Service stage determination method, service stage determination device, operation and maintenance method and medium

Abstract

The application provides a service stage determination method for a charging pile gun head, a service stage determination device, an operation and maintenance method and a computer readable storage medium, wherein the service stage comprises a health stage and an abnormal stage, and the service stage determination method comprises the following steps: acquiring a time series data set about a to-be-tested gun head; acquiring a time series feature matrix of a preset dimension of the time series data set and a center distance of the time series feature matrix relative to a reference feature matrix, wherein the reference feature matrix is constructed in advance based on a gun head in the health stage; and determining the service stage of the to-be-tested gun head based on a comparison result of the center distance and a preset threshold value, wherein the preset threshold value is determined in advance based on a gun head in the abnormal stage. The service stage determination method can determine the running condition of the gun head more simply and efficiently, and makes it possible for the early intervention of relevant measures.

Description

Technical Field

[0001] This invention relates to a method for determining the service phase of a charging station nozzle, a service phase determination apparatus, a maintenance method for a charging station nozzle, and a computer-readable storage medium for performing such a service phase determination method. The charging station nozzle can be used to provide charging services for electric vehicles, and may relate to pure electric vehicles and hybrid electric vehicles. Background Technology

[0002] The emergence of charging stations, especially supercharging stations, has alleviated or eliminated range anxiety for electric vehicle owners. To achieve fast charging of the battery, supercharging stations on the market can provide charging power of 480kW or charging voltage as high as 800V. However, with the increased frequency of supercharging, the temperature rise and wear of the charging gun heads intensify, accelerating their failure. Generally, charging gun heads are replaced when they reach their failure point or service time thresholds. This one-size-fits-all approach to supervision and maintenance is not only wasteful of resources but also inefficient. Furthermore, calling in maintenance personnel only when the failure point is reached introduces a delay. Therefore, there is an urgent need for a comprehensive supervision mechanism and method that can accurately and proactively predict charging gun head failures, thereby improving the efficiency of charging gun head maintenance. Summary of the Invention

[0003] Depending on the specific aspects, the present invention aims to provide a method, apparatus, maintenance method, and computer-readable storage medium for determining the service phase of a charging station gun.

[0004] Furthermore, the present invention aims to solve or alleviate other technical problems existing in the prior art.

[0005] This invention addresses the aforementioned problems by providing a service phase for a charging station gun head. Specifically, the service phase includes a healthy phase and an abnormal phase, and the method for determining the service phase includes the following steps:

[0006] Obtain the time-series dataset for the test nozzle;

[0007] Obtain the temporal representation matrix of the time series dataset with a preset dimension and the center distance of the temporal representation matrix relative to the reference representation matrix, wherein the reference representation matrix is ​​pre-constructed based on the gun head in the healthy stage;

[0008] Based on the comparison between the center distance and the preset threshold, the service stage of the test head is determined, wherein the preset threshold is pre-determined based on test heads in an abnormal stage.

[0009] According to a service phase determination method proposed in one aspect of the present invention, the abnormal phase includes a flow-limiting phase and a degradation phase, and the preset threshold correspondingly includes a flow-limiting preset threshold and a degradation preset threshold, wherein the flow-limiting preset threshold is predetermined based on the nozzle in the flow-limiting phase, and the degradation preset threshold is predetermined based on the nozzle in the degradation phase.

[0010] According to one aspect of the service phase determination method of the present invention, the acquisition of the time series dataset about the test head includes the following sub-steps:

[0011] Obtain a feature dataset of the charging service for the gun head under test at a fixed time step, wherein the fixed time step covers at least one charging service;

[0012] Based on the assigned timestamp information, the feature datasets are integrated into a time-series dataset;

[0013] The element dataset is associated with the duration of the charging service, the ambient temperature during the charging service, and the temperature of the charging gun during the charging service.

[0014] According to one aspect of the present invention, the service phase determination method obtains the temporal representation matrix and the center distance by means of a trained representation learning model, the representation learning model including a reservoir computation sub-model, a temporal representation sub-model and a center distance generation sub-model.

