New energy charging pile fault prediction and maintenance system and method based on Internet of Things

Abstract

The invention relates to the technical field of new energy charging pile fault prediction and maintenance, in particular to a new energy charging pile fault prediction and maintenance system and method based on the Internet of Things, and the method comprises the steps: collecting data, generating multi-dimensional information of a charging pile, processing the multi-dimensional information of the charging pile, obtaining the operation data of the charging pile based on the Internet of Things, and carrying out the fault prediction and maintenance of the charging pile. And comparing and analyzing the operation data of the charging pile, determining a charging pile fault prediction method based on the Internet of Things, and implementing an alarm decision based on the charging pile fault prediction method. According to the method, high-frequency hard data acquisition is abandoned, and the cost and difficulty of acquiring data required for fault prediction of the charging pile are reduced on the basis of acquisition of related soft data; when the charging pile is in the standby state and the working state, soft data are processed and analyzed, advanced prediction and warning of the fault of the charging pile are achieved, the freedom of self-fault prediction of the charging pile is improved, and hidden dangers of the charging pile can be eliminated in advance.

Description

Technical Field

[0001] The present invention relates to the technical field of new energy charging pile fault prediction and maintenance, and in particular to a new energy charging pile fault prediction and maintenance system and method based on the Internet of Things. Background Art

[0002] With the country's strong support for the development of electric vehicles, the construction of new energy charging piles that provide electricity for electric vehicles has also developed rapidly. Nowadays, many regions have built charging stations of varying sizes. At the same time, the frequency of use of charging piles in charging stations is getting higher and higher, which makes the possibility of failure and damage of charging piles gradually increase. Therefore, the safety management of charging piles has become particularly important.

[0003] A Chinese patent with authorization announcement number CN118246750B discloses an abnormal diagnosis and early warning system for charging piles based on artificial intelligence; it includes a charging pile operation and management platform, an information collection unit, a status supervision unit, a charging safety assessment unit, a self-test charging analysis unit, a fusion feedback unit, a fault prediction assessment unit, and a management execution unit; in the above patent document, its early warning system evaluates the safety and performance of the charging pile based on the real-time data of the charging pile, diagnoses the charging pile based on the evaluation results, and then issues relevant early warnings, which is real-time. At the same time, the data required for the early warning are hard data with high collection frequency such as real-time current, real-time voltage, and real-time temperature. The cost of collecting and analyzing relevant hard data on each charging pile is extremely high, and the requirements for system data processing are extremely high, which is easy to cause redundancy and has great limitations; at the same time, the technical support it provides does not have the freedom to predict faults in the standby and working states of the charging pile, resulting in a delay in the discovery of charging pile problems, and cannot provide technical personnel with advance inspections of safety hazards in the charging pile, and the effect is not good. Summary of the invention

[0004] The purpose of the present invention is to address the problems existing in the background technology and to propose a new energy charging pile fault prediction and maintenance system and method based on the Internet of Things.

[0005] The technical solution of the present invention: a new energy charging pile fault prediction and maintenance method based on the Internet of Things, comprising the following steps: S1. Collect the vehicle model connected to the charging pile, the actual charge amount of the vehicle, the output power of the charging pile, the number of charging pile services, the number of charging pile maintenance and the current health status of the charging pile, and determine the multi-dimensional information of the charging pile based on the Internet of Things; S2. Process the multi-dimensional information of the charging pile to obtain the charging pile operation data based on the Internet of Things; S3. Compare and analyze the charging pile operation data to determine the charging pile fault prediction method based on the Internet of Things; S4. Implement an alarm decision based on the charging pile fault prediction method, and notify the technical staff to perform maintenance on the charging pile based on the implementation of the alarm decision.

