A charging pile health state evaluation method and system based on multi-dimensional data driving

Abstract

The application discloses a kind of based on multi-dimensional data driving's charging pile health state evaluation method and system, by real-time acquisition charging pile's three dimensions of original data of communication condition, function condition and operating condition, the weight of each dimension internal index is automatically calculated using entropy weight method and CRITIC method geometric mean combination weighting, respectively generate communication score, function score and operating score, wherein operating score introduces operating degradation state detection mechanism and carries out dynamic correction;Three dimensions of score are three channels parallel hierarchical aggregation again, and the basic health degree score is obtained by secondary combination weighting and adding cross correction term;Finally, the trend health coefficient and fluctuation health coefficient are calculated by extracting historical score sequence, the basic health degree score is exponentially weighted dynamic compensation, and the final health degree score is output and divided into health grade.The application completely changes charging pile operation and maintenance mode from passive post-maintenance to active early warning, and significantly improves the reliability of charging pile.

Description

Technical Field

[0001] This invention belongs to the field of electric vehicle charging facilities and intelligent operation and maintenance technology, and in particular relates to a method and system for assessing the health status of charging piles based on multi-dimensional data. Background Technology

[0002] With the rapid growth of electric vehicle ownership, charging stations, as critical infrastructure, directly impact the user experience and the economic benefits for operating companies through their operational reliability and safety. Charging stations face complex operating conditions in practice, including frequent plugging and unplugging operations, high-power electrical stress, outdoor environmental corrosion, and communication network fluctuations. These factors combined can lead to a gradual decline in equipment performance, ultimately causing charging failures, safety hazards, and even equipment shutdowns. Therefore, scientifically assessing the health status of charging stations, promptly identifying potential faults, and implementing preventative maintenance have become urgent technical challenges for the industry.

[0003] In existing charging pile operation and maintenance management, a reactive fault handling mode based on threshold alarms is commonly adopted. Specifically, the operation and maintenance system only monitors a few key electrical parameters (such as output voltage, current, and temperature). When a parameter exceeds a preset upper or lower limit, an alarm is triggered, and maintenance personnel then go to the site to handle the issue. The limitation of this approach is that it cannot quantify the overall health status of the equipment, only providing a binary judgment of normal or fault. It lacks effective monitoring methods for gradual degradation processes such as slow parameter deterioration, communication quality decline, and partial failure of functional modules. Many hidden dangers have been developing for a considerable period of time before triggering an alarm, but maintenance personnel cannot detect them in advance. As a result, faults often emerge in the form of sudden power outages or charging interruptions, seriously affecting user experience and equipment availability. In addition, although there are some data monitoring solutions for charging piles in existing technologies, most of them only focus on a single dimension. For example, they only collect communication data to assess online rate, or only collect electrical parameters to assess output performance. They rarely comprehensively evaluate communication status, functional status, and operational status as an organic whole. This kind of one-dimensional assessment is prone to blind spots. For example, the electrical output indicators of a charging pile may be completely normal, but the card reader's recognition success rate may have dropped significantly. This problem cannot be detected by electrical parameters alone, and it will only be passively noticed when the user repeatedly fails to swipe the card.

[0004] Another type of existing technology attempts to guide operation and maintenance by establishing an equipment health scoring system. However, these solutions often rely heavily on the experience of human experts to determine the weights and scoring rules for each indicator. Different operation and maintenance personnel have subjective differences in their emphasis on different indicators, and as the equipment's operating environment changes and aging patterns become apparent, fixed weights and rules are difficult to adjust in a timely manner, resulting in insufficient objectivity and adaptability of the evaluation results. Furthermore, some methods employ fuzzy comprehensive evaluation or hierarchical analysis, which includes a large number of manually set membership functions and judgment matrices. This not only makes the implementation process cumbersome but also makes it difficult to avoid subjective bias.

