Embedded dynamic monitoring module of electric vehicle charging pile and multi-source data processing method

Abstract

The invention provides an electric vehicle charging pile embedded dynamic monitoring module and a multi-source data processing method, and relates to the technical field of charging pile monitoring. The monitoring module is used for collecting various load information of a single charging facility and can exchange data with collection terminal equipment; the device is composed of a data processing unit, a communication unit, a safety unit, a storage unit and an electric energy measuring unit. The data processing module is used for calculating and analyzing the collected charging facility state data; various communication modes of HPLC, 4G, RS485, CAN and Bluetooth are supported, and data interaction with a charging facility and a cloud server is realized; an encryption algorithm and a signature verification function are built in, so that the security of data transmission and storage is ensured; the data storage module is used for storing collected original data and processed result data; the data acquisition module has a B-level measurement accuracy level and is used for acquiring electric parameter data such as voltage, current and power of a charging facility; module operation states are displayed through the indicating lamps.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of charging pile monitoring, and more particularly relates to an embedded dynamic monitoring module for electric vehicle charging piles and a multi-source data processing method. BACKGROUND

[0002] As an important part of the power grid energy-saving and emission-reducing industry, electric vehicles are an indispensable part of building a new environment-friendly city intelligent transportation. With the gradual deepening of urban energy reform and the gradual development of electric vehicles, the corresponding charging piles also show a scale development. The rapid expansion of the scale of charging infrastructure poses new challenges to digital supervision technologies such as charging facility metering evaluation, remote verification, and electric-carbon conversion.

[0003] The research on the online evaluation technology of charging facility metering performance lays the metering foundation for the digital supervision of charging facilities. Public charging facilities are widely distributed, numerous, and of varying quality. The existing online metering method cannot fully meet the demand for the accuracy of charging facility metering verification. Therefore, it is urgent to explore new technologies to realize efficient supervision and control of charging facility metering performance.

[0004] Based on the above problems, by developing an embedded monitoring module for charging facilities, online acquisition of operation big data is realized, and based on the collected data, multi-source data is identified and repaired, which helps to improve data quality and provide good basic data for remote online monitoring of charging facilities. A DC charging pile AC-DC conversion efficiency-charging state quantitative model is proposed to improve the estimation accuracy of DC charging facility conversion efficiency and provide data support for remote online monitoring model. Based on the operation and transaction data of charging facilities, a charging facility operation state monitoring model is constructed, and the uncertainty of the model is evaluated, which helps to improve the detection accuracy and efficiency of charging facilities. Finally, the monitoring model is evaluated and verified, which helps to significantly improve the digital supervision service level of charging facilities. SUMMARY

[0005] To solve the above technical problems, the application provides an embedded dynamic monitoring module for electric vehicle charging piles and a multi-source data processing method to solve the above problems.

[0006] An embedded dynamic monitoring module for electric vehicle charging piles and a multi-source data processing method, the monitoring module collects charging facility state data in real time, and provides multi-source data support for remote online monitoring of charging piles;

[0007] The monitoring module is used for collecting various load information of a single charging facility and can exchange data with a collection terminal device, and is composed of a data processing unit, a communication unit, a security unit, a storage unit, and an electric energy measurement unit;

[0008] A data processing unit is configured to perform operation and analysis on the collected charging facility status data.

[0009] A communication unit supports multiple communication modes such as HPLC, 4G, RS485, CAN and Bluetooth, and realizes data interaction with the charging facility and the cloud server.

[0010] A security unit is configured to have encryption algorithm and signature verification function to ensure the security of data transmission and storage.

[0011] A storage unit is configured to store the collected raw data and the processed result data.

[0012] An electric energy measurement unit is configured to have B-level measurement accuracy level and is used to collect electric parameter data such as voltage, current and power of the charging facility.

[0013] A status indication unit is configured to display the running state through an indicator light display module.

[0014] Preferably, the dynamic monitoring module has the functions of protocol conversion, power quality analysis, single-point flexible control, external circuit breaker control, information storage and processing, and software upgrade, and has HPLC communication, RS-485 communication, CAN communication and Bluetooth communication, and finally completes the development of the monitoring module.