[0006] Preferably, for S2, the method for processing the multi-dimensional information of the charging pile includes the following steps: Storing multi-dimensional information of charging piles and establishing a dimensional information library of multi-dimensional information of charging piles; Obtain multi-dimensional information of the most recent k1 charging piles in the dimensional information database and establish a first sample set; Acquire the multi-dimensional information of the most recent k2 charging piles under the unique identification of the charging pile in the dimensional information library, and establish a second sample set, wherein the first sample set and the second sample set are mutually related sample sets; Determine whether the charging pile is in operation; if the charging pile is in operation, establish a feature vector based on the current multi-dimensional information of the charging pile, construct sample vectors corresponding to the multi-dimensional information of several charging piles in the first sample set, and construct a sample vector set based on the several sample vectors; Generate charging pile operation data based on the first sample set, the second sample set, the feature vector and the sample vector set; If the charging pile is in standby mode, the charging pile operation data is generated based on the first sample set and the second sample set.

[0007] Preferably, for S3, the method for comparing and analyzing the charging pile operation data is as follows: S301, performing abnormal information identification on a first sample set to obtain an abnormal information identification result, and calculating a first risk parameter of the associated sample set based on the identification result; S302, conditionally constraining the associated sample set; S303, performing dimensional correlation analysis on the associated sample set to obtain analysis results, and calculating a second risk parameter of the associated sample set based on the analysis results; S304, calculating the average correlation between the feature vector and the sample vector set based on whether the charging pile operation data contains the feature vector and the sample vector set; S305 . Perform weighted processing on the second risk parameter based on the average correlation degree, input the first risk parameter, the weighted second risk parameter and the average correlation degree into a fault risk prediction model, and output a fault risk value.

[0008] Preferably, with respect to S301, abnormal information is identified on the first sample set to obtain an abnormal information identification result, and a method for calculating a first risk parameter of the associated sample set based on the identification result is: Mark the health status of the charging piles in the multi-dimensional information of the charging piles in the sample set as the key search term, mark the multi-dimensional information of the charging piles with the key search term "fault" as abnormal information, calculate the frequency of occurrence of the abnormal information, and use it as the frequency of abnormal occurrence of the sample set; The frequency of abnormal occurrence in the first sample set is used as the risk frequency threshold, and the ratio of the number of multi-dimensional information of charging piles in the second sample set to the risk frequency threshold is calculated to obtain the first risk parameter α1 of the associated sample set.

[0009] Preferably, for S302, conditional constraints are imposed on the associated sample set, and the conditional constraints satisfy: Constraint 1: The vehicle models included in the k1 first information must include the vehicle models included in the k2 second information; Constraint 2: The frequency of abnormal occurrence of the first sample set is not 0; Constraint three: The frequency of abnormal occurrence of the second sample set is 0; Constraint condition 4: The number k1 of multi-dimensional information of charging piles contained in the first sample set and the number k2 of multi-dimensional information of charging piles contained in the second sample set must satisfy: k1>>k2>n, where n is the dimensionality value of the multi-dimensional information of charging piles.

[0010] Preferably, with respect to S303, a dimensional correlation analysis is performed on the associated sample set to obtain an analysis result, and a method for calculating a second risk parameter of the associated sample set based on the analysis result includes the following steps: By formula Calculate and obtain the power loss analysis value L of the associated sample set; Wherein, ELi is the charging coefficient corresponding to the vehicle model included in the first sample set, and ELj is the charging coefficient corresponding to the vehicle model included in the second sample set; the charging coefficient is obtained based on the vehicle charging big data; OEi is the output power of the charging pile included in the first sample set, and OEj is the output power of the charging pile included in the second sample set; IEi is the actual charge amount of the vehicle included in the first sample set, and IEj is the actual charge amount of the vehicle included in the second sample set; i is the multi-dimensional information number of the charging pile in the first sample set, i∈[1, k1], k1 is a positive integer; j is the multi-dimensional information number of the charging pile in the second sample set, i∈[1, k2], k2 is a positive integer; lc is the basic loss constant; The total number of charging services of the charging piles included in the multi-dimensional prediction information in the first sample set and the second sample set are counted respectively, and marked as T1 and T2 respectively; The maintenance times corresponding to the charging piles included in the multi-dimensional prediction information in the first sample set and the second sample set are counted respectively, and marked as W1 and W2 respectively; By formula Calculate and obtain the maintenance frequency analysis value P of the associated sample set; The second risk parameter α2 is calculated by the formula α2=β1×L+β2×P; where β1 and β2 are the risk coefficients of the power loss analysis value L and the maintenance frequency analysis value P, respectively.

