is used to prevent the denominator from being 0. The implementer can set it according to the implementation scenario. In one embodiment of the present invention, the surge hyperparameter is 0.001, and the implementer can set it according to the length of the surge phase generated at the beginning of the function operation. The longer the duration, the smaller the surge hyperparameter.

[0074] In the formula, considering that there will be a short surge at the beginning of the timing process, the data points corresponding to the surge stage are difficult to reflect the true performance of the energy storage battery. right Proportional normalization is performed so that when the data point is in the surge phase generated at the beginning of the timing process, the data point Low; Considering the surge phase generated at the beginning of the timing process, there will be a short surge phenomenon of performance parameter value surge, and the corresponding performance parameter values ​​around the data point are unstable. The local fluctuation degree of the data point is reflected. The greater the local fluctuation degree, the more likely it is that the data point is in the surge stage generated at the beginning of the timing process. The initial confidence can preliminarily reflect the credibility of the data point. The larger the initial confidence, the less likely the data point is to be in the surge stage generated at the beginning of the timing process, and the more credible the data point is.

[0075] In order to more accurately express the credibility of the data points, considering that the area where the energy storage battery is located will produce a surge stage when the weather is abnormal, in the surge stage caused by abnormal weather, the data points corresponding to the same time will show similar changes at the same time due to the change of the electromagnetic environment around the charging pile. At this time, the data points are difficult to reflect the real performance. Therefore, it is necessary to analyze multiple data points corresponding to the same time to study whether the data points are in the surge stage caused by abnormal weather.

[0076] In one embodiment of the present invention, a method for obtaining reference time data of a data point includes:

[0077] Each data point at the same time as the data point is used as each reference time data of the data point. Each reference time data of the data point represents each data at the same time as the data point, so as to facilitate subsequent analysis of whether it is affected by abnormal weather at the same time.

[0078] Preferably, in one embodiment of the present invention, the data points in the surge stage caused by abnormal weather are unreliable, and such data points are difficult to reflect the real performance value. The initial data points are adjusted to obtain adjusted confidences that can more accurately reflect the credibility of the data points. The method for obtaining the adjusted confidences includes:

[0079] The data points and the reference time data are used as the distribution data points to be analyzed; and the preset neighborhood area time series data of the distribution data points to be analyzed are determined;

[0080] In order to study the local distribution change of the distribution data point to be analyzed and obtain the surrounding data of the distribution data point to be analyzed, in one embodiment of the present invention, the method for obtaining the time series data of the preset neighborhood area of ​​the distribution data point to be analyzed includes:

[0081] The preset neighborhood area is an area constructed with the distribution data point to be analyzed as the center and the preset neighborhood specification as the size; the preset neighborhood specification is the minimum circumscribed rectangle including the preset neighborhood data volume. In one embodiment of the present invention, the preset neighborhood data volume is 31, which can be set by the implementer according to the implementation scenario and is not limited here. All data points in the preset neighborhood area are used as the preset neighborhood area time series data of the distribution data point to be analyzed.

[0082] The reference degree value of the data point is obtained according to the irrelevance between the preset neighborhood area time series data corresponding to the data point and the preset neighborhood area time series data corresponding to each reference moment data; the larger the reference degree value, the less likely the data point is to be in the power surge stage caused by abnormal weather, the more credible the data point is, and the greater the adjusted confidence level is.

[0083] In one embodiment of the present invention, a method for obtaining a reference level value is as follows:

[0084] Based on the Pearson correlation formula, the Pearson coefficient of the preset neighborhood area time series data corresponding to the data point and the preset neighborhood area time series data corresponding to the reference time data is calculated, and the Pearson coefficient is proportionally normalized to obtain the correlation coefficient value between the data point and the reference time data; in the adjusted confidence formula In one embodiment of the present invention, the data point and reference time data The correlation coefficient value between them. It should be noted that the Pearson coefficient is a prior art well known to those skilled in the art and will not be described in detail here. It should be noted that in the embodiment of the present invention, Norm is used as a normalization function to normalize the Pearson coefficient to between 0 and 1.

