A method and system for detecting metering performance of an off-board charger for electric vehicles

Abstract

The present invention relates to a method and system for detecting the metering performance of an off-board charger for electric vehicles. The method comprises the following steps: S1, collecting historical power data, ambient temperature data and charging time data of the charger, and preprocessing the historical power data; S2, calculating the relationship parameters between the preprocessed historical power data and the ambient temperature data and charging time data, and constructing fusion features; S3, using the fusion features as input data, and conducting model training on multiple different types of models; S4, constructing fusion features for the charger data to be tested and importing them into multiple different types of trained models, and jointly calculating the final detection result through the output of multiple different types of models. The method and system do not require power outage detection, and the detection is accurate, efficient and convenient.

Description

Technical Field

[0001] The present invention belongs to the technical field of electric vehicle charging equipment, and in particular relates to a method and system for detecting metering performance of an off-board charger for an electric vehicle. Background Art

[0002] At present, the energy metering test method is usually used to detect abnormal metering performance of non-on-board chargers for electric vehicles, that is, to use professional testing equipment to perform energy metering tests on the charger to verify whether the power provided by the charger matches the nominal value. However, this method cannot be used to charge the electric vehicle using a charging pile, which affects the efficiency of the charger, consumes a lot of manpower and material resources, and has a high detection cost. Summary of the invention

[0003] The object of the present invention is to provide a method and system for detecting the metering performance of an off-board charger for an electric vehicle. The method and system do not require power outage detection and the detection is accurate, efficient and convenient.

[0004] In order to achieve the above object, the technical solution adopted by the present invention is: a method for detecting the metering performance of an off-board charger for an electric vehicle, comprising the following steps:

[0005] S1. Collect historical power data, ambient temperature data and charging time data of the charger, and pre-process the historical power data;

[0006] S2, calculating the relationship parameters between the pre-processed historical power data, the ambient temperature data and the charging time data, and constructing fusion features;

[0007] S3, using the fusion features as input data, to train multiple different types of models;

[0008] S4. Construct fusion features for the charger data to be tested and import them into multiple trained models of different types, and calculate the final detection result through the outputs of multiple different types of models.

[0009] Furthermore, in step S1, historical power data and historical environmental data of multiple chargers of the same model at the same location at multiple different time points are collected; the historical power data includes:

[0010] 1) Status data, which is discrete data: normal or abnormal;

[0011] 2) Voltage amplitude data, which is the voltage amplitude per second time series data one hour before the acquisition time point, denoted as X U , X U ={x U1 ,x U2 ,...,x U3600};

[0012] 3) Current amplitude data, where the current amplitude data is the time-series data of the current amplitude per second in the hour before the acquisition time point, denoted as X I , X I = {x I1 , x I2 ,..., x I3600};

[0013] 4) Third-harmonic amplitude data, where the third-harmonic amplitude data is the time-series data of the third-harmonic voltage amplitude per second in the hour before the acquisition time point, denoted as X H , X H = {x H1 , x H2 ,..., x H3600};

[0014] The ambient temperature data is the time-series data of the ambient temperature per second in the hour before the acquisition time point, denoted as X T , X T = {x T1 , x T2 ,..., x T3600};

[0015] The charging duration data is the charging duration data from the start of charging to the acquisition time point, denoted as X E .

[0016] Furthermore, in step S1, preprocessing the historical power data specifically includes:

[0017] S101. Two-dimensionalize the data, using time as the abscissa and the voltage amplitude data X U , the current amplitude data X I , and the third-harmonic amplitude data X H as the ordinates to form three two-dimensional scatter plots. Use the DBSCAN clustering algorithm to separately select the cluster with the highest density. The sample points within this cluster are recorded as majority-class sample points, and the sample points not within this cluster are recorded as minority-class sample points;

[0018] S102. For X U , use the k-nearest neighbor algorithm to search for h nearest neighbor samples of the sample point x Ui . The number of majority-class sample points among the h nearest neighbor samples is denoted as h'. If h / 2 ≤ h' < h, the sample point x Ui is regarded as a boundary point. If 0 ≤ h' < h / 2, the sample point x Ui is regarded as a minority-class sample point; Take out all the sample points that meet h / 2 ≤ h' < h to form the boundary point set B U ;

[0019] For the boundary point set B U Perform equalization processing; for the boundary point set B U Each element in is processed as follows:

[0020] b new = b g + λ ( b h - b g )

[0021] in, b new For the processed samples, λ is a random number in the interval (0,1), b g B is the boundary point set U The samples in b h for b g The nearest neighbor sample of b new Form a new boundary point set B Unew , replace the corresponding sample with X U The samples in form new voltage amplitude data, denoted as X UN ;

[0022] S103, use the same method as S102 to I , X H Processing is performed to obtain new current amplitude data X IN , new third harmonic amplitude data X HN .