[0011] Preferably, for S304, the method for calculating the average correlation between the feature vector and the sample vector set is: If the charging pile operation data includes a feature vector and a sample vector set, the cosine similarity between the feature vector and several sample vectors in the sample vector set is calculated using the cosine similarity algorithm, and the cosine similarity is used as the correlation between the feature vector and the sample vector. The correlation between the feature vector and several sample vectors is summed and averaged, and the calculated result is used as the average correlation between the feature vector and the sample vector set. If the charging pile operation data does not contain feature vectors and sample vector sets, no calculation is performed.

[0012] Preferably, for S305, the expression of the fault risk prediction model is: ; Where Ω is the average correlation between the feature vector and the sample vector set, X is the state variable, and F is the fault risk value; The specific method for implementing alarm decision-making based on the charging pile fault prediction method is: Perform data analysis on the fault risk value F. If F is less than 0, no alarm is issued. If F is not less than 0, an alarm is issued to notify the technicians to perform maintenance and set the health status of the charging pile to fault.

[0013] The present invention also discloses a new energy charging pile fault prediction and maintenance system based on the Internet of Things, which applies the new energy charging pile fault prediction and maintenance method based on the Internet of Things, and specifically includes: Charging operation management module, used to monitor and manage charging piles; The data collection module is used to collect the model of the vehicle connected to the charging pile, the actual charge amount of the vehicle, the output power of the charging pile, the number of charging pile services, the number of charging pile maintenance times and the current health status of the charging pile, and determine the multi-dimensional information of the charging pile based on the Internet of Things; the current health status of the charging pile is divided into good and faulty; The data processing module is used to process the multi-dimensional information of the charging pile to obtain the charging pile operation data based on the Internet of Things; The data analysis module is used to compare and analyze the charging pile operation data and determine the charging pile fault prediction method based on the Internet of Things; The maintenance alarm module is used to implement alarm decisions based on the charging pile fault prediction method. Based on the implementation of the alarm decision, the charging operation management module is used to notify the technicians to maintain the charging piles.

[0014] Preferably, the charging operation management module includes an identification unit and a supervision unit; An identification unit, used to add a unique identification to each charging pile, and bind the multi-dimensional data of the charging pile collected by the charging pile based on the unique identification of each charging pile; The supervision unit is used to monitor the status of the charging pile in real time and identify whether the charging pile is in operation or in standby.

[0015] Compared with the prior art, the above technical solution of the present invention has the following beneficial technical effects: The present invention abandons the collection of high-frequency hard data such as the real-time current, real-time voltage, connection confirmation voltage and real-time temperature of the charging head of each charging pile during charging. The data collection module collects the soft data related to the charging pile service, which greatly reduces the cost and difficulty of acquiring the data required for the prediction of the charging pile fault. Based on the big data level, the power loss analysis value of the associated sample set is calculated for the actual charging amount of the vehicle by the charging pile and the output power of the charging pile. Based on the increase of the power loss analysis value, it can be found in time that when the corresponding charging pile in the second sample set fails, the heating increases, the connection voltage of the charging pile component is high, and the charging energy loss caused by the aging or damage of the charging pile component increases. In addition, based on the big data level, the maintenance frequency analysis value of the associated sample set is calculated for the service times and maintenance times of the charging pile. It can be known that for the charging pile corresponding to the second sample set, after reaching a certain number of charging service times, the hidden dangers increase; The present invention also has the advantage of analyzing soft data when the charging pile is in standby state and working state, outputting a fault risk value through a charging pile fault prediction method, realizing early prediction and alarm of charging pile faults, improving the freedom of self-fault prediction of the charging pile, and facilitating early elimination of hidden dangers of the charging pile and reducing the possibility of charging pile failures. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A flowchart of the method of Embodiment 1 proposed by the present invention; Figure 2 This is a flow chart of the method S3 in the first embodiment of the present invention; Figure 3 This is a module flow chart of the second embodiment of the present invention. DETAILED DESCRIPTION