[0015] According to the service phase determination method proposed in one aspect of the present invention, the time series representation sub-model solves for the time series representation matrix C of the time series dataset using the following formula:

[0016]

[0017] Wherein, E(t) i () represents the loss function;

[0018] x(t i ) indicates that at t i The output of the reservoir calculation sub-model at time step;

[0019] || || 2 This indicates a search for the norm.

[0020] δ is a control parameter and is fixed;

[0021] This indicates the calculation of the Frobenius norm.

[0022] According to one aspect of the service phase determination method proposed by the present invention, the solution of the time series characterization matrix C is optimized using the spectral clustering method, wherein the time series characterization matrix C is expressed as:

[0023] C=R(R+δ 2 I) -1 ,

[0024] Where R = E(t) i )[x(t i )x(t i ) T ];

[0025] I is an N*N identity matrix;

[0026] δ is a control parameter and is fixed.

[0027] According to one aspect of the service phase determination method of the present invention, the representation learning model is trained based on an unsupervised learning method according to the following steps:

[0028] Obtain the health sample dataset, current-limiting sample dataset, and degradation sample dataset of the gun head;

[0029] Based on a portion of the rate-limiting sample dataset and a portion of the degraded sample dataset, the representation learning model is pre-trained and pre-tested to obtain the rate-limiting preset threshold and the degraded preset threshold.

[0030] Based on another portion of the flow-limited sample dataset and another portion of the degraded sample dataset, the representation learning model is fused and retrained until the loss function of the representation learning model converges.

[0031] According to the service phase determination method proposed in one aspect of the present invention, the acquisition of the health sample dataset, the current-limiting sample dataset, and the degradation sample dataset of the nozzle includes the following sub-steps:

[0032] Acquire charging service data for multiple charging heads throughout their entire lifecycle and categorize it into health service data, current-limiting service data, and degradation service data.

[0033] Using the fixed time step, the health service data, flow restriction service data, and degradation service data of each nozzle are divided and integrated into the health sample dataset, flow restriction sample dataset, and degradation sample dataset, respectively.

[0034] According to one aspect of the service phase determination method proposed by the present invention, the flow limiting preset threshold and the degradation preset threshold are obtained by using a clustering method or an averaging method.

[0035] According to another aspect of the invention, a service phase determination apparatus for performing such a service phase determination method is also provided, comprising:

[0036] The acquisition module is configured to acquire a time-series dataset about the gun head under test;

[0037] The representation learning module is configured to obtain a temporal representation matrix of a preset dimension for the temporal dataset and the center distance of the temporal representation matrix relative to a reference representation matrix, wherein the reference representation matrix is ​​pre-constructed based on the gun head in a healthy phase.

[0038] The determination module is configured to determine the service stage of the test head based on the comparison result between the center distance and a preset threshold, wherein the preset threshold is pre-determined based on test heads in an abnormal stage.

[0039] According to another aspect of the present invention, a method for the operation and maintenance of charging pile gun heads is also proposed, which includes the following steps:

[0040] Obtain the attribute information of the test nozzle within a preset spatial range;

[0041] Based on the service stage determination method described above, the service stage of each test head is determined.

[0042] Information about the service phase of each identified test head is integrated and displayed.

[0043] According to another aspect of the present invention, the operation and maintenance method includes an abnormal phase and a degradation phase. If it is determined that the test head is in the current limiting phase, the maximum charging voltage or maximum charging power of the test head is reduced. If it is determined that the test head is in the degradation phase, the test head is replaced.

[0044] Finally, according to another aspect of the invention, a computer-readable storage medium is also provided, on which a computer program is stored, wherein the computer program, when executed by a processor, implements the service phase determination method for charging pile gun heads described above.