[0015] Preferably, the communication unit includes a local communication module and a remote communication module, the local communication module uses RS485 or CAN protocol to realize near-distance data interaction with the charging pile, and the remote communication module uses HPLC or 4G technology to realize remote data transmission with the cloud platform.

[0016] Preferably, a multi-source data processing method of an embedded dynamic monitoring module of an electric vehicle charging pile includes the steps of:

[0017] Data acquisition: collecting multi-source running data of the charging pile through the monitoring module, including voltage, current, power and electric quantity data;

[0018] Data anomaly identification: analyzing the use scenario, working process and running state of the charging pile, and identifying abnormal types such as data loss and data abnormal jump based on clustering algorithm and classification algorithm;

[0019] Data repair: repairing the missing data by using U-Net convolutional network;

[0020] Model establishment: establishing a charging pile conversion efficiency parameterization model and a remote metering error calculation model to realize online evaluation of the metering performance of the charging pile.

[0021] Preferably, characterized in that, data anomaly recognition: firstly analyze the use scene, the whole process and the running state of different types of charging piles, obtain the index information that can identify the running state of the charging pile in the running process of the charging pile through different sensors, carry out analog-digital conversion, fuse multi-source data, calculate the contribution rate of each index information to the charging pile metering error based on the principal component analysis idea, the contribution rate of each index information is respectively, and determine the main factors based on the contribution rate sorting, realize dimension reduction, and quantify the correlation degree of each index information through Pearson correlation

[0022]

[0023] In the formula, r is the radius of the nearest good cluster of the anomaly point, the Euclidean distance between the center point o of the good cluster and the anomaly point a.

[0024] Preferably, characterized in that, secondly, the clustering algorithm in the data mining theory is used to extract feature patterns from a large number of data curves that may contain abnormal data, and then the classification algorithm is used to separate the curve patterns containing abnormal data from the feature curve patterns to realize the recognition of abnormal data;

[0025] The clustering algorithm in the data mining theory is used to extract feature patterns from a large number of data curves that may contain abnormal data, and then the classification algorithm is used to separate the curve patterns containing abnormal data from the feature curve patterns to realize the recognition of abnormal data, and the formula is as follows:

[0026]

[0027] Preferably, characterized in that, the data repair: in order to ensure the integrity of the charging pile power data, the project repairs the missing of multi-source data, studies the data-driven charging pile multi-source data missing completion method, and repairs the missing data by using the U-Net convolution network;

[0028] According to the characteristics of charging pile multi-source heterogeneous data, the charging pile data is formed into a spatio-temporal tensor data format available for convolution operation, the multi-source data is spliced into an input vector graph, and then the coding and decoding ability of the network is used to realize the reconstruction of the missing data.

[0029] Preferably, characterized in that, the conversion efficiency parameterization model:

[0030] Firstly, the corresponding file information of the charging station and the charging pile is obtained, and according to the relevant specifications of the charging station design and the corresponding electrical physical principles of the direct current power module, the electrical topological relationship between the charging facilities in the charging station and the correlation between the charging settlement power and the total metering power in the station are established;

[0031] Then, the embedded charging pile operation state monitoring module is installed in the pilot charging station to record the electric signals of the input side and output side of the charging pile, so as to monitor the conversion efficiency of the power module;

[0032] Finally, the instantaneous current and voltage data of the charging pile are acquired by the embedded charging pile operation state monitoring module, and the obtained data are calculated locally to obtain the power and energy curves of the input side and output side of the charging pile;

[0033] During the establishment process, the voltage and current recording data of the input side and output side of the charging pile are acquired by the monitoring module, the power and energy curves are calculated, the conversion efficiency characteristic curve and its relationship with the working state are obtained through processing and analysis, the model is established in combination with the conversion efficiency statistical results and the physical principles and design specifications of the power module.