[0017] Embodiment 1, as Figure 1 As shown, the new energy charging pile fault prediction and maintenance method based on the Internet of Things proposed by the present invention includes the following steps: S1. Collect the vehicle model connected to the charging pile, the actual charge amount of the vehicle, the output power of the charging pile, the number of charging pile services, the number of charging pile maintenance and the current health status of the charging pile, and determine the multi-dimensional information of the charging pile based on the Internet of Things; S2. Process the multi-dimensional information of the charging pile to obtain the charging pile operation data based on the Internet of Things; The method of S2 comprises the following steps: Storing multi-dimensional information of charging piles and establishing a dimensional information library of multi-dimensional information of charging piles; Obtain multi-dimensional information of the most recent k1 charging piles in the dimensional information database and establish a first sample set; Acquire the multi-dimensional information of the most recent k2 charging piles under the unique identification of the charging pile in the dimensional information library, and establish a second sample set, wherein the first sample set and the second sample set are mutually related sample sets; Determine whether the charging pile is in operation; if the charging pile is in operation, establish a feature vector based on the current multi-dimensional information of the charging pile, construct sample vectors corresponding to the multi-dimensional information of several charging piles in the first sample set, and construct a sample vector set based on the several sample vectors; Generate charging pile operation data based on the first sample set, the second sample set, the feature vector and the sample vector set; If the charging pile is in standby mode, generating charging pile operation data based on the first sample set and the second sample set; S3. Compare and analyze the charging pile operation data to determine the charging pile fault prediction method based on the Internet of Things; like Figure 2 As shown, the method of S3 includes the following steps: S301, performing abnormal information identification on a first sample set to obtain an abnormal information identification result, and calculating a first risk parameter of the associated sample set based on the identification result; In S301, the specific implementation method is to identify abnormal information on the first sample set to obtain an abnormal information identification result, and the method for calculating the first risk parameter of the associated sample set based on the identification result is: Mark the health status of the charging piles in the multi-dimensional information of the charging piles in the sample set as the key search term, mark the multi-dimensional information of the charging piles with the key search term "fault" as abnormal information, calculate the frequency of occurrence of the abnormal information, and use it as the frequency of abnormal occurrence of the sample set; The frequency of abnormal occurrence of the first sample set is used as the risk frequency threshold, and the ratio of the number of multi-dimensional information of charging piles in the second sample set to the risk frequency threshold is calculated to obtain the first risk parameter α1 of the associated sample set; S302, conditionally constraining the associated sample set; In S302, the specific implementation method is to associate the sample set with conditional constraints, and the conditional constraints meet: Constraint 1: The vehicle models included in the k1 first information must include the vehicle models included in the k2 second information; Constraint 2: The frequency of abnormal occurrence of the first sample set is not 0; Constraint three: The frequency of abnormal occurrence of the second sample set is 0; Constraint condition 4: The number k1 of multi-dimensional information of charging piles contained in the first sample set and the number k2 of multi-dimensional information of charging piles contained in the second sample set must satisfy: k1>>k2>n, where n is the dimension value of the multi-dimensional information of charging piles; it should be noted that the symbol ">>" means much greater than; S303, performing dimensional correlation analysis on the associated sample set to obtain analysis results, and calculating a second risk parameter of the associated sample set based on the analysis results; In S303, the specific implementation method is to perform dimensional correlation analysis on the associated sample set to obtain analysis results, and the method for calculating the second risk parameter of the associated sample set based on the analysis results includes the following steps: By formula Calculate and obtain the power loss analysis value L of the associated sample set; Wherein, ELi is the charging coefficient corresponding to the vehicle model included in the first sample set, and ELj is the charging coefficient corresponding to the vehicle model included in the second sample set; the charging coefficient is obtained based on the vehicle charging big data; OEi is the output power of the charging pile included in the first sample set, and OEj is the output power of the charging pile included in the second sample set; IEi is the actual charge amount of the vehicle included in the first sample set, and IEj is the actual charge amount of the vehicle included in the second sample set; i is the multi-dimensional information number of the charging pile in the first sample set, i∈[1, k1], k1 is a positive integer; j is the multi-dimensional information number of the charging pile in the second sample set, i∈[1, k2], k2 is a positive integer; lc is the basic loss constant; The total number of charging services of the charging piles included in the multi-dimensional prediction information in the first sample set and the second sample set are counted respectively, and marked as T1 and T2 respectively; The maintenance times corresponding to the charging piles included in the multi-dimensional prediction information in the first sample set and the second sample set are counted respectively, and marked as W1 and W2 respectively; By formula Calculate and obtain the maintenance frequency analysis value P of the associated sample set; It should be noted that the default maintenance times for charging piles is 1, that is, one maintenance is counted when the charging pile is manufactured or put into operation; The second risk parameter α2 is calculated by the formula α2=β1×L+β2×P; where β1 and β2 are the risk coefficients of the power loss analysis value L and the maintenance frequency analysis value