[0023] Furthermore, in step S2, the relationship parameters between the pre-processed historical power data, the ambient temperature data, and the charging time data are calculated to construct fusion features. The specific method is as follows:

[0024] The obtained X UN , X IN , X HN As input data, the corresponding state data is used as output data, and is respectively imported into three models: recurrent neural network RNN, support vector machine SVM, and random forest RF for model training. The model accuracy of the iteration a is extracted: m1, m2, m3, and the number of iterations p1, p2, p3 at convergence is recorded;

[0025] For X UN , collect X UN The ambient temperature data and charging time data at the corresponding time are recorded as X TNU, X ENU ; Design parameter K to reflect X UN , X IN , X HN One-to-one correspondence with ambient temperature data and charging time data;

[0026] X UN With X TNU The relationship parameter K 11 for:

[0027]

[0028] in, Indicates the permutation and combination symbol, , count ( ) is the counting function, K 11 For X UN With X TNU The relationship parameter, n(x UNU ) is x UNU Similarly, collect X IN , X HN The ambient temperature data and charging time data at the corresponding time are recorded as X TNI , X ENI , X TNH , X ENH , all parameters are calculated using the same method: X UN With X ENU The relationship parameter is K 12 , X IN With X TNI The relationship parameter is K 21 , X IN With X ENI The relationship parameter is K 22 , X HN With X TNH The relationship parameter is K 31 , X HN With X ENH The relationship parameter is K 32 ;

[0029] Using input data X UN , X IN , X HN Construct the fusion feature T as follows:

[0030]

[0031] In the formula, X mean For X UN , X IN , X HN The numerical average value, X max For XUN , X IN , X HN The maximum value of X min For X UN , X IN , X HN The minimum value of .

[0032] Furthermore, in step S3, the fusion feature is used as input data to perform model training on multiple different types of models, and the specific method is as follows:

[0033] The fused feature T is used as input data, and the corresponding state data is used as output data, which are respectively imported into the recurrent neural network RNN, support vector machine SVM, and random forest RF models for model training to obtain three models: M1, M2, and M3.

[0034] Furthermore, in step S4, a fusion feature is constructed for the charger data to be tested and imported into a plurality of trained models of different types, and the final detection result is calculated by combining the outputs of the plurality of different models. The specific method is as follows:

[0035] Construct fusion features for the charger data to be tested, import the M1, M2, and M3 models respectively, and obtain three results, recorded as r1, r2, and r3. The results are normal or abnormal. The normal result is set to 1, and the abnormal result is set to 0. Calculate the judgment parameters R and R0 according to the following formula:

[0036]

[0037]

[0038] Determine whether the charger to be tested is normal according to the following rules:

[0039]

[0040] The above judgment results are output as the final charger measurement performance test results.

[0041] The present invention also provides a metering performance detection system for an off-board charger of an electric vehicle, comprising a memory, a processor, and computer program instructions stored in the memory and capable of being executed by the processor. When the processor executes the computer program instructions, the above-mentioned method steps can be implemented.

[0042] Compared with the prior art, the present invention has the following beneficial effects: the present invention provides a method and system for detecting the metering performance of an off-board charger for electric vehicles without power outage detection. The method and system are data-driven, and train the model by calculating the relationship parameters between power data and environmental factors and constructing new feature parameters, and then fuse the outputs of multiple different types of models to obtain the final detection result, so that the metering performance of the off-board charger for electric vehicles can be accurately and efficiently detected without affecting the use of the charger due to power outages. The detection is convenient, does not require a large amount of manpower and material resources, and has a low detection cost. Therefore, the present invention has strong practicality and broad application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 is a flow chart of a method implementation of an embodiment of the present invention;

[0044] Figure 2 2 is a diagram of the detection results of an embodiment of the present invention. DETAILED DESCRIPTION

[0045] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0046] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present application belongs.

[0047] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0048] like Figure 1 As shown, this embodiment provides a method for detecting the metering performance of an off-board charger for an electric vehicle, comprising the following steps:

[0049] S1. Collect historical power data, ambient temperature data, and charging time data of the charger, and pre-process the historical power data.