[0034] Preferably, the remote metering error calculation model comprises:

[0035] Firstly, based on the topology structure of the load of the electric vehicle centralized charging station, the parameters of the model are determined, and a remote metering error calculation model of the charging pile is established according to the generalized energy conservation theorem, as follows:

[0036]

[0037] Secondly, considering the complex working conditions and data transmission errors on site, disturbance modeling is performed, and a disturbance term is added to the model; then, the stability of the model is evaluated, the convergence and divergence of the model are studied, and the stability of the model is improved through variational decomposition; the model characteristics are analyzed, such as measuring the stability of the model according to the condition number, and an optimization solving method with strong robustness of the model is proposed; further, the overcharged charging pile detected remotely is detected on site to test the accuracy of the algorithm;

[0038] Finally, by adding disturbance data, the influence of data uncertainty on the model and the error calculation accuracy is analyzed and judged, and the mean absolute percentage error and the root mean square error are used as the judgment basis, which are represented by and respectively:

[0039]

[0040] In the formula, p is the total number of samples, and and are the outputs of the disturbance data and the non-disturbance data respectively;

[0041] Based on the generalized energy conservation theorem, the disturbance term is added to consider the complex working conditions and data transmission errors on site, and the stability of the model is improved through variational decomposition.

[0042] Preferably, characterized in that the verification method is: first, according to the test point situation, select typical charging station data as sample data, according to the actual charging data and the conversion efficiency model, set the number of charging direct current piles, the number of alternating current piles, the power supply mode of the charging station, the line loss and the fixed loss in the station, establish a variety of charging station models, and add the line loss item and the fixed loss item, establish a charging station simulation model, simulate the ideal charging station data of various scenes, including charging settlement data and total metering data;

[0043] Then, on the basis of ideal data, generate data containing metering error in accordance with error distribution law through simulation, and add the scene of quantization error and clock deviation, superimposed on the charging station data, generate model verification data with metering error and known error and each parameter true value;

[0044] Finally, according to the time and space dimension requirements of the model reliability, the model verification index and evaluation dimension are established;

[0045] Among them, the verification index includes detection rate, false detection rate and accuracy, and the evaluation dimension includes charging station scale, charging station type, average charging per pile per day, and average use per pile per day;

[0046] The model verification data obtained by simulation is substituted into the model and calculated to obtain the calculation results of the corresponding charging pile error parameters and other parameters of each charging station based on blind test data, the statistical rules of the model verification indexes of detection rate, false detection rate and accuracy are established, the division standards of the evaluation dimensions of charging station scale, charging station type, average charging per pile per day and average use per pile per day are determined, and the error calculation results of the test data are counted to obtain the results of the corresponding verification indexes.

[0047] Compared with the prior art, the present application has the following beneficial effects:

[0048] 1. An electric vehicle charging pile embedded dynamic monitoring module and a multi-source data processing method, the embedded monitoring module supports multiple communication modes such as HPLC, 4G and RS-485, can flexibly adapt to complex communication scenarios of charging facilities, and can collect multi-source operation data such as voltage, current and power of the input side and the output side of the charging pile in real time, build a comprehensive and stable data acquisition system for remote online monitoring of the charging pile, and solve the problems of single data source and limited transmission of traditional monitoring data.

[0049] 2. An embedded dynamic monitoring module for electric vehicle charging piles and a multi-source data processing method, which has functions such as protocol conversion, power quality analysis, and single-point flexible control. Protocol conversion breaks down communication barriers between different devices, ensuring effective fusion of multi-source data; power quality analysis can identify voltage sag, harmonics and other problems, providing quality verification for subsequent data processing and measurement evaluation; single-point flexible control combined with external circuit breaker control can accurately intervene in the operation state of the charging pile, improve the intelligent control level of the charging pile, and reduce the risk of abnormal operation.

[0050] 3. An embedded dynamic monitoring module for electric vehicle charging piles and a multi-source data processing method based on the principal component analysis idea, which calculates the contribution rate of each index to the measurement error of the charging pile, and combines the Pearson correlation coefficient

[0051] to quantify the correlation degree of the index, screen the key factors, reduce redundant calculation, and improve the data processing efficiency. At the same time, the core index affecting the measurement error is clearly defined, providing a direction for abnormal tracing.