P, respectively; In the formula, the calculation method of risk coefficients β1 and β2 is: Based on calculating the actual charging capacity of the vehicle and the output power of the charging pile of each charging pile in the multi-dimensional information of the first sample set, the proportion of charging power loss is calculated. Specifically, the loss of charging pile power is obtained by calculating the difference between the output power of the charging pile and the actual charging capacity of the vehicle, and then the ratio of the loss of charging pile power to the output power of the charging pile is calculated, and the calculated information is used as the proportion of charging power loss of the multi-dimensional information of the charging pile; the charging power loss proportions of the multi-dimensional information of several charging piles in the first sample set are summed and averaged to obtain the average of the sample set power loss proportion of the first sample set; Accordingly, the average value of the charging power loss ratio of the abnormal information in the first sample set is calculated and used as the average value of the abnormal power loss ratio; Calculate the ratio of the average value of abnormal power loss ratio to the average value of power loss ratio in the sample set to obtain the first coefficient parameter; Calculate the ratio of the total number of charging services T1 of the charging piles included in the multi-dimensional prediction information of the first sample set to the number of maintenance times W1 corresponding to T1 charging piles to obtain the maintenance frequency of the sample set; Correspondingly, the charging piles included in the abnormal information in the first sample set are counted as target charging piles, the total number of service times of the target charging piles is counted to obtain the total number of abnormal services, the total number of maintenance times of the target charging piles is counted to obtain the total number of abnormal maintenances, and the total number of abnormal services is calculated by ratio with the total number of abnormal maintenances to obtain the abnormal maintenance frequency; Calculate the ratio of the sample set maintenance frequency to the abnormal maintenance frequency to obtain a second coefficient parameter; The risk coefficients β1 and β2 are constrained, β1+β2=1, and the risk coefficients β1 and β2 are obtained based on the weight relationship between the first coefficient parameter and the second coefficient parameter; Specifically, for example, if the first coefficient parameter is 0.9 and the second coefficient parameter is 0.6, the risk coefficients β1 and β2 are 0.6 and 0.4 respectively; The advantages of the present invention are that the collection of high-frequency hard data such as the real-time current, real-time voltage, connection confirmation voltage and real-time temperature of the charging head of each charging pile during charging is abandoned, and the collection of soft data related to the charging pile service by the data collection module greatly reduces the cost and difficulty of data acquisition required for charging pile fault prediction. Based on the big data level, the power loss analysis value of the associated sample set is calculated for the actual charging amount of the vehicle by the charging pile and the output power of the charging pile. Based on the increase of the power loss analysis value, it can be timely discovered that when the corresponding charging pile in the second sample set fails, the heating increases, the connection voltage of the charging pile component is high, and the charging energy loss caused by the aging or damage of the charging pile component increases; and based on the big data level, the maintenance frequency analysis value of the associated sample set is calculated for the service times and maintenance times of the charging pile, and it can be known that for the charging pile corresponding to the second sample set, the hidden dangers increase after reaching a certain number of charging services. S304, calculating the average correlation between the feature vector and the sample vector set based on whether the charging pile operation data contains the feature vector and the sample vector set; In S304, the specific implementation method is that if the charging pile operation data includes a feature vector and a sample vector set, the cosine similarity between the feature vector and several sample vectors in the sample vector set is calculated using a cosine similarity algorithm, and the cosine similarity is used as the correlation between the feature vector and the sample vector, and the correlation between the feature vector and the several sample vectors is summed and averaged, and the calculated result is used as the average correlation between the feature vector and the sample vector set; If the charging pile operation data does not contain the feature vector and the sample vector set, no calculation is performed; the cosine similarity algorithm is an existing technology and will not be described in detail here; S305, weighting the second risk parameter based on the average correlation, inputting the first risk parameter, the weighted second risk parameter and the average correlation into a fault risk prediction model, and outputting a fault risk value; In S305, the specific implementation method is as follows: The expression of the failure risk prediction model is: ; Where Ω is the average correlation between the feature vector and the sample vector set, X is the state variable, and F is the fault risk value; It should be noted that when the charging pile is in standby state, the feature vector cannot be constructed, so the average correlation degree Ω does not exist at this time, and the state variable X takes 1; S4. Implementing an alarm decision based on the charging pile fault prediction method, and notifying the technicians to perform maintenance on the charging pile based on the implementation of the alarm decision; For S4, the specific implementation plan is to perform data analysis on the fault risk value F. If F is less than 0, no alarm is issued; if F is not less than 0, an alarm is issued to notify the technicians to perform maintenance, and the health status of the charging pile is set to fault; The present invention also has the advantage of analyzing soft data when the charging pile is in standby state and working state, outputting a fault risk value through a charging pile fault prediction method, realizing early prediction and alarm of charging pile faults, improving the freedom of self-fault prediction of the charging pile, and facilitating early elimination of hidden dangers of the charging pile and reducing the possibility of charging pile failures.