[0050] Specifically, historical power data and historical environmental data of multiple chargers of the same model at the same location at multiple different time points are collected.

[0051] The historical power data includes:

[0052] 1) Status data, which is discrete data: normal or abnormal.

[0053] 2) Voltage amplitude data, which is the voltage amplitude per second time series data one hour before the acquisition time point, denoted as X U , X U ={x U1 ,x U2 ,...,x U3600}.

[0054] 3) Current amplitude data, which is the current amplitude per second time series data one hour before the acquisition time point, denoted as X I , X I ={x I1 ,x I2 ,...,x I3600}.

[0055] 4) Third harmonic amplitude data, which is the third harmonic voltage amplitude per second time series data one hour before the acquisition time point, recorded as X H , X H ={x H1 ,x H2 ,...,x H3600}.

[0056] The ambient temperature data is the ambient temperature per second time series data one hour before the collection time point, denoted as X T , X T ={x T1 ,x T2 ,...,x T3600}.

[0057] The charging time data is the charging time data from the start of charging to the collection time point, recorded as X E .

[0058] On this basis, the historical power data is preprocessed, including:

[0059] S101, convert the data into two dimensions, with time as the horizontal axis and voltage amplitude data X as the horizontal axis. U , current amplitude data X I , third harmonic amplitude data X H As the vertical axis, three two-dimensional scatter plots are formed, and the DBSCAN clustering algorithm is used to select the cluster with the largest density. The sample points in the cluster are recorded as the majority class sample points, and the sample points not in the cluster are recorded as the minority class sample points.

[0060] X U For example, set the minimum value Minpts=X U / 10, neighborhood Eps=0.2; (1) in X USelect any unprocessed data point p from it, then calculate the number of data points within the Eps neighborhood of the data point p, and determine whether this quantity meets or exceeds the minimum point threshold. If the number of data points can reach the minimum value of Minpts, create a new cluster C and incorporate all the data points within its neighborhood into the cluster C; (2) Similarly, select an unprocessed data point q, detect its Eps neighborhood. If the number of data points can reach the minimum value of Minpts, add the data points in its neighborhood that do not belong to any cluster to the cluster C; (3) Continue step (2) in the same way until no new objects are added to the current cluster C. (4) Repeat (1) to (3) until all data points are processed. For X I and X H perform the same processing.

[0061] S102. Taking X U as an example, use the k-nearest neighbor algorithm to search for h nearest neighbor samples of the sample point x Ui . Denote the number of sample points belonging to the majority class among the h nearest neighbor samples as h'. If h / 2 ≤ h' < h, the sample point x Ui is regarded as a boundary point. If 0 ≤ h' < h / 2, the sample point x Ui is regarded as a minority class sample point; Take out all the sample points that meet h / 2 ≤ h' < h to form the boundary point set B U .

[0062] Perform an equalization process on the boundary point set B U ; For each element in the boundary point set B U , its processing method is:

[0063] b new = b g + λ ( b h - b g )

[0064] where, b new is the processed sample, λ is a random number in the interval (0, 1), b g is the sample in the boundary point set B U , b h is b g 's nearest neighbor sample; Use the sample b new to form a new boundary point set B Unew , and replace the corresponding sample with XU The samples in form new voltage amplitude data, denoted as X UN .

[0065] S103, use the same method as S102 to I , X H Processing is performed to obtain new current amplitude data X IN , new third harmonic amplitude data X HN .

[0066] S2. Calculate the relationship parameters between the preprocessed historical power data, ambient temperature data, and charging time data, and construct fusion features.

[0067] The obtained X UN , X IN , X HN As input data, the collected and verified corresponding state data is used as output data and imported into three models: recurrent neural network RNN, support vector machine SVM and random forest RF for model training respectively. The model accuracy of the ath iteration is extracted (preferably, a is 10): m1, m2, m3, and the number of iterations p1, p2, p3 at convergence is recorded.

[0068] X UN For example, collect X UN The ambient temperature data and charging time data at the corresponding time are recorded as X TNU , X ENU ; Design parameter K to reflect X UN , X IN , X HN One-to-one correspondence with ambient temperature data and charging time data.