[0005] Due to the aforementioned shortcomings and deficiencies, the current operation and maintenance of charging piles is characterized by high cost and low efficiency. Maintenance personnel often need to invest a lot of manpower in regular inspections and emergency repairs, but they still cannot prevent sudden failures. Equipment lifecycle management lacks data support, making it difficult to make scientific decisions on when to replace old equipment or adjust maintenance strategies. Summary of the Invention

[0006] To address the aforementioned technical issues, this invention provides a method and system for assessing the health status of charging piles based on multi-dimensional data. By collecting and intelligently integrating data from three dimensions—communication status, functional status, and operational status—of the charging pile equipment, a scientific comprehensive health scoring system is constructed. This accurately captures the non-linear degradation process of equipment performance, enabling a shift from "post-event maintenance" to "pre-event early warning" operation and maintenance mode.

[0007] Specifically, the technical solution provided by this invention is as follows: A method for assessing the health status of charging piles based on multidimensional data includes the following steps: S1. Collect raw indicators of the communication status, functional status and operational status of the charging pile during the evaluation period. S2. Standardize the raw indicators in the three dimensions of communication status, functional status and operation status respectively, and automatically calculate the weight of the indicators in each dimension. The weighted sum is used to obtain the raw scores of communication dimension, functional dimension and operation dimension. S3. Using the original scores of the communication dimension, the function dimension, and the operation dimension as input, calculate the secondary weights of the three dimensions again, obtain the preliminary fusion score by weighted summation, and introduce a cross correction term to obtain the basic health score. S4. Extract the historical basic health score sequence and calculate the trend health coefficient; at the same time, collect short-term data of auxiliary indicators and calculate the fluctuation health coefficient; weight and fuse the basic health score with the trend health coefficient and the fluctuation health coefficient to output the final health score.

[0008] Furthermore, in step S2: The weights of the internal indicators of each dimension are calculated using a combination of entropy weighting and CRITIC geometric mean weighting strategies. First, calculate the information entropy of the standardized indicator data:

[0009] in, Indicates the first j Information entropy of each indicator m The number of sampling points. For the first i The sampling point at the th sampling point j Standardized values ​​for each indicator; Then calculate the entropy weight method weights:

[0010] in, Indicates the first j Entropy weighting of each indicator n The total number of indicators within the dimension; Next, calculate the first... j Standard deviation of each indicator:

[0011] in, Indicates the first j The standard deviation of each indicator For the first time during the evaluation period j The average of the indicators; Calculate the next j Correlation coefficient between each indicator and other indicators within the dimension ; Then calculate the information content:

[0012] in, Indicates the first j The amount of information in each indicator; Finally, calculate the CRITIC method weights:

[0013] in, Indicates the first j CRITIC weighting of each indicator; The entropy weight method weights and the CRITIC method weights are then geometrically averaged and fused.

[0014] in, Indicates the first j The combined weights of each indicator; For each dimension, the original score for each dimension is obtained by weighted summation based on the combined weights of each indicator within the dimension.

[0015] Furthermore, in step S3: The past M The original scores of the communication dimension, the original scores of the function dimension, and the original scores of the operation dimension within each evaluation period are used as the input matrix. The score sequence of each dimension is regarded as an indicator. The entropy weight method weight and the CRITIC method weight are calculated respectively, and then the geometric mean is fused to obtain the secondary weight of each dimension. The entropy weight method calculates the weights as follows: for the th j First, calculate the information entropy in each dimension:

[0016] in, For the first i Within the evaluation period, the first j The original scores for each dimension; Then calculate the entropy weight method's second-order weights: ; The CRITIC method calculates the weights as follows: first calculate the weight of the first weight... j Standard deviation of the score series for each dimension:

[0017] Calculate the next j Correlation coefficients between one dimension and the other dimensions ; Then calculate the information content: ; Obtain the CRITIC method's quadratic weights: ; By fusing the entropy weight method's quadratic weights and the CRITIC method's quadratic weights, we obtain the quadratic weights for each dimension:

[0018] The preliminary fusion score is as follows:

[0019] in, , and The original scores are for the communication dimension, the functionality dimension, and the operation dimension, respectively. , and These are the corresponding secondary weights; Then, a cross-correction term is introduced:

[0020] The basic health score is then: .