[0018] Embodiment 2, as Figure 3 As shown, the new energy charging pile fault prediction and maintenance system based on the Internet of Things proposed by the present invention is applied to the new energy charging pile fault prediction and maintenance method based on the Internet of Things proposed in Example 1, including: Charging operation management module, used to monitor and manage charging piles; The data collection module is used to collect the model of the vehicle connected to the charging pile, the actual charge amount of the vehicle, the output power of the charging pile, the number of charging pile services, the number of charging pile maintenance times and the current health status of the charging pile, and determine the multi-dimensional information of the charging pile based on the Internet of Things; the current health status of the charging pile is divided into good and faulty; The data processing module is used to process the multi-dimensional information of the charging pile to obtain the charging pile operation data based on the Internet of Things; The data analysis module is used to compare and analyze the charging pile operation data and determine the charging pile fault prediction method based on the Internet of Things; The maintenance alarm module is used to implement alarm decisions based on the charging pile fault prediction method, and to notify technicians to maintain the charging piles through the charging operation management module based on the implementation of the alarm decision; The charging operation management module includes an identification unit and a supervision unit; An identification unit, used to add a unique identification to each charging pile, and bind the multi-dimensional data of the charging pile collected by the charging pile based on the unique identification of each charging pile; The supervision unit is used to monitor the status of the charging pile in real time and identify whether the charging pile is in operation or in standby.