[0052] 4. An embedded dynamic monitoring module for electric vehicle charging piles and a multi-source data processing method, which uses FCM combined with genetic algorithm for data classification, and then fine-tunes the abnormal points through K-nearest neighbor algorithm to accurately identify data missing, jumping and other abnormalities. Compared with traditional single algorithm, this collaborative strategy enhances the ability to identify abnormal data in complex scenarios, ensuring data quality and providing a reliable data foundation for subsequent model establishment.

[0053] 5. An embedded dynamic monitoring module for electric vehicle charging piles and a multi-source data processing method, which uses U-Net convolutional network to construct the multi-source data of the charging pile into a spatiotemporal tensor input through compression-expansion path and skip connection, effectively reconstructing the missing data. The repair process does not require human intervention, adapts to the dynamic data characteristics of the charging pile, fills the data missing gap, ensures data integrity, and solves the problem of measurement evaluation deviation caused by data missing.

[0054] 6. An embedded dynamic monitoring module for electric vehicle charging piles and a multi-source data processing method, which provides a whole-process technical solution from data acquisition, processing to performance evaluation, enabling the charging pile to have more intelligent state perception and self-diagnosis capabilities, meeting the development needs of the electric vehicle industry for efficient, accurate and intelligent charging facilities, accelerating the intelligent upgrading process of charging piles, and promoting the collaborative development of the new energy vehicle industry chain.

[0055] 7. An embedded dynamic monitoring module for electric vehicle charging piles and a multi-source data processing method, which provides quantitative and accurate monitoring and evaluation tools for charging pile operation enterprises and regulatory departments. Operation enterprises can optimize operation and maintenance plans based on data to reduce equipment downtime. Regulatory departments can use online evaluation to standardize market order, improve overall industry management level, reduce social resource waste caused by measurement disputes and equipment failures, and produce significant economic and social benefits. BRIEF DESCRIPTION OF DRAWINGS

[0056] Fig. 1 is a charging pile state monitoring and multi-source data anomaly processing technology architecture of the present application;

[0057] Fig. 2 is a monitoring module implementation scheme of the present application;

[0058] Fig. 3 is a schematic diagram of installation of a direct current charging pile edge computing module of the present application. DETAILED DESCRIPTION

[0059] The embodiments of the present application will be further described in detail below in conjunction with the drawings and examples. The following examples are used to illustrate the present application, but cannot be used to limit the scope of the present application.

[0060] Please refer to Figs. 1-3 The present application provides an embedded dynamic monitoring module for electric vehicle charging piles and a multi-source data processing method. An embedded monitoring module is developed to collect charging facility state data in real time, providing multi-source data support for remote online monitoring of charging piles. For the multi-element characteristics of AC and DC charging facilities such as structure, communication, and load, a prototype of a charging facility embedded monitoring module is developed to support load state sensing and adaptive access, support remote real-time communication with charging facilities, and support flexible interaction and aggregated regulation and control of charging facility loads.

[0061] Multi-source data anomaly identification and repair technology supporting online evaluation of charging pile metering performance

[0062] Different types of information acquisition methods are studied to support online evaluation of charging facility metering performance, and the data characteristics of multi-source information are statistically analyzed. The anomaly identification method of multi-source measurement data is studied to achieve efficient and accurate information processing, improve error identification of big data, and remove noise. Based on the data anomaly identification results, the potential laws and internal relations among multi-source data are mined, and the repair method for missing and abnormal problems of charging facility multi-source measurement big data is studied to improve data quality and provide good basic data for remote online monitoring of charging piles.

[0063] An embedded monitoring module is developed to collect charging facility state data in real time, providing multi-source data support for remote online monitoring of charging piles

[0064] The three aspects include charging facility load state perception and adaptive access technology, remote real-time communication technology with charging facilities, and flexible interaction and aggregation control technology with charging facility load.