[0021] Furthermore, the trend health coefficient mentioned in step S4 is calculated as follows: Extract the past T The historical baseline health score sequence for each assessment period is denoted as . ,in, t Indicates the current evaluation period. T The preset number of historical backtracking cycles; the time points are numbered as follows. The corresponding basic health score is recorded as Calculate the slope of the first-order linear regression: , ,

[0022] in, The slope of the first-order linear regression. It is the average value at different points in time. The mean of the basic health score; The trend health coefficient is defined as:

[0023] in, The preset trend correction amplitude coefficient is greater than 0. The preset trend decay index is greater than 0 and less than 1; For symbolic functions: when When +1 is taken, When -1 is taken, Take 0 at that time.

[0024] Furthermore, the calculation method for the fluctuation health coefficient in step S4 is as follows: Several auxiliary indicators sensitive to operational stability were selected, and the numerical sequences of each auxiliary indicator were collected over the most recent L evaluation periods; for the nth auxiliary indicator, its numerical sequence is denoted as... Calculate the mean of the sequence. and standard deviation This leads to the coefficient of variation of the auxiliary indicator. ; Let N be the total number of auxiliary indicators selected, then the fluctuation health coefficient is:

[0025] Fluctuation Health Index The value range is [0,1], and the more drastic the fluctuation, the smaller the value.

[0026] Furthermore, the final health score in step S4 is calculated as follows:

[0027] in, For the final health score, Basic health score, As the trend health coefficient, The fluctuation health coefficient; and The preset exponential adjustment coefficient and satisfying .

[0028] Preferably, the original score for the operational dimension is corrected by introducing a degradation coefficient based on the operational degradation status detection results: If the load rate of a charging pile exceeds a preset load rate threshold for multiple consecutive evaluation periods, and the actual output power is lower than a preset percentage of the rated power, it is determined to be in a degraded operation state, and a degrade factor is set. The value must be less than 1 within the preset range; otherwise... The original score for the operational dimension is corrected under the following formula when the rating is downgraded:

[0029] in, The overall efficiency of a charging pile is equal to the ratio of output electrical energy to input electrical energy. The power factor of a charging pile is equal to the ratio of active power to apparent power. The input voltage error of the charging pile is equal to the absolute value of the relative error between the actual input voltage and the rated input voltage. The output voltage error of the charging pile is equal to the absolute value of the relative error between the actual output voltage and the target output voltage. For voltage regulation accuracy, For stable flow accuracy, This is the historical average load rate. This represents the current load rate.

[0030] Furthermore, the voltage regulation accuracy The calculation method is as follows: In constant voltage charging mode, the instantaneous voltage value at the output terminal is continuously collected at a frequency of not less than 1 Hz. Take all sampling points The maximum value divided by the target voltage value Multiply by 100%; the current stabilization accuracy The calculation method is as follows: In constant current charging mode, the instantaneous current value at the output terminal is continuously collected at a frequency of not less than 1 Hz. Take all sampling points The maximum value divided by the target current value Then multiply by 100%.

[0031] A charging pile health status assessment system based on multidimensional data-driven approach, the system comprising: The data acquisition module is deployed on the charging pile terminal equipment to collect raw indicator data in three dimensions: communication status, functional status, and operational status in real time. The data preprocessing and feature extraction module runs on the edge computing node and is used to clean the raw data, fill in missing values, remove outliers and standardize the data, and calculate the statistical characteristics of each indicator. The multidimensional assessment and dynamic compensation module is used to execute steps S2 to S4 of the above method and output the final health score. The information display module presents a detailed health analysis of a single charging pile device through a graphical interface, including at least a comprehensive score display area, a detailed communication status data area, a detailed functional status data area, a detailed operational status data area, a lifecycle data area, and a historical record area. The operation and maintenance early warning and decision support module is used to proactively generate early warning information based on health scores and trend changes, and recommend operation and maintenance action plans.