[0085] Calculate the mean of the correlation coefficient between the data point and all reference time data to obtain the non-reference value of the data point; in the adjusted confidence formula In one embodiment of the present invention, the data point The non-reference value.

[0086] According to the non-reference value of the data point, the reference degree value of the data point is obtained; the non-reference value and the reference degree value are negatively correlated.

[0087] According to the reference degree value of the data point, the initial confidence degree of the data point is adjusted to obtain the adjusted confidence degree of the data point; the reference degree value, the initial confidence degree and the adjusted confidence degree are all positively correlated.

[0088] Among them, a positive correlation means that the dependent variable will increase as the independent variable increases, and the dependent variable will decrease as the independent variable decreases. The specific relationship can be a multiplication relationship, an addition relationship, or the power of an exponential function, which is determined by actual application; a negative correlation means that the dependent variable will decrease as the independent variable increases, and the dependent variable will increase as the independent variable decreases. It can be a subtraction relationship, a division relationship, etc., which is determined by actual application.

[0089] The adjusted confidence is obtained according to the adjusted confidence formula, and the adjusted confidence formula includes:

[0090] ;in, For data points The adjusted confidence level; For data points The total number of all reference time data; For data points The preset neighborhood area time series data; For data points No. Reference time data; Reference time data The preset neighborhood area time series data; For data points The initial confidence level of For the In the performance parameter timing data of the timing process, data points; Time series data for preset neighborhood areas and preset neighborhood area time series data The Pearson coefficient. For data points The reference level value of is the normalization function.

[0091] In the formula, it is considered that in the same parking lot area, the weather has similar effects on each energy storage battery at the same time. For example, when thunder strikes, it will have similar effects on the fluctuation of the performance parameter values ​​of each energy storage battery, that is, the correlation coefficient of the change of the performance parameter value during the surge stage caused by abnormal weather is higher; The larger it is, the more dissimilar the changes of all data at the moment of the data point are, and the less likely the data point is in the power surge stage caused by abnormal weather. right Adjust to obtain the adjusted confidence of the data point. The larger the adjusted confidence, the less likely the data point is to be in the surge phase caused by abnormal weather and the beginning of the time series process, and the more accurate the data point is, the more credible it is.

[0092] Step S3: In the performance parameter time series data of the two time series processes, the target distance measurement values ​​of the two time series processes are obtained according to the difference between the performance parameter values ​​corresponding to the data points, the degree of outliers of the data points, and the adjusted confidence of the data points.

[0093] Considering that the performance difference between two time series processes is mainly measured based on the difference between all data points, abnormal weather and the surge phase generated at the beginning of the time series process will cause data points to be unreliable, and unreliable data points will make the distance measurement between the two time series processes inaccurate; since the greater the degree of outliers of data points, the more likely it is that the data points are outliers caused by performance anomalies, and the difference between the corresponding performance parameter values ​​of the data points can better reflect the performance difference; obtain the target distance measurement value of the two time series processes. The target distance measurement value can more accurately reflect the performance difference between the two time series processes.

[0094] Preferably, in one embodiment of the present invention, a method for obtaining a target distance measurement value includes:

[0095] Considering that when measuring the distance between the performance parameter time series data of two time series processes, it is necessary to consider the distance measurement between the performance parameter time series data of different durations, the DTW (Dynamic Time Warping) matching method can match data points to facilitate the measurement of the distance between the performance parameter time series data of different durations.