[0045] By introducing continuous time-series datasets and time-series representation matrices, this service phase determination method can determine the operating status of the gun head in a simple and efficient manner, and makes it possible to intervene in related measures in advance. Attached Figure Description

[0046] Referring to the accompanying drawings, the above and other features of the present invention will become apparent, wherein,

[0047] Figure 1 and Figure 2 The main steps of the service phase determination method according to the present invention are shown respectively;

[0048] Figure 3A simplified structural diagram of the service phase determination device according to the present invention is shown. Detailed Implementation

[0049] It is readily understood that, based on the technical solution of this invention, those skilled in the art can propose various interchangeable structural methods and implementations without altering the essential spirit of the invention. Therefore, the following detailed embodiments and accompanying drawings are merely illustrative examples of the technical solution of this invention and should not be considered as the entirety of the invention or as limitations or restrictions on the technical solution of this invention.

[0050] The directional terms such as up, down, left, right, front, back, front, back, top, and bottom mentioned or possibly used in this specification are defined relative to the structures shown in the accompanying drawings. These are relative concepts and may therefore vary depending on their location and usage. Therefore, these or other directional terms should not be interpreted as restrictive. Furthermore, the terms "first," "second," "third," and similar expressions are used for descriptive and distinguishing purposes only and should not be construed as indicating or implying the relative importance of the corresponding components.

[0051] The following is combined with Figure 1 and 2 To elaborate on the method according to the present invention in more detail, and in conjunction with Figure 3 The following describes in detail the apparatus for performing such a method, which is used to determine the service stage of a charging gun under test (of a charging station), particularly its current service stage. Here, the service stage determination apparatus according to the invention is used to detect the charging gun under test at predetermined time intervals or in real time. Exemplarily, this apparatus can be integrated into the charging station, particularly into the associated control system of the charging station; that is, the service stage determination apparatus is assigned to the charging station under its responsibility. In this case, the apparatus can exist as an edge computing device and be effectively wired connected to the charging gun under test and associated sensors.

[0052] Instead, the service phase determination device can be positioned remotely from the charging gun under test, particularly from the entire charging station, and communicate with the charging gun under test and its associated sensors, particularly wirelessly. For example, the service phase determination device can exist as a cloud device or be integrated into a cloud server, whereby it communicates with the charging gun under test and its associated sensors via cloud communication. In this way, it becomes possible for the service phase determination device to simultaneously handle multiple charging guns under test.

[0053] Specifically, the entire service phase of a charging station gun is divided into a healthy phase or an abnormal phase. The healthy phase indicates that the charging station gun's operating status meets the requirements and does not require maintenance or adjustment. In contrast, the abnormal phase indicates that relevant measures need to be taken to ensure the completion of the charging task. These measures may involve not only replacing the gun or its related components, but also adjusting the gun's characteristic parameters (such as maximum charging voltage, maximum charging current, or maximum charging power).

[0054] Optionally, the service phases of the charging gun can also be divided as follows: the abnormal phase further includes a current-limiting phase and a degradation phase. The current-limiting phase indicates that the charging gun's operating state has deviated to a certain extent (this could be caused, for example, by gun wear), and for safety reasons, its maximum charging current / voltage / power needs to be limited. Similarly, the degradation phase indicates that the charging gun can no longer continue to be used and needs to be replaced, for example, when the charging gun is severely worn. Dividing the abnormal phase into a current-limiting phase and a degradation phase more closely reflects the characteristics of the charging gun's service cycle, and compared to a two-phase division method, this can reduce the charging gun's maintenance costs while ensuring normal operation. The invention will be further described below with reference to this three-phase division method.

[0055] Overall, the service stage determination method is based on the following concept: by comparing the characteristic information of the historical data of the test gun during the charging service with a predetermined benchmark, the service stage of the gun can be determined. The predetermined benchmark represents the general characteristics of each service stage or a certain service stage.

[0056] Based on this, the service phase determination method according to the present invention generally includes the following steps S100, S200 and S300 executed in sequence, which will be described step by step below.