[0065] The monitoring module is used for collecting various load information of a single charging facility, and can exchange data with a collection terminal device. The monitoring module is composed of a data processing unit, a communication unit, a security unit, a storage unit, and an electric energy measurement unit, and has the functions of protocol conversion, power quality analysis, single-point flexible control, external circuit breaker control, information storage and processing, software upgrade, HPLC communication (4G communication), RS-485 communication (RS232 communication), CAN communication, and Bluetooth communication. Finally, the development of the monitoring module is completed. The implementation scheme is shown in Fig. 2 .

[0066] Supporting the online evaluation of the multi-source data anomaly identification and repair technology of the charging pile metering performance

[0067] Firstly, the use scenarios, whole process and running state of different types of charging piles are analyzed. The index information measurement technology that can identify the running state of the charging pile in the running process of the charging pile is obtained through different sensors, analog-digital conversion is performed, multi-source data is fused, the contribution rate of each index information to the charging pile metering error is calculated based on the principal component analysis idea, the contribution rates of the index information are respectively and the main factors are determined based on the contribution rate ranking to realize dimension reduction. The correlation degree of each index information is quantified through Pearson correlation

[0068]

[0069] In the formula, r is the radius of the nearest good cluster of the abnormal point, and the Euclidean distance between the center point o of the good cluster and the abnormal point a.

[0070] Secondly, common data anomaly types include data missing and data abnormal jump. The clustering algorithm in the data mining theory is used to extract feature patterns from a large number of data curves that may contain abnormal data, and then the classification algorithm is used to separate the curve patterns containing abnormal data from the feature curve patterns to realize the identification of abnormal data. Through the combination of FCM and genetic algorithm, the training database is classified to obtain multiple good clusters and a small amount of abnormal data. The K nearest neighbor algorithm is used to push the abnormal points to the good cluster to realize fine tuning of the abnormal data and accurately identify the abnormal data in the big data. The formula is as follows:

[0071]

[0072] Finally, in order to ensure the integrity of the charging pile power data, the project repairs the missing data of multiple sources, studies the charging pile multi-source data missing completion method based on data driving, and plans to use U-Net convolutional network to repair the missing data. U-Net is an unsupervised deep learning network that can reconstruct missing data through end-to-end learning. The network consists of a compression path and an expansion path. The network is similar in structure to a convolutional autoencoder, except that it has a skip connection designed specifically for the encoding and decoding parts, which allows the network to capture more spatiotemporal detail information. According to the characteristics of charging pile multi-source heterogeneous data, the charging pile data is formed into a spatiotemporal tensor data format that can be used for convolutional operations, and the multi-source data is spliced into an input vector graph. Then, the network's encoding and decoding capabilities are used to reconstruct the missing data.

[0073] Research on DC charging pile AC-DC conversion characteristics and conversion efficiency - charging state quantitative parameterization model

[0074] First, the corresponding file information of the charging station and the charging pile is obtained, and the electrical topological relationship between the charging facilities in the charging station and the correlation between the charging settlement power and the total metering power in the station are established according to the relevant design specifications of the charging station and the corresponding electrical and physical principles of the DC power module. Through the research content (1), the charging operation data and the charging station total metering data collected by the embedded charging pile operation state monitoring module are processed, such as metering cycle alignment, to obtain data that can be used for power comparison. Using the above data, the average conversion efficiency of the corresponding charging pile is calculated according to different batches, and the average conversion efficiency of each charging station is calculated as an evaluation index of the conversion efficiency of the charging facility, and as the basis for establishing the conversion efficiency model.

[0075] Then, the embedded charging pile operation state monitoring module is installed in the pilot charging station to record the electrical signals on the input side and output side of the charging pile, and to monitor the conversion efficiency of the power module. The installation is shown in Fig. 3 The voltage and current recording data of the AC input side and DC output side of the charging pile are obtained.