[0032] Furthermore, in the information display module, the comprehensive score display area displays the final health score and health level in the form of a progress ring chart and numbers, and simultaneously displays the independent scores of three dimensions: communication status, functional status, and operational status. The detailed data area of ​​each dimension allows for expansion to view the current values ​​of each indicator within that dimension, historical comparison curves, and automatically calculated weights of each indicator. The lifecycle data area displays the cumulative running time, cumulative charging volume, number of charging times, number of major failures, maintenance record summary, and remaining service life range predicted based on the health score since the charging pile was put into operation. The historical record area provides interactive time range selection, historical health score curves, evolution trends of scores in each dimension, fluctuation curves of key raw parameters, and detailed reports for each evaluation.

[0033] Compared with the prior art, the present invention has at least the following beneficial effects: First, this invention completely abandons the weighting method that relies on human experience. It adopts a weighting strategy that combines entropy weighting and CRITIC geometric average. Whether it is the internal indicators of each dimension or the secondary weights between the three dimensions, they are all automatically calculated and generated by the dispersion and correlation of the data itself. This eliminates the bias caused by subjective judgment in traditional fuzzy comprehensive evaluation or hierarchical analysis, making the evaluation results more objective, stable and adaptable to changes in different equipment types and operating environments.

[0034] Secondly, this invention breaks through the limitations of traditional single-dimensional monitoring by collecting and merging three dimensions—communication status, functional status, and operational status—in parallel and in a hierarchical manner. This allows for a comprehensive capture of the health status of charging piles, avoiding blind spots in evaluation caused by focusing only on electrical parameters while ignoring communication quality or functional module degradation. For example, a device with normal output electrical indicators but a significantly reduced card reader recognition success rate will be accurately reflected by the low score in the functional dimension in this invention, thus triggering timely maintenance.

[0035] This invention also introduces a degraded operation state detection mechanism. Through joint analysis of load rate and output capacity, it proactively identifies the critical changes in equipment transitioning from a healthy state to a degraded state, and dynamically corrects the operation score using a degradation coefficient. This enables it to capture the slow performance degradation process that traditional threshold alarms cannot detect. Building upon this, the invention further constructs trend health coefficients and fluctuation health coefficients. By linearly regressing historical health score sequences, it captures the long-term trend of equipment health. Simultaneously, it uses short-term coefficients of variation to measure the volatility of key indicators. These two dynamic coefficients are then exponentially weighted and integrated into the base score, ensuring that the final health score not only reflects the current state but also contains information on degradation trends and stability.

[0036] Furthermore, the cross-correction term designed in the three-channel hierarchical aggregation process of this invention uses the square of the lowest score among the three dimensions as the correction factor, which effectively highlights the impact of the weak link effect on the overall health, avoids the problem that the weak link may be masked by simple weighted averaging, and enables maintenance personnel to clearly locate the equipment dimension that needs the most intervention.

[0037] In summary, this invention completely transforms the charging pile operation and maintenance model from passive, reactive repair to proactive, pre-emptive early warning. Maintenance personnel can schedule maintenance plans in advance based on health scores, significantly reducing the number of sudden failures and emergency repairs. Practical application verification shows that charging piles using this invention have significantly improved mean time between failures (MTBF), and significantly reduced operation and maintenance costs and spare parts replacement costs. Simultaneously, the entire lifecycle management of the equipment gains scientific data support, thereby improving the reliability of charging infrastructure and user satisfaction while creating considerable economic benefits for operating companies. Attached Figure Description

[0038] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof.