[0057] First, regarding step S100: acquiring time-series data about the charging station under test, this can be performed by the acquisition module 110 of the service phase determination device 100 according to the present invention, wherein the acquisition module communicates with the charging station under test or its associated sensors in order to retrieve relevant data. Preferably, to ensure timely data reception, when the service phase determination device is located at or near the charging station, the acquisition module 110 is wired to the charging station under test or its associated sensors. Conversely, when the service phase determination device is located away from the charging station under test, for example, at an operation and maintenance center, the acquisition module 110 communicates wirelessly with the charging station under test or its sensors.

[0058] Here, the time-series dataset may involve the operating status parameters of the charging gun during the charging service, or environmental parameters during the charging service, or a combination of both. This time-series dataset is a multi-feature dataset, optionally including the following types of data: data characterizing the duration of the charging service for the tested charging gun, data characterizing the ambient temperature during the charging service, and data characterizing the temperature of the charging gun during the charging service (including the starting temperature of the charging service, the ending temperature of the charging service, and the real-time temperature of the charging gun at each moment). These types of data, after preprocessing, can be used to obtain data that can be input into the model described below through physical modeling. For example, a fitting error operation is performed on the original data to reduce the difference between the extracted features and the original data, thereby reducing data noise. Specifically, a multi-feature time-series dataset including the following six types of information can be input into the representation learning model described below: charging service duration data, ambient temperature data, charging service starting temperature data, charging service ending temperature data, charging gun temperature data, and fitting error data, wherein the fitting error is used to preprocess the other data mentioned above. For simplicity, the duration, ambient temperature, and nozzle temperature can be expressed as standardized, dimensionless values.

[0059] Furthermore, this time-series dataset can also include current parameters, voltage parameters, and resistance parameters derived from the charging gun during the charging service, which can also serve as the basis for obtaining the aforementioned types of data. The types of feature parameters included in the time-series dataset can be modified accordingly based on specific needs.

[0060] Optionally, when testing the charging gun head under test, the time-series dataset is obtained as follows: A feature dataset of the charging gun head under test is acquired at a fixed time step regarding charging services; and then, based on the timestamp information associated with each feature data point, the feature datasets are integrated into a time-series dataset. The specified fixed time step should cover at least one charging service; for example, the fixed time step can be set to one week. Generally, a fixed time step of one week can cover both peak and off-peak charging periods.

[0061] As a component of time-series datasets, feature datasets also involve the types of data mentioned above. For example, each type of data from the previous week constitutes a feature dataset. Multiple feature datasets are then arranged sequentially according to timestamp information to form a time-series dataset that can serve as the basis for subsequently determining service stages. A multi-feature time-series dataset is defined in a matrix sense as X∈R. F×T Where F represents the number of feature types; T represents the fixed time step mentioned above (which can also be called the time series length, such as one week mentioned above). Therefore, it can be concluded that... in, This represents the timestamp data of the i-th feature at the j-th time point.

[0062] Compared to observing the instantaneous operating parameters of the charging station gun and determining the service status based on the parameters at this point in time, the present invention, based on continuous data and multiple feature parameters, can eliminate external interference and thereby improve the detection accuracy of the charging station's service phase.

[0063] Furthermore, the service stage determination method according to the present invention further includes the following step S200: obtaining a time-series representation matrix of a preset dimension of the time-series dataset and the center distance of the time-series representation matrix relative to a reference representation matrix, wherein the reference representation matrix is ​​pre-constructed based on probes in a healthy stage. Here, the time-series representation matrix carries dynamic information and feature information (e.g., trends, periods, waveforms) of the data in the time-series dataset, and the center distance relative to the reference representation matrix implies the degree of difference between the features of the observed time-series dataset and the features of the reference representation matrix. In a practical sense, this center distance can be used to characterize the difference between the probe to be tested and a reference healthy probe. By quantitatively assessing this degree of difference, the service stage of the probe to be tested can be determined.