[0076] Finally, the instantaneous current and voltage data of the charging pile are obtained by high-frequency acquisition through the embedded charging pile operation state monitoring module, and the obtained data is calculated locally to obtain the power and energy curves of the charging pile output side and output side. The conversion efficiency curve, input side current, voltage, power curve, output side current, voltage, power curve, etc. are processed and analyzed by multiple point merging, averaging, etc. to obtain the conversion efficiency characteristic curve of the charging pile, and its relationship with the working state (power, current, voltage). Based on the above data analysis results, and the conversion efficiency statistical results, and by comprehensively investigating the physical principles and design specifications of the power module, the conversion efficiency parameterization model of the charging pile is established.

[0077] The application discloses a charging station operation state monitoring model based on charging pile operation and transaction data and a model uncertainty evaluation method, and realizes remote online monitoring of the charging pile

[0078] Firstly, based on the topology structure of the load of the centralized charging station of the electric vehicle, the parameters of the model are determined, and a remote metering error calculation model of the charging pile is established according to the generalized energy conservation theorem, as follows:

[0079]

[0080] Secondly, considering the complex working conditions and data transmission errors on site, disturbance modeling is carried out, and a disturbance term is added to the model; model stability evaluation is carried out, the convergence of the model is studied, and the stability of the model is improved through variational decomposition; model characteristic analysis is carried out, such as measuring the stability of the model according to the condition number, and a strong robustness optimization solving method of the model is proposed; further, on-site detection is carried out on the charging pile with an error detected remotely, to test the accuracy of the algorithm (the accuracy calculation method is: accuracy η = actual error pile number / remote detection pile number).

[0081] Finally, by adding disturbance data, the influence of data uncertainty on the model and the error calculation accuracy is analyzed and judged, and the mean absolute percentage error and the root mean square error are used as the judgment basis, and e MAPE and e RMSE are respectively represented as:

[0082]

[0083] In the formula, p is the total number of samples, y' i and y i are respectively the disturbance data output and the non-disturbance data output.

[0084] Charging station metering performance monitoring model evaluation and verification technology research based on simulation of multiple charging station scenarios

[0085] Firstly, according to the trial situation, typical charging station data is selected as sample data and the charging pile metering error result data obtained by the charging pile remote verification technology in the task 2, according to the actual charging data and the conversion efficiency model, multiple charging DC pile numbers, AC pile numbers, charging station power supply modes (high supply high metering, high supply low metering, etc.), station internal line loss and fixed loss conditions and the like are set, multiple charging station models are established, and line loss items, fixed loss items and the like are added, and a charging station simulation model is established to simulate ideal charging station data of multiple scenarios, including charging settlement data and total metering data.

[0086] Then, on the basis of the ideal data, data containing metering errors in conformity with error distribution rules are generated through simulation, and quantitative errors, clock deviation and the like are added to the charging station data to generate charging station data with metering errors and known error

[0087] Difference and model validation data of each parameter true value.

[0088] Finally, according to the time and space dimension requirements of model reliability, the model validation index and evaluation dimension are established. The validation index includes detection rate, false detection rate and accuracy rate, and the evaluation dimension includes charging station scale, charging station type (AC, DC, hybrid), average daily charging per pile, average daily use frequency per pile, etc. The model validation data obtained by simulation is substituted into the model and calculated to obtain the calculation results of the error parameters and other parameters of each charging pile of the charging station based on the blind test data, the statistical rules of the model validation indexes such as detection rate, false detection rate and accuracy rate are established, the division standards of the evaluation dimensions such as charging station scale, charging station type, average daily charging per pile and average daily use frequency per pile are determined, and the error calculation results of the test data are counted to obtain the results of the corresponding validation indexes.

[0089] Embodiments of the present application are given for the purpose of illustration and description, and are not intended to be exhaustive or to limit the application to the forms disclosed. Many modifications and variations will be apparent to those of ordinary skill in the art. Embodiments were chosen and described in order to best explain the principles of the application and its practical application, and to enable others skilled in the art to understand the application for various embodiments with various modifications as are suited to the particular use contemplated.