[0019] The embodiments of the present invention are described in detail above with reference to the accompanying drawings, but the present invention is not limited thereto, and various changes can be made within the knowledge scope of technicians in the relevant technical field without departing from the purpose of the present invention.

Claims

1. A new energy charging pile fault prediction and maintenance method based on the Internet of Things, characterized in that: The following steps are involved: S1. Collect the vehicle model connected to the charging pile, the actual charge amount of the vehicle, the output power of the charging pile, the number of charging pile services, the number of charging pile maintenance and the current health status of the charging pile, and determine the multi-dimensional information of the charging pile based on the Internet of Things; S2. Process the multi-dimensional information of the charging pile to obtain the charging pile operation data based on the Internet of Things; S3. Compare and analyze the charging pile operation data to determine the charging pile fault prediction method based on the Internet of Things; S4. Implement an alarm decision based on the charging pile fault prediction method, and notify the technical staff to perform maintenance on the charging pile based on the implementation of the alarm decision.

2. The method for fault prediction and maintenance of new energy charging piles based on the Internet of Things according to claim 1 is characterized in that: With respect to S2, the method for processing the multi-dimensional information of the charging pile includes the following steps: Storing multi-dimensional information of charging piles and establishing a dimensional information library of multi-dimensional information of charging piles; Obtain multi-dimensional information of the most recent k1 charging piles in the dimensional information database and establish a first sample set; Acquire the multi-dimensional information of the most recent k2 charging piles under the unique identification of the charging pile in the dimensional information library, and establish a second sample set, wherein the first sample set and the second sample set are mutually related sample sets; Determine whether the charging pile is in operation; if the charging pile is in operation, establish a feature vector based on the current multi-dimensional information of the charging pile, construct sample vectors corresponding to the multi-dimensional information of several charging piles in the first sample set, and construct a sample vector set based on the several sample vectors; Generate charging pile operation data based on the first sample set, the second sample set, the feature vector and the sample vector set; If the charging pile is in standby mode, the charging pile operation data is generated based on the first sample set and the second sample set.

3. The method for fault prediction and maintenance of new energy charging piles based on the Internet of Things according to claim 2 is characterized in that: For S3, the method for comparative analysis of charging pile operation data is as follows: S301, performing abnormal information identification on a first sample set to obtain an abnormal information identification result, and calculating a first risk parameter of the associated sample set based on the identification result; S302, conditionally constraining the associated sample set; S303, performing dimensional correlation analysis on the associated sample set to obtain analysis results, and calculating a second risk parameter of the associated sample set based on the analysis results; S304, calculating the average correlation between the feature vector and the sample vector set based on whether the charging pile operation data contains the feature vector and the sample vector set; S305 . Perform weighted processing on the second risk parameter based on the average correlation degree, input the first risk parameter, the weighted second risk parameter and the average correlation degree into a fault risk prediction model, and output a fault risk value.

4. The method for fault prediction and maintenance of new energy charging piles based on the Internet of Things according to claim 3 is characterized in that: In S301, abnormal information is identified on the first sample set to obtain an abnormal information identification result, and a method for calculating a first risk parameter of the associated sample set based on the identification result is: Mark the health status of the charging piles in the multi-dimensional information of the charging piles in the sample set as the key search term, mark the multi-dimensional information of the charging piles with the key search term "fault" as abnormal information, calculate the frequency of occurrence of the abnormal information, and use it as the frequency of abnormal occurrence of the sample set; The frequency of abnormal occurrence in the first sample set is used as the risk frequency threshold, and the ratio of the number of multi-dimensional information of charging piles in the second sample set to the risk frequency threshold is calculated to obtain the first risk parameter α1 of the associated sample set.

5. The method for fault prediction and maintenance of new energy charging piles based on the Internet of Things according to claim 4 is characterized in that: For S302, condition constraints are imposed on the associated sample set, and the condition constraints satisfy: Constraint 1: The vehicle models included in the k1 first information must include the vehicle models included in the k2 second information; Constraint 2: The frequency of abnormal occurrence of the first sample set is not 0; Constraint three: The frequency of abnormal occurrence of the second sample set is 0; Constraint condition 4: The number k1 of multi-dimensional information of charging piles contained in the first sample set and the number k2 of multi-dimensional information of charging piles contained in the second sample set must satisfy: k1>>k2>n, where n is the dimensionality value of the multi-dimensional information of charging piles.