[0064] Based on this, the operating status of the test head is evaluated according to the size of the center distance. That is, the service stage determination method includes step S300: determining the service stage of the test head based on the comparison result of the center distance and the preset threshold, wherein the preset threshold is pre-determined based on the test head in the abnormal stage.

[0065] Specifically, when the service phase is divided into a healthy phase and an abnormal phase, if the calculated center distance is near or falls within a preset threshold range of a certain value, the test probe can be determined to be in an abnormal phase and appropriate measures need to be taken. Conversely, if the calculated center distance does not fall into the above categories, the test probe can be determined to be in a healthy phase.

[0066] More specifically, when the service phase is divided into a health phase, a flow-limiting phase, and a degradation phase, there can be a flow-limiting preset threshold (pre-determined based on the probe being in the flow-limiting phase) and a degradation preset threshold (pre-determined based on the probe being in the degradation phase). In step S300, the calculated center distance is compared with these two preset thresholds respectively. By determining which of the flow-limiting preset thresholds and degradation preset thresholds the calculated center distance is near, or particularly within the range defined by either of them, the service phase of the probe being tested can be determined. If the calculated center distance is neither near nor within the range defined by either of these preset thresholds, the probe being tested can be determined to be in a healthy state.

[0067] Accordingly, step S300 can be performed by means of a determination module 130 of the service stage determination device according to the invention, which is downstream of the representation learning module 120, which will be described below, and whose input is communicatively connected to the output of the representation learning module 120. The determination module 130 is configured to determine the service stage of the test gun head based on a comparison between the center distance and a preset threshold.

[0068] Preferably, the preset threshold for the center distance includes a numerical range, which to some extent provides redundancy and favorable early prediction for the service stage determination method. For example, in a dimensionless sense, the current limiting preset threshold is set to 0.3-0.5, and the degradation preset threshold can be set to 0.5-0.8. When the calculated center distance is close to 0.3, the charging head under test is determined to be in the current limiting stage, thereby limiting the maximum current of the charging head in advance and ensuring its safe operation. In addition, the specific position of the calculated center distance within 0.3-0.5 can indicate the degree of abnormality of the charging head or its service stage in the near future, which is undoubtedly beneficial to charging pile maintenance personnel. Specifically, if the calculated center distance is 0.48, the maintenance personnel can recognize that although the charging head under test is in the current limiting stage, it will soon enter the degradation stage. Based on this recognition, replacement operations can be performed in advance under certain circumstances.

[0069] The temporal representation matrix used above and its central distance relative to the reference representation matrix can be obtained using a neural network model or other arbitrary deep learning model. Optionally, a method based on an unsupervised recurrent neural network model is provided, wherein step S200 can be performed using a trained representation learning model, which includes a reservoir calculation sub-model, a temporal representation sub-model, and a central distance generation sub-model. That is, the service phase determination device 100 according to the present invention can also have a representation learning module 120, whose input is communicatively connected to the output of the acquisition module 110 and on which a representation learning model is deployed, wherein the representation learning module 120 is configured to acquire a temporal representation matrix of a preset dimension of the aforementioned temporal dataset and calculate the central distance of the temporal representation matrix relative to the reference representation matrix.

[0070] First, the representation learning model and its initial training process are described. The training step S400 of this model includes the following sub-steps:

[0071] S410: Obtain the health sample dataset, current-limiting sample dataset, and degradation sample dataset of the nozzle;

[0072] S420: Based on a portion of the rate-limiting sample dataset and a portion of the degraded sample dataset, the representation learning model is pre-trained and pre-tested, whereby the rate-limiting preset threshold and the degraded preset threshold can be obtained;

[0073] S430: Based on another part of the flow-limited sample dataset and another part of the degraded sample dataset, the representation learning model is fused and retrained until the loss function of the representation learning model converges.