[0069] X UN With X TNU For example, the relationship between UN With X TNU The relationship parameter K 11 for:

[0070]

[0071] in, Represents the permutation and combination symbol, , count ( ) is the counting function, K 11 For X UN With X TNU The relationship parameter, n(x UNU ) is x UNU Similarly, collect X IN , X HNThe ambient temperature data and charging time data at the corresponding time are recorded as X TNI , X ENI , X TNH , X ENH , all parameters are calculated using the same method: X UN With X ENU The relationship parameter is K 12 , X IN With X TNI The relationship parameter is K 21 , X IN With X ENI The relationship parameter is K 22 , X HN With X TNH The relationship parameter is K 31 , X HN With X ENH The relationship parameter is K 32 .

[0072] Using input data X UN , X IN , X HN Construct the fusion feature T as follows:

[0073]

[0074] In the formula, X mean For X UN , X IN , X HN The numerical average value, X max For X UN , X IN , X HN The maximum value of X min For X UN , X IN , X HN The minimum value of .

[0075] S3. Use the fused features as input data to train multiple different types of models.

[0076] Specifically, the fused feature T is used as input data, and the corresponding state data is used as output data, which are respectively imported into three models: recurrent neural network RNN, support vector machine SVM, and random forest RF for model training to obtain three models: M1, M2, and M3.

[0077] S4. Construct fusion features for the charger data to be tested and import them into multiple trained models of different types, and calculate the final detection result through the outputs of multiple different types of models.

[0078] Construct fusion features for the charger data to be tested, import the M1, M2, and M3 models respectively, and obtain three results, recorded as r1, r2, and r3. The results are normal or abnormal. The normal result is set to 1, and the abnormal result is set to 0. Calculate the judgment parameters R and R0 according to the following formula:

[0079]

[0080]

[0081] Determine whether the charger to be tested is normal according to the following rules:

[0082]

[0083] The above judgment results are output as the final charger measurement performance test results.

[0084] This embodiment also provides a metering performance detection system for an off-board charger of an electric vehicle, including a memory, a processor, and computer program instructions stored in the memory and capable of being executed by the processor. When the processor executes the computer program instructions, the above-mentioned method steps can be implemented.

[0085] In this embodiment, an application test is conducted based on the method and system provided above. The corresponding data of 74 charging piles with known metering performance status are imported into the model constructed by this method. The test results are as follows: Figure 2 As shown in the figure, 73 predictions were correct, 1 was wrong, and the accuracy rate was 98.6%, indicating that the detection accuracy of the present invention is high and has a wide range of practical application prospects.

[0086] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0087] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0088] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0089] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0090] The above is only a preferred embodiment of the present invention, and does not limit the present invention in other forms. Any technician familiar with the profession may use the above disclosed technical content to change or modify it into an equivalent embodiment with equivalent changes. However, any simple modification, equivalent change and modification made to the above embodiment according to the technical essence of the present invention without departing from the technical solution of the present invention still belongs to the protection scope of the technical solution of the present invention.