[0096] Using the DTW matching method, the data points in the performance parameter time series data of the two time series processes are matched to obtain each matching pair; the two data points corresponding to each matching pair are used as matching data points. It should be noted that the technical means are well known to those skilled in the art, and only the method for obtaining the matching pair is briefly described here:

[0097] Assume that there are two time series data of performance parameters of the time series process, the length of performance parameter time series data A is N, and the length of performance parameter time series data B is M. Initialize an N The cumulative distance matrix D of M, where D[i][j] represents the minimum cumulative distance between the first i points in A and the first j points in B. Calculate the Euclidean distances between all possible point pairs in the two performance parameter time series data, and fill these distances into the corresponding positions as the initial values ​​of the matrix D. Calculate the cumulative distances, and find the optimal regularized path of the performance parameter time series data of the two time series processes by constructing the cumulative distance matrix of the performance parameter time series data of the two time series processes. Each position (i, j) passed by the optimal regularized path represents a matching pair, that is, the i-th point in the performance parameter time series data A and the j-th point in the performance parameter time series data B are matched together. A series of matching pairs are obtained, which describe how to align two performance parameter time series data of different lengths with the minimum cumulative distance. The specific form of the matching pair is a series of coordinates (i, j), and the two data points corresponding to each matching pair are used as matching data points.

[0098] According to the corresponding outlier degree of two matching data points of the matching pair, the matching outlier value of the matching pair is obtained;

[0099] In one embodiment of the present invention, considering that the greater the degree of outlier of a data point is, the more likely it is that the data point outlier is caused by performance abnormality, special attention should be paid when measuring the distance metric value of the matching pair. In order to analyze the degree of outlier of the matching pair, the method for obtaining the matching outlier value includes:

[0100] According to the degree of outlier of the matching data point in the corresponding performance parameter time series data, the outlier degree value of the matching data point is obtained; the outlier degree value can reflect the degree to which the performance parameter value of the matching data deviates from the overall performance parameter value. The larger the value, the greater the degree of deviation.

[0101] In order to obtain the overall performance parameter value corresponding to the performance parameter time series data of the matching data point, in one embodiment of the present invention, the mean of the performance parameter values ​​in the performance parameter time series data of the matching data point is calculated to obtain the overall distribution characteristics of the performance parameter time series data of the matching data point; in other embodiments of the present invention, the median of the performance parameter values ​​in the performance parameter time series data of the matching data point can also be used as the overall distribution characteristics of the performance parameter time series data of the matching data point.

[0102] The absolute value of the difference between the performance parameter value of the matching data point and the overall distribution characteristics is calculated to obtain the outlier degree value of the matching data point. The larger the outlier degree value, the more likely the matching data point is an outlier caused by abnormal performance.

[0103] In order to obtain the outlier degree of the matching pair, the matching outlier value of the matching pair is obtained according to the outlier degree values ​​corresponding to the two matching data points of the matching pair; in one embodiment of the present invention, the mean of the outlier degree values ​​corresponding to the two matching data points of the matching pair is calculated as the matching outlier value of the matching pair; in other embodiments of the present invention, the cumulative value of the outlier degree values ​​corresponding to the two matching data points of the matching pair can also be calculated, and the cumulative value is proportionally normalized to obtain the matching outlier value of the matching pair; it should be noted that in the embodiment of the present invention, Norm is used as the normalization function to normalize the cumulative value to between 0 and 1.

[0104] In order to obtain the credibility of the matching pair, the matching confidence value of the matching pair is obtained according to the corresponding adjusted confidence of the two matching data points of the matching pair; the larger the matching confidence value, the more credible the matching pair.

[0105] In one embodiment of the present invention, the average of the adjusted confidences corresponding to the two matching data points of the matching pair is calculated as the matching confidence value of the matching pair; in other embodiments of the present invention, the cumulative value of the adjusted confidences corresponding to the two matching data points of the matching pair is also calculated as the matching confidence value of the matching pair.

[0106] According to the difference between the corresponding performance parameter values ​​of the two matching data points of the matching pair, the performance difference value of the matching pair is obtained; the larger the performance difference value is, the greater the performance difference of the matching data points is.