Claims

1. An embedded dynamic monitoring module for electric vehicle charging piles and a multi-source data processing method, characterized in that: The monitoring module collects charging facility state data in real time to provide multi-source data support for remote online monitoring of charging piles; The monitoring module is used for collecting various load information of a single charging facility, and can exchange data with a collection terminal device, and is composed of a data processing unit, a communication unit, a security unit, a storage unit and an electric energy measurement unit; The data processing unit is used for operation and analysis of the collected charging facility state data; The communication unit supports HPLC, 4G, RS485, CAN and Bluetooth communication modes, and realizes data interaction with the charging facility and the cloud server; The security unit has a built-in encryption algorithm and signature verification function to ensure the security of data transmission and storage; The storage unit is used for storing the collected raw data and the processed result data; The electric energy measurement unit has a B-level measurement accuracy level and is used for collecting voltage, current and power electric parameter data of the charging facility; The state indication unit displays the operating state through an indicator light display module.

2. The embedded dynamic monitoring module and multi-source data processing method for electric vehicle charging piles according to claim 1, characterized in that, The dynamic monitoring module has the functions of protocol conversion, power quality analysis, single-point flexible control, external circuit breaker control, information storage and processing, and software upgrading, and has HPLC communication, RS-485 communication, CAN communication and Bluetooth communication, and finally completes the development of the monitoring module.

3. The embedded dynamic monitoring module and multi-source data processing method for electric vehicle charging piles according to claim 2, characterized in that, The communication unit includes a local communication module and a remote communication module, the local communication module realizes near-distance data interaction with the charging pile by using RS485 or CAN protocol, and the remote communication module realizes remote data transmission with the cloud platform by using HPLC or 4G technology.

4. A multi-source data processing method of an embedded dynamic monitoring module of an electric vehicle charging pile, according to any one of claims 1-3, characterized in that, The method comprises the following steps: Data collection: collecting multi-source operation data of the charging pile through the monitoring module, including voltage, current, power and electric quantity data; Data anomaly identification: analyzing the use scene, working process and operating state of the charging pile, identifying abnormal types such as data loss and data abnormal jump based on clustering algorithm and classification algorithm; Data repair: repairing the missing data by using U-Net convolutional network; Model establishment: establishing a charging pile conversion efficiency parameterization model and a remote metering error calculation model to realize online evaluation of the metering performance of the charging pile.

5. The multi-source data processing method of the embedded dynamic monitoring module of the electric vehicle charging pile according to claim 4, characterized in that, Data anomaly identification: first, analyze the use scene, working process and operating state of different types of charging piles, obtain index information of the charging pile operating state through different sensor acquisition technologies during the charging pile operation process, perform analog-digital conversion, fuse multi-source data, calculate the contribution rate of each index information to the charging pile metering error based on the principal component analysis idea, and determine the main factors based on the contribution rate sorting to realize dimension reduction, and quantify the correlation degree of each index information through Pearson correlation In the formula, r is the radius of the nearest good cluster of the abnormal point, and the Euclidean distance between the center point o of the good cluster and the abnormal point a.

6. The multi-source data processing method of the embedded dynamic monitoring module of the electric vehicle charging pile according to claim 5, characterized in that, Secondly, the clustering algorithm in the data mining theory is used to extract feature patterns from a large number of data curves that may contain abnormal data, and then the classification algorithm is used to separate the curve patterns containing abnormal data from the feature curve patterns to realize the identification of abnormal data; The feature mode is extracted from a large number of data curves that may contain abnormal data by using a clustering algorithm in the data mining theory, and then a classification algorithm is used to separate the curve mode containing abnormal data from the feature curve mode to realize the identification of abnormal data, and the formula is as follows:

7. The multi-source data processing method of the embedded dynamic monitoring module of the electric vehicle charging pile according to claim 6, characterized in that, In order to ensure the integrity of the charging pile power data, the project repairs the multi-source data missing, studies the charging pile multi-source data missing completion method based on data driving, and repairs the missing data by using the U-Net convolution network. According to the characteristics of charging pile multi-source heterogeneous data, the charging pile data is formed into a spatio-temporal tensor data format available for convolution operation, the multi-source data is spliced into an input vector graph, and then the coding and decoding ability of the network is used to realize the reconstruction of the missing data.