[0039] Figure 1 This is a schematic diagram of the charging pile health status assessment method provided in an embodiment of the present invention; Figure 2This is a schematic diagram of the module composition of the charging pile health status assessment system provided in an embodiment of the present invention. Detailed Implementation

[0040] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, other embodiments obtained by those skilled in the art without creative effort are all within the scope of protection of the present invention.

[0041] Example 1 This embodiment provides a charging pile health status assessment method based on multi-dimensional data, using a DC fast charging pile as the assessment object, with an assessment cycle of once a day (i.e., generating a health score once a day), and the data collection time window being the most recent 24 hours. Figure 1 As shown, the implementation process is mainly divided into three stages: Phase 1: Collect communication status indicators, functional status indicators, and operational status indicators of charging piles. Standardize the raw indicators within each dimension, automatically calculate the weight of each indicator, and calculate the communication score, functional score, and operational score respectively. At the same time, introduce an operational degradation detection mechanism to correct the operational score.

[0042] 1. Data collection and scoring in the communication dimension (1) Raw data collection Data exchange between the charging station and the backend server is conducted via TCP / IP protocol. The following parameters are collected in real time: Device online time The total duration during which the charging pile maintains a heartbeat connection with the server within the statistical evaluation period; Total evaluation period duration In this embodiment, we take Seconds (24 hours); Heartbeat timeout count The number of times a heartbeat timeout occurred during the assessment period; Command issuance success count The number of times that control commands (such as start charging and stop charging) issued by the server were successfully executed by the charging pile within the evaluation period; Total number of instructions issued The total number of commands issued by the server during the evaluation period.

[0043] (2) Calculation of basic indicators Based on the above original parameters, calculate three basic indicators: Communication online rate This indicates the proportion of time that the charging station maintains a long-term connection with the server:

[0044] Packet loss rate A lower packet loss rate indicates better communication quality.

[0045] like Then take

[0046] Command response success rate :

[0047] like Then take .

[0048] (3) Standardization process Suppose there are m sampling points in total (for example, data from 7 consecutive days, then m = 7), and each sampling point corresponds to a set of data. Each indicator is standardized individually, mapping the original values ​​to the [0,1] interval.

[0049] Communication online rate Command response success rate Using the forward standardization formula:

[0050] in For the first i The sampling point at the th sampling point j The original values ​​of each indicator and These are the minimum and maximum values ​​of the indicator across all sampling points, respectively.

[0051] Packet loss rate Using the inverse standardization formula:

[0052] After standardization, it is obtained .

[0053] (4) Calculate weights using the entropy weight method First calculate the... j Information entropy of each indicator :

[0054] Then calculate the entropy weight method weights. :

[0055] Where n represents the number of indicators, which are three indicators: online communication rate, packet loss rate, and command response success rate.

[0056] (5) Calculate the weights using the CRITIC method The CRITIC method considers both the standard deviation of the indicators (reflecting the strength of the contrast) and the correlation coefficient between the indicators (reflecting the conflict). First, the first... j Standard deviation of each indicator :

[0057] Calculate the next j Correlation coefficient between one indicator and the other indicators .

[0058] Information content Defined as:

[0059] The weights for the CRITIC method are:

[0060] (6) Combination weighting The weights obtained from the entropy weight method and the CRITIC method are then geometrically averaged and fused.

[0061] (7) Communication dimension score Standardized data for the current evaluation period The original score for the communication dimension is:

[0062] The value ranges from [0,1], and the closer it is to 1, the better the communication status.

[0063] 2. Functional Dimension Data Collection and Scoring (1) Raw data collection The following four parameters are collected inside the charging station through a self-test program and sensors: Measurement module accuracy deviation Error value of the built-in power metering module in the charging pile; Display status : A binary variable, 1 for normal and 0 for fault; Card reader recognition success rate The percentage of times the card reader successfully recognized the charging card (or RFID tag) out of the total number of card swipes during the evaluation period; Validity of insulation testing of protection module The percentage of times the insulation test function was executed normally and the result was correct out of the total number of tests.