[0074] The training process of this representation learning model is divided into a pre-training process and a retraining process. The pre-training process is mainly used to obtain the preset thresholds mentioned above (e.g., rate limiting and degradation preset thresholds). The retraining process is used to fuse and match the model with the input time-series dataset. After obtaining the preset thresholds, the loss function and retraining strategy of the representation learning model are designed. Training stops when the loss on the validation dataset does not decrease for several consecutive rounds (i.e., the loss function converges), and the model with the optimal weight parameters in the reserve pool calculation submodule is saved at this point. During the pre-training and retraining processes, the training sample dataset is divided into a training sample dataset and a validation sample dataset according to a certain ratio. The model is trained multiple times using the training sample dataset, stopping when the loss on the validation sample dataset does not decrease for several consecutive rounds (i.e., the loss function converges). Here, to ensure training stability in the early stages, a warm-up strategy is used for the training learning rate, which initially increases and then gradually decreases.

[0075] For step S410, a large amount of charging service data for charging guns is obtained from the database, and the charging service data for each charging gun is divided according to the service stage. For each charging gun, a sample dataset in the form of the time-series dataset X defined above is obtained.

[0076] Specifically, firstly, charging service data for the entire lifecycle of multiple charging nozzles within the maintenance scope is acquired and divided into health service data, current-limiting service data, and degradation service data, i.e., sub-step S411. Next, in the subsequent sub-step S412, using a sliding time window method with a fixed time step, the health service data, current-limiting service data, and degradation service data for each charging nozzle are divided to obtain time series for training, and then integrated into the health sample dataset, current-limiting sample dataset, and degradation sample dataset, respectively.

[0077] The sample dataset assigned to a specific service phase (see the time-series dataset X defined above) is then fed into the reservoir computation sub-model, which is similar to a large sparse unsupervised recurrent neural network. In this sub-model, the connectivity and weight parameters of each neuron are fixed and randomly assigned when each neuron receives a nonlinear response signal. Flexible learning is achieved by processing the weight parameters in this way. This reservoir computation sub-model includes an input layer, a reservoir computation layer, and an output layer, and can be described by the following formula:

[0078] x(t i )=f(W in u(t i )+Wx(t i-1 )),

[0079] Where x(t) i ) indicates that at t i The current computational state at time t; u(t) i ) indicates that at t i The input to the sample dataset at time t; x(t) i-1 ) expressed in t i-1 The calculation state at time; W in W is the learning parameter matrix and is fixed after training.

[0080] Next, the current computational state x(t) of the reserve pool computational sub-model is calculated. i The input is fed to the temporal representation sub-model, which generates a fixed-dimensional temporal representation matrix C based on a variable-length multi-feature sample dataset (or, when performing detection using a trained representation learning model, a temporal dataset for the gun head to be tested). Here, the temporal representation sub-model can be described as a filter for the reservoir sub-model, where the representation matrix C can be defined as minimizing the current computational state of the reservoir:

[0081]

[0082] Here, E(t) i () represents the loss function;

[0083] x(t i ) indicates that at t i The output of the reservoir calculation sub-model at time step;

[0084] || || 2 This indicates a search for the norm.

[0085] δ is a control parameter and is fixed after training;

[0086] This indicates the calculation of the Frobenius norm.

[0087] Optionally, a simplified method for solving the representation matrix is ​​provided, namely, using spectral clustering to obtain the time series representation matrix C. The time series representation matrix C is:

[0088] C=R(R+δ 2 I) -1 ,

[0089] Where R = E(t) i )[x(t i )x(t i ) T ];

[0090] I is an N*N identity matrix;

[0091] δ is a control parameter and is fixed.

[0092] In the initial training phase, a large number of temporal representation matrices can be obtained from the healthy sample dataset using the above method. The baseline representation matrix corresponding to the healthy stage is obtained by averaging these matrices. Here, clustering methods, such as density clustering or Gaussian clustering, can also be used to obtain this baseline representation matrix, which serves as the basis for calculating the current limiting preset threshold and degradation preset threshold, as well as the center distance in subsequent detection processes. For the trained service stage determination process, a defined temporal representation matrix can be obtained for the input temporal dataset.