[0107] In one embodiment of the present invention, the absolute value of the difference between the corresponding performance parameter values ​​of two matching data points of a matching pair is calculated to obtain the performance difference value of the matching pair; in other embodiments of the present invention, the square of the difference between the corresponding performance parameter values ​​of two matching data points of a matching pair is calculated to obtain the performance difference value of the matching pair.

[0108] According to the performance difference value, matching outlier value and matching confidence value of the matching pair, the local distance measurement value of the matching pair is obtained; the performance difference value, matching outlier value and matching confidence value are all positively correlated with the local distance measurement value; in the target distance measurement formula In one embodiment of the present invention, The local distance measure of the matching pairs.

[0109] Among all matching pairs corresponding to the two time series processes, target distance measurement values ​​of the two time series processes are obtained according to local distance measurement values ​​of all matching pairs; the local distance measurement value and the target distance measurement value are positively correlated.

[0110] In one embodiment of the present invention, the formula for obtaining the target distance measurement value includes:

[0111] ;in, For the and The target distance metric of a temporal process; For the and The total number of matching pairs corresponding to each sequential process; For the and The sequential process corresponds to all matching pairs. The match confidence value of the matching pairs; For the and The sequential process corresponds to all matching pairs. Matching outliers for matching pairs; For the and The sequential process corresponds to all matching pairs. The performance difference between the matched pairs.

[0112] In the formula, through right Make adjustments, The larger it is, the less likely the matching pair is to be in the abnormal weather and surge stage generated by the timing process. The more credible the matching pair is, and the performance difference value corresponding to the matching pair can better reflect the real performance difference. The larger the value, the more likely it is that the data is outliers due to performance anomalies, and more attention should be paid to the performance difference values ​​of matching pairs; Reflects the performance difference value of the matching pair; ultimately, the target distance metric can more accurately reflect the performance difference between the two timing processes.

[0113] Step S4: In all the time series processes of the energy storage battery, anomaly detection is performed based on the target distance measurement value of every two time series processes to obtain the performance test results of each energy storage battery.

[0114] Through the above steps, a target distance measurement value that can accurately reflect the performance difference between the two time series processes is obtained, thereby improving the reliability of anomaly detection and making the performance test results more credible.

[0115] Preferably, in one embodiment of the present invention, the method for obtaining performance test results includes:

[0116] Based on the LOF anomaly detection method, in all the time series processes of all the energy storage batteries, the anomaly score of each time series process is obtained according to the target distance measurement value of each two time series processes; each time series process with an anomaly score greater than the preset anomaly threshold is regarded as each abnormal time series process; it should be noted that the LOF anomaly detection method is a technical means well known to those skilled in the art and will not be described in detail here. In one embodiment of the present invention, the preset anomaly threshold is 1.1, and the implementer can set it according to the implementation scenario.

[0117] According to the abnormal timing process of each energy storage battery, the performance test result of each energy storage battery is obtained.

[0118] In order to analyze the performance of the energy storage battery, considering that the timing process of the energy storage battery gradually increases with time, the shorter the interval between abnormal timing processes, the more likely the performance of the energy storage battery is to be abnormal.

[0119] The interval time between the abnormal time series process and the previous abnormal time series process is used as the abnormal interval value of the abnormal time series process;

[0120] When the abnormal interval value of any abnormal timing process of the energy storage battery is less than the preset abnormal interval value, the performance test result of the energy storage battery is marked as performance abnormality; otherwise, the performance test result of the energy storage battery is marked as normal performance. In one embodiment of the present invention, the preset abnormal interval value is 15, and the implementer can set it according to the implementation scenario. Specifically, when the performance test result of the energy storage battery is marked as performance abnormality, an alarm is sent to the system, and a test engineer is dispatched to test and repair the energy storage battery to ensure the efficient use of energy of the distributed charging pile energy storage battery and the safe and stable operation of the charging pile system.