Claims

1. A method for detecting the metering performance of an off-board charger for an electric vehicle, characterized in that: The following steps are involved: S1. Collect historical power data, ambient temperature data and charging time data of the charger, and pre-process the historical power data; S2, calculating the relationship parameters between the pre-processed historical power data, the ambient temperature data and the charging time data, and constructing fusion features; S3, using the fusion features as input data, to train multiple different types of models; S4, constructing fusion features for the charger data to be tested and importing them into multiple trained models of different types, and jointly calculating the final detection results through the outputs of multiple different models; In step S1, historical power data and historical environmental data of multiple chargers of the same model at the same location at multiple different time points are collected; The historical power data includes: 1) Status data, which is discrete data: normal or abnormal; 2) Voltage amplitude data, which is the voltage amplitude per second time series data one hour before the acquisition time point, denoted as X U , X U ={x U1 ,x U2 ,...,x U3600 }; 3) Current amplitude data, which is the current amplitude per second time series data one hour before the acquisition time point, denoted as X I , X I ={x I1 ,x I2 ,...,x I3600 }; 4) Third harmonic amplitude data, which is the third harmonic voltage amplitude per second time series data one hour before the acquisition time point, recorded as X H , X H ={x H1 ,x H2 ,...,x H3600 }; The ambient temperature data is the ambient temperature per second time series data one hour before the collection time point, denoted as X T , X T ={x T1 ,x T2 ,...,x T3600 }; The charging time data is the charging time data from the start of charging to the collection time point, recorded as X E ; In step S1, the historical power data is preprocessed, specifically including: S101, convert the data into two dimensions, with time as the horizontal axis and voltage amplitude data X as the horizontal axis. U , current amplitude data X I , third harmonic amplitude data X H As the vertical coordinate, three two-dimensional scatter plots are formed, and the cluster with the largest density is selected using the DBSCAN clustering algorithm. The sample points in the cluster are recorded as the majority class sample points, and the sample points not in the cluster are recorded as the minority class sample points; S102. For X U , use the k-nearest neighbor algorithm to search for h nearest neighbor samples of the sample point x Ui . Denote the number of sample points belonging to the majority class among the h nearest neighbor samples as h'. If h / 2 ≤ h' < h, the sample point x Ui is regarded as a boundary point. If 0 ≤ h' < h / 2, the sample point x Ui is regarded as a minority class sample point; take out all sample points that meet h / 2 ≤ h' < h to form the boundary point set B U ; For the boundary point set B U Perform equalization processing; for the boundary point set B U Each element in is processed as follows: b new = b g + λ ( b h - b g ) in, b new For the processed samples, λ is a random number in the interval (0,1), b g B is the boundary point set U The samples in b h for b g The nearest neighbor sample of b new Form a new boundary point set B Unew , replace the corresponding sample with X U The samples in form new voltage amplitude data, denoted as X UN ; S103, use the same method as S102 to I , X H Processing is performed to obtain new current amplitude data X IN , new third harmonic amplitude data X HN ; In step S2, the relationship parameters between the pre-processed historical power data, the ambient temperature data, and the charging time data are calculated to construct fusion features. The specific method is as follows: The obtained X UN , X IN , X HN As input data, the corresponding state data is used as output data, and is respectively imported into three models: recurrent neural network RNN, support vector machine SVM, and random forest RF for model training. The model accuracy of the iteration a is extracted: m1, m2, m3, and the number of iterations p1, p2, p3 at convergence is recorded; For X UN , collect X UN The ambient temperature data and charging time data at the corresponding time are recorded as X TNU , X ENU ; Design parameter K to reflect X UN , X IN , X HN One-to-one correspondence with ambient temperature data and charging time data; X UN With X TNU The relationship parameter K 11 for: ; in, Represents the permutation and combination symbol, , count ( ) is the counting function, K 11 For X UN With X TNU The relationship parameter, n(X UN ) is X UN Similarly, collect X IN , X HN The ambient temperature data and charging time data at the corresponding time are recorded as X TNI , X ENI , X TNH , X ENH , all parameters are calculated using the same method: X UN With X ENU The relationship parameter is K 12 , X IN With X TNI The relationship parameter is K 21 , X IN With X ENI The relationship parameter is K 22 , X HN With X TNH The relationship parameter is K 31 , X HN With X ENH The relationship parameter is K 32 ; Using input data X UN , X IN , X HN Construct the fusion feature T as follows: ; Where, X mean For X UN , X IN , X HN The numerical average value, X max For X UN , X IN , X HN The maximum value of X min For X UN , X IN , X HN The minimum value of .

2. A method for detecting the metering performance of an off-board charger for an electric vehicle according to claim 1, characterized in that: In step S3, the fusion features are used as input data to train multiple different types of models. The specific method is as follows: The fused feature T is used as input data, and the corresponding state data is used as output data, which are respectively imported into the recurrent neural network RNN, support vector machine SVM, and random forest RF models for model training to obtain three models: M1, M2, and M3.

3. A method for detecting the metering performance of an off-board charger for an electric vehicle according to claim 2, characterized in that: In step S4, a fusion feature is constructed for the charger data to be tested and imported into multiple trained models of different types, and the final detection result is calculated by combining the outputs of multiple different models. The specific method is as follows: Construct fusion features for the charger data to be tested, import the M1, M2, and M3 models respectively, and obtain three results, recorded as r1, r2, and r3. The results are normal or abnormal. The normal result is set to 1, and the abnormal result is set to 0. Calculate the judgment parameters R and R0 according to the following formula: ; ; Determine whether the charger to be tested is normal according to the following rules: ; The above judgment results are output as the final charger measurement performance test results.

4. A metering performance detection system for an off-board charger of an electric vehicle, characterized in that: The method comprises a memory, a processor, and computer program instructions stored in the memory and executable by the processor. When the processor executes the computer program instructions, the method steps as described in any one of claims 1 to 3 can be implemented.

Images