6. The method for fault prediction and maintenance of new energy charging piles based on the Internet of Things according to claim 5 is characterized in that: In S303, a dimensional correlation analysis is performed on the associated sample set to obtain an analysis result. The method for calculating the second risk parameter of the associated sample set based on the analysis result includes the following steps: By formula Calculate and obtain the power loss analysis value L of the associated sample set; Wherein, ELi is the charging coefficient corresponding to the vehicle model included in the first sample set, and ELj is the charging coefficient corresponding to the vehicle model included in the second sample set; the charging coefficient is obtained based on the vehicle charging big data; OEi is the output power of the charging pile included in the first sample set, and OEj is the output power of the charging pile included in the second sample set; IEi is the actual charge amount of the vehicle included in the first sample set, and IEj is the actual charge amount of the vehicle included in the second sample set; i is the multi-dimensional information number of the charging pile in the first sample set, i∈[1, k1], k1 is a positive integer; j is the multi-dimensional information number of the charging pile in the second sample set, i∈[1, k2], k2 is a positive integer; lc is the basic loss constant; The total number of charging services of the charging piles included in the multi-dimensional prediction information in the first sample set and the second sample set are counted respectively, and marked as T1 and T2 respectively; The maintenance times corresponding to the charging piles included in the multi-dimensional prediction information in the first sample set and the second sample set are counted respectively, and marked as W1 and W2 respectively; By formula Calculate and obtain the maintenance frequency analysis value P of the associated sample set; The second risk parameter α2 is calculated by the formula α2=β1×L+β2×P; where β1 and β2 are the risk coefficients of the power loss analysis value L and the maintenance frequency analysis value P, respectively.

7. The method for fault prediction and maintenance of new energy charging piles based on the Internet of Things according to claim 6 is characterized in that: For S304, the method for calculating the average correlation between the feature vector and the sample vector set is: If the charging pile operation data includes a feature vector and a sample vector set, the cosine similarity between the feature vector and several sample vectors in the sample vector set is calculated using the cosine similarity algorithm, and the cosine similarity is used as the correlation between the feature vector and the sample vector. The correlation between the feature vector and several sample vectors is summed and averaged, and the calculated result is used as the average correlation between the feature vector and the sample vector set. If the charging pile operation data does not contain feature vectors and sample vector sets, no calculation is performed.

8. The method for fault prediction and maintenance of new energy charging piles based on the Internet of Things according to claim 7 is characterized in that: For S305, the expression of the fault risk prediction model is: ； Where Ω is the average correlation between the feature vector and the sample vector set, X is the state variable, and F is the fault risk value; The specific method for implementing alarm decision-making based on the charging pile fault prediction method is: Perform data analysis on the fault risk value F. If F is less than 0, no alarm is issued. If F is not less than 0, an alarm is issued to notify the technicians to perform maintenance and set the health status of the charging pile to fault.

9. A new energy charging pile fault prediction and maintenance system based on the Internet of Things, which applies the new energy charging pile fault prediction and maintenance method based on the Internet of Things as described in claim 1 above, characterized in that: include: Charging operation management module, used to monitor and manage charging piles; The data collection module is used to collect the vehicle model connected to the charging pile, the actual charge amount of the vehicle, the output power of the charging pile, the number of charging pile services, the number of charging pile maintenance and the current health status of the charging pile, and determine the multi-dimensional information of the charging pile based on the Internet of Things; The current health status of the charging pile is divided into good and faulty; The data processing module is used to process the multi-dimensional information of the charging pile to obtain the charging pile operation data based on the Internet of Things; The data analysis module is used to compare and analyze the charging pile operation data and determine the charging pile fault prediction method based on the Internet of Things; The maintenance alarm module is used to implement alarm decisions based on the charging pile fault prediction method. Based on the implementation of the alarm decision, the charging operation management module is used to notify the technicians to maintain the charging piles.

10. The new energy charging pile fault prediction and maintenance system based on the Internet of Things according to claim 9 is characterized in that: The charging operation management module includes an identification unit and a supervision unit; An identification unit, used to add a unique identification to each charging pile, and bind the multi-dimensional data of the charging pile collected by the charging pile based on the unique identification of each charging pile; The supervision unit is used to monitor the status of the charging pile in real time and identify whether the charging pile is in operation or in standby.

Images