[0121] In summary, the embodiment of the present invention provides a distributed charging pile energy storage battery intelligent testing method and system, firstly, the performance parameter time series data of different time series processes under the same functional type of each energy storage battery are obtained; in the performance parameter time series data of two time series processes, the target distance measurement value of the two time series processes is obtained according to the difference between the corresponding performance parameter values ​​of the data points, the degree of outliers of the data points, and the adjusted confidence of the data points; in all time series processes of all energy storage batteries, according to the target distance measurement value of every two time series processes, anomaly detection is performed to obtain the performance test results of each energy storage battery. In the embodiment of the present invention, by constructing a distance measurement value reflecting performance anomalies, the error of anomaly detection is reduced and the accuracy of performance test results is improved.

[0122] This embodiment also provides a distributed charging pile energy storage battery intelligent testing system, the system includes a processor, a memory and a computer program, wherein the memory is used to store the corresponding computer program, the processor is used to run the corresponding computer program, and when the computer program runs on the processor, it can implement any one of the steps of a distributed charging pile energy storage battery intelligent testing method.

[0123] See also Figure 2 , which shows a structure diagram of a distributed charging pile energy storage battery intelligent testing system provided by an embodiment of the present invention, including: a data acquisition module 101: used to implement step S1 in the above method, an adjusted confidence analysis module 102: used to implement step S2 in the above method, a target distance metric analysis module 103: used to implement step S3 in the above method, and a performance testing module 104: used to implement step S4 in the above method.

[0124] The present invention also proposes a computer device for intelligent testing of distributed charging pile energy storage batteries, please refer to Figure 3, which shows a schematic diagram of a computer device for intelligent testing of distributed charging pile energy storage batteries provided by an embodiment of the present invention. The computer device includes a memory 401, a processor 402, and a computer program 403 stored in the memory 401 and running on the processor 402, wherein when the processor 402 executes the computer program 403, the computer device can execute any of the above-mentioned intelligent testing methods for distributed charging pile energy storage batteries.

[0125] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0126] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