[0093] During training, multiple time-series representation matrices C are obtained based on the current-limited sample dataset and the degraded sample dataset. i And calculate its relationship with the benchmark representation matrix C. j Center distance Dis ij The center is located near Dis ij , is represented as:

[0094] Dis ij =||C i -C j || fro .

[0095] Similarly, during the training process (including the pre-training and retraining processes), the distribution of multiple center distances that match it can be studied by averaging or clustering methods, thereby obtaining a representative center threshold and setting the current limiting preset threshold and degradation preset threshold based on it.

[0096] In the actual service phase determination process, the acquired time-series dataset of the test gun head is input into the trained representation learning model. This allows the calculation of its associated time-series representation matrix and center distance. The service phase can be determined by the relationship between these matrix and the preset current-limiting and degradation thresholds. The derivation of this time-series representation matrix and center distance can be referenced from the model training process described above.

[0097] Furthermore, the present invention also relates to a maintenance method for charging pile gun heads, comprising the following steps:

[0098] Obtain the attribute information of the test nozzle within a preset spatial range;

[0099] Based on the service stage determination method described above, the service stage of each test head is determined.

[0100] Information about the service phase of each identified test head is integrated and displayed.

[0101] Here, the attribute information includes the spatial location of the charging station nozzle, the time when the charging station nozzle was put into use, and the specific model of the charging station nozzle, etc. In this way, maintenance personnel can be given more intuitive guidance information about abnormal charging station nozzles.

[0102] In a more specific embodiment, the service stages of all test probes within the preset spatial range can be displayed to relevant maintenance personnel in the form of a dashboard or chart (e.g., using a human-computer interaction device), so that maintenance personnel can gain a general understanding of the probes and facilitate the analysis of the probes' operating status.

[0103] Optionally, if it is determined that the test head is in the current-limiting stage, the maximum charging voltage or maximum charging power of the test head is reduced; if it is determined that the test head is in the deterioration stage, the test head is replaced.

[0104] Finally, the present invention also relates to a computer-readable storage medium for implementing a service phase determination method for charging station gun heads according to one or more embodiments of the present invention, having stored thereon a computer program, wherein the computer program, when executed by a processor, implements the service phase determination method for charging station gun heads described above. The computer-readable storage medium referred to herein includes various types of computer storage media, and can be any available medium accessible to a general-purpose or special-purpose computer. For example, a computer-readable storage medium may include RAM, ROM, EPROM, E2PROM, registers, hard disk, removable disk, CD-ROM or other optical disk storage, magnetic disk storage or other magnetic storage devices, or any other temporary or non-temporary medium capable of carrying or storing desired program code units having the form of instructions or data structures and accessible by a general-purpose or special-purpose computer, or a general-purpose or special-purpose processor. The description of the computer-readable storage medium according to the present invention can be referenced to the explanation of the service phase determination method according to the present invention, and will not be repeated here.

[0105] It should be understood that all the above preferred embodiments are exemplary and not restrictive, and various modifications or variations made by those skilled in the art to the specific embodiments described above under the concept of the present invention should be within the legal protection scope of the present invention.

Claims

1. A method for determining the service stage of a charging station gun head, characterized in that, The service phase includes a healthy phase and an abnormal phase, and the method for determining the service phase includes the following steps: Obtain the time-series dataset for the test nozzle; Obtain the temporal representation matrix of the time series dataset with a preset dimension and the center distance of the temporal representation matrix relative to the reference representation matrix, wherein the reference representation matrix is ​​pre-constructed based on the gun head in the healthy stage; Based on the comparison between the center distance and a preset threshold, the service stage of the test probe is determined, wherein the preset threshold is pre-determined based on probes in an abnormal stage. The abnormal phase includes a flow-limiting phase and a degradation phase. Correspondingly, the preset thresholds include a flow-limiting preset threshold and a degradation preset threshold. The flow-limiting preset threshold is pre-determined based on the nozzle in the flow-limiting phase, and the degradation preset threshold is pre-determined based on the nozzle in the degradation phase. The temporal representation matrix and the center distance are obtained using a trained representation learning model, wherein the representation learning model is trained based on an unsupervised learning method according to the following steps: Obtain the health sample dataset, current-limiting sample dataset, and degradation sample dataset of the gun head; Based on a portion of the rate-limiting sample dataset and a portion of the degraded sample dataset, the representation learning model is pre-trained and pre-tested to obtain the rate-limiting preset threshold and the degraded preset threshold. Based on another portion of the flow-limited sample dataset and another portion of the degraded sample dataset, the representation learning model is fused and retrained until the loss function of the representation learning model converges.