8. The multi-source data processing method of the embedded dynamic monitoring module of the electric vehicle charging pile according to claim 7, characterized in that, The conversion efficiency parameterization model: Firstly, the corresponding file information of the charging station and the charging pile is obtained, and the electrical topological relationship between the charging facilities in the charging station and the correlation between the charging settlement power and the total meter power in the station are established according to the relevant design specifications of the charging station and the corresponding electrical physical principles of the DC power module; Then, the embedded charging pile operation state monitoring module is installed in the pilot charging station to record the electrical signals of the input side and the output side of the charging pile, so as to monitor the conversion efficiency of the power module; Finally, the instantaneous current and voltage data of the charging pile are obtained by high-frequency acquisition through the embedded charging pile operation state monitoring module, and the power and energy curves of the input side and the output side of the charging pile are obtained by local calculation; During the establishment process, the voltage and current recording data of the input side and the output side of the charging pile are obtained through the monitoring module, the power and energy curves are calculated, the conversion efficiency characteristic curve and its relationship with the working state are obtained through processing and analysis, and the model is established by combining the conversion efficiency statistical results, the physical principles of the power module and the design specifications.

9. The multi-source data processing method of the embedded dynamic monitoring module of the electric vehicle charging pile according to claim 8, characterized in that, The remote metering error calculation model: Firstly, based on the topological structure of the load of the electric vehicle centralized charging station, the parameters of the model are determined, and the remote metering error calculation model of the charging pile is established according to the generalized energy conservation theorem, as follows: Secondly, considering the complex working conditions and data transmission errors on site, the disturbance term is added to the model; then the stability of the model is evaluated, the convergence and divergence of the model are studied, and the stability of the model is improved through variational decomposition; the model characteristics are analyzed, such as measuring the stability of the model according to the condition number, and an optimization solving method with strong robustness of the model is proposed; further, the on-site detection of the charging pile with excessive error detected by remote detection is carried out to test the accuracy of the algorithm; Finally, the influence of data uncertainty on the model and the error calculation accuracy is analyzed and evaluated by adding disturbance data, and the mean absolute percentage error and the root mean square error are used as evaluation criteria, which are represented by and respectively: In the formula, p is the total number of samples, and and are the output of the disturbance data and the output of the non-disturbance data respectively; Based on the generalized energy conservation theorem, the disturbance term is added to consider the complex working conditions and data transmission errors on site, and the stability of the model is improved through variational decomposition.

10. The multi-source data processing method of the embedded dynamic monitoring module of the electric vehicle charging pile according to claim 9, characterized in that, The verification method is: first, according to the test point situation, select typical charging station data as sample data, according to the actual charging data and conversion efficiency model, set the number of charging direct current piles, the number of alternating current piles, the power supply mode of charging station, the line loss and fixed loss in the station, establish a variety of charging station models, add line loss and fixed loss, establish a charging station simulation model, simulate ideal charging station data of various scenes, including charging settlement data and total metering data; Then, on the basis of ideal data, generate data containing measurement error conforming to error distribution law through simulation, and add the scene of quantization error and clock deviation, superimposed on the charging station data, generate model verification data with measurement error and known error and true value of each parameter; Finally, according to the time and space dimension requirements of model reliability, establish model verification index and evaluation dimension; Among them, the verification index includes detection rate, false detection rate and accuracy, and the evaluation dimension includes charging station scale, charging station type, average charging per pile per day, and average use per pile per day; The model verification data obtained by simulation is substituted into the model and calculated to obtain the calculation results of the error parameters and other parameters of each charging pile of the charging station based on the blind test data, establish the statistical rules of the model verification indexes of detection rate, false detection rate and accuracy, determine the division standards of the evaluation dimensions of charging station scale, charging station type, average charging per pile per day and average use per pile per day, and count the error calculation results of the test data to obtain the results of the corresponding verification indexes.

Images