Claims

1. A distributed charging pile energy storage battery intelligent testing method, characterized in that: The method comprises: Acquire performance parameter time series data of different time series processes of each energy storage battery under the same functional type; the performance parameter time series data includes performance parameter values ​​at each moment; In the performance parameter time series data of the time series process, the initial confidence of the data point is obtained according to the time series position corresponding to the data point and the local fluctuation of the performance parameter value corresponding to the data point; each data point at the same time as the data point is used as each reference time data of the data point; according to the local distribution of the data point and the irrelevance between the local distribution of the reference time data, the initial confidence of the data point is adjusted to obtain the adjusted confidence of the data point; the adjusted confidence is used to reflect the degree to which the data point is unlikely to be in abnormal weather and the surge stage generated at the beginning of the time series process; In the performance parameter time series data of two time series processes, the target distance measurement value of the two time series processes is obtained according to the difference between the performance parameter values ​​corresponding to the data points, the degree of outliers of the data points, and the adjusted confidence of the data points; the target distance is used to accurately reflect the performance difference between the two time series processes; In all the time series processes of all the energy storage batteries, anomaly detection is performed according to the target distance measurement value of every two time series processes to obtain the performance test result of each energy storage battery; The method for obtaining the initial confidence level includes: Get the surrounding data points of a data point; According to the volatility of the corresponding performance parameter values ​​of all surrounding data points of the data point, the local volatility value of the data point is obtained; Obtaining a performance confidence value of a data point according to a time series position of the data point in the performance parameter time series data to which it belongs; According to the local fluctuation degree value and the performance confidence value, the initial confidence of the data point is obtained; the performance confidence value and the initial confidence are positively correlated; the local fluctuation degree value and the initial confidence are negatively correlated; The method for obtaining the performance confidence value includes: Obtaining a performance confidence value according to a time interval between a data point and an initial data point in the performance parameter time series data to which the data point belongs; the time interval is positively correlated with the performance confidence value; The method for obtaining the adjusted confidence level includes: Taking the data point and the reference time data as the distribution data point to be analyzed; determining the preset neighborhood area time series data of the distribution data point to be analyzed; Obtaining a reference degree value of a data point according to the irrelevance between the preset neighborhood area time series data corresponding to the data point and the preset neighborhood area time series data corresponding to each reference time point data; According to the reference degree value of the data point, the initial confidence degree of the data point is adjusted to obtain the adjusted confidence degree of the data point; the reference degree value, the initial confidence degree and the adjusted confidence degree are all positively correlated; The method for obtaining the reference level value includes: Based on the Pearson correlation formula, according to the correlation between the preset neighborhood area time series data corresponding to the data point and the preset neighborhood area time series data corresponding to the reference time data, the correlation coefficient value between the data point and the reference time data is obtained; Obtaining a reference degree value of the data point according to a correlation coefficient value between the data point and all reference time data; the correlation coefficient value and the reference degree value are negatively correlated; The method for obtaining the target distance measurement value includes: Based on the DTW matching method, data points in the performance parameter time series data of two time series processes are matched to obtain each matching pair; the matching pair includes two matching data points; According to the corresponding outlier degree of two matching data points of the matching pair, the matching outlier value of the matching pair is obtained; Obtaining a matching confidence value of the matching pair according to the adjusted confidence corresponding to the two matching data points of the matching pair; Obtaining a performance difference value of the matching pair according to a difference between corresponding performance parameter values ​​of two matching data points of the matching pair; Obtaining a local distance measurement value of the matching pair according to a performance difference value, a matching outlier value, and a matching confidence value of the matching pair; the performance difference value, the matching outlier value, and the matching confidence value are all positively correlated with the local distance measurement value; Among all matching pairs corresponding to the two time series processes, a target distance measurement value of the two time series processes is obtained according to the local distance measurement values ​​of all the matching pairs; the local distance measurement value and the target distance measurement value are positively correlated; The method for obtaining the matching confidence value includes: calculating the average of the adjusted confidences corresponding to the two matching data points of the matching pair as the matching confidence value of the matching pair; The method for obtaining the performance difference value comprises: calculating the absolute value of the difference between the performance parameter values ​​corresponding to two matching data points of the matching pair to obtain the performance difference value of the matching pair; The target distance measurement value acquisition formula includes: ;in, For the and The target distance metric of a temporal process; For the and The total number of matching pairs corresponding to each sequential process; For the and The sequential process corresponds to all matching pairs. The match confidence value of the matching pairs; For the and The sequential process corresponds to all matching pairs. Matching outliers for matching pairs; For the and The sequential process corresponds to all matching pairs. The performance difference value of the matched pairs; For the and The sequential process corresponds to all matching pairs. The local distance measure of the matching pairs.

2. A distributed charging pile energy storage battery intelligent testing method according to claim 1, characterized in that: The method for obtaining the matching outlier value comprises: According to the outlier degree of the matching data point in the corresponding performance parameter time series data, the outlier degree value of the matching data point is obtained; According to the outlier degree values ​​corresponding to the two matching data points of the matching pair, the matching outlier value of the matching pair is obtained.

3. A distributed charging pile energy storage battery intelligent testing method according to claim 1, characterized in that: The method for obtaining the performance test results includes: In all the time sequence processes of all the energy storage batteries, anomaly detection is performed according to the target distance measurement value between every two time sequence processes to obtain each abnormal time sequence process; According to the abnormal timing process of each energy storage battery, the performance test result of each energy storage battery is obtained.

4. A distributed charging pile energy storage battery intelligent testing method according to claim 3, characterized in that: The method for acquiring the abnormal timing process includes: Perform anomaly detection according to the target distance measurement value of each two time series processes to obtain an anomaly score for each time series process; Each time series process whose abnormality score is greater than a preset abnormality threshold is regarded as each abnormal time series process.

5. A distributed charging pile energy storage battery intelligent testing system, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 4 are implemented.

Images