2. The service stage determination method according to claim 1, characterized in that, The process of obtaining the time-series dataset about the gun head under test includes the following sub-steps: Obtain a feature dataset of the charging service for the gun head under test at a fixed time step, wherein the fixed time step covers at least one charging service; Based on the assigned timestamp information, the feature datasets are integrated into a time-series dataset; The element dataset is associated with the duration of the charging service, the ambient temperature during the charging service, and the temperature of the charging gun during the charging service.

3. The service stage determination method according to claim 1, characterized in that, The representation learning model includes a reservoir computation sub-model, a time-series representation sub-model, and a center distance generation sub-model.

4. The service stage determination method according to claim 3, characterized in that, The temporal representation submodel solves for the temporal representation matrix C of the temporal dataset using the following formula: ， in, Represents the loss function; Indicates in The output of the reservoir calculation sub-model at time step; This indicates a search for the norm. These are control parameters and are fixed. This indicates the calculation of the Frobenius norm.

5. The service stage determination method according to claim 4, characterized in that, The temporal representation matrix C is optimized using the spectral clustering method. The temporal representation matrix C is expressed as: ， in, ; It is an N*N identity matrix; These are control parameters and are fixed.

6. The service stage determination method according to claim 1, characterized in that, The process of obtaining the healthy sample dataset, the current-limiting sample dataset, and the degraded sample dataset of the gun head includes the following sub-steps: Acquire charging service data for multiple charging heads throughout their entire lifecycle and categorize it into health service data, current-limiting service data, and degradation service data. With a fixed time step, the health service data, flow restriction service data, and degradation service data of each nozzle are divided and integrated into the health sample dataset, flow restriction sample dataset, and degradation sample dataset, respectively.

7. The service stage determination method according to claim 6, characterized in that, The current limiting preset threshold and the degradation preset threshold are obtained by using clustering methods or averaging methods.

8. A service phase determination device for a charging pile gun head, characterized in that, It is used to perform the service phase determination method according to any one of claims 1 to 7 and includes: The acquisition module is configured to acquire a time-series dataset about the gun head under test; The representation learning module is configured to obtain a temporal representation matrix of a preset dimension for the temporal dataset and the center distance of the temporal representation matrix relative to a reference representation matrix, wherein the reference representation matrix is ​​pre-constructed based on the gun head in a healthy phase. The determination module is configured to determine the service stage of the test head based on the comparison result between the center distance and a preset threshold, wherein the preset threshold is pre-determined based on test heads in an abnormal stage.

9. A method for the operation and maintenance of charging pile gun heads, characterized in that, Includes the following steps: Obtain the attribute information of the test nozzle within a preset spatial range; Based on the service stage determination method according to any one of claims 1 to 7, the service stage of each test head is determined; Information about the service phase of each identified test head is integrated and displayed.

10. The operation and maintenance method according to claim 9, characterized in that, The abnormal phase includes a current limiting phase and a degradation phase. If it is determined that the test head is in the current limiting phase, the maximum charging voltage or maximum charging power of the test head is reduced. If it is determined that the test head is in the degradation phase, the test head is replaced.

11. A computer-readable storage medium on which a computer program is stored, characterized in that, When the computer program is executed by the processor, it implements the service phase determination method for charging pile gun heads according to any one of claims 1 to 7.

Images