[0064] (2) Differentiation and standardization For accuracy deviation Nonlinear transformation is employed:

[0065] That is, when the deviation is ≤0.5%, the score is 1. For every doubling of the deviation, the score decreases by the reciprocal.

[0066] For display status Take directly Regarding the card reader's recognition success rate And the effectiveness of insulation testing of protection modules The positive standardization formula is adopted (same as the communication dimension).

[0067] (3) Weight calculation and functional scoring The same entropy weighting method and CRITIC method used in the communication dimension were employed for weighting (using standardized data as input) to calculate the weights of the four indicators, and then the weighted sum was used to obtain the original score for the functional dimension. .

[0068] 3. Operational Dimension Data Collection and Scoring (1) Raw data collection The following parameters (all average values ​​within the evaluation period) were obtained from the electrical parameter acquisition module of the charging pile: Input voltage error The absolute value of the relative error between the actual input voltage and the rated input voltage; Output voltage error The absolute value of the relative error between the actual output voltage and the target output voltage; Voltage regulation accuracy Output voltage stability is measured; a smaller value is better. Steady flow accuracy Output current stability is measured; a smaller value is better. Overall efficiency The ratio of output electrical energy to input electrical energy; Power factor PF: The ratio of active power to apparent power, with a value range of [0,1]. Output power (kW): Average power output; Rated power (kW): The nominal power indicated on the nameplate of the charging pile.

[0069] Among them, voltage regulation accuracy is used to measure the stability of the output voltage of the charging pile relative to the target voltage during the charging process. In specific calculations, it is necessary to measure the voltage when the charging pile is in constant voltage charging mode and set a fixed target voltage value. This value is determined by the charging strategy, for example, set to the rated charging voltage of the battery pack during the constant voltage charging phase. Within a single evaluation cycle (typically 24 hours), the instantaneous voltage value at the charging pile output is continuously acquired at a sampling frequency of no less than 1 Hz. A total of N There are 10 sampling points. The formula for calculating the voltage regulation accuracy is:

[0070] The numerator is the maximum absolute value of the deviation between the output voltage and the target voltage across all sampling points, the denominator is the target voltage value, and the result is expressed as a percentage.

[0071] The calculation method for constant current accuracy is similar to that for voltage regulation accuracy, but it applies to constant current charging mode and is used to measure the stability of the output current relative to the target current. A fixed target current value is set. During the constant current charging phase, the instantaneous current value at the output terminal of the charging pile is continuously collected at a frequency of not less than 1 Hz. N sampling points are obtained. The formula for calculating the steady-state accuracy is:

[0072] The numerator is the maximum absolute value of the deviation between the output current and the target current among all sampling points, the denominator is the target current value, and the result is also expressed as a percentage.

[0073] (2) Load rate calculation

[0074] (3) Historical average load factor Calculate the moving average of the load rate for each day within the most recent week (7 days):

[0075] (4) Detection of degraded operation status The judgment logic is as follows: If the conditions are met for three consecutive evaluation periods (i.e., three consecutive days)... and If the charging pile is in a degraded operating state, a degraded coefficient correction item will be set. (A value between 0.85 and 0.95 can be selected based on actual operation and maintenance experience). Otherwise .

[0076] (5) Original score of the operational dimension When a degraded running state is detected:

[0077] When no degradation status is detected:

[0078] in, This represents the larger of the input voltage error and the output voltage error, reflecting the most severe voltage error. The physical meaning of this formula is: efficiency is directly multiplied by the power factor as the baseline performance; the part in parentheses is the error penalty term, which decreases as the error increases; the last term... The load gain item is awarded when the equipment is under load for an extended period and the current load rate is higher. This reflects the value of the equipment under actual workload. Even if the indicators are perfect, the equipment is meaningless if it is not under load.

[0079] Phase Two: The communication score, function score, and operation score are aggregated in parallel through three channels in a hierarchical manner. The secondary weights of the three dimensions are calculated, and a cross-correction term is introduced to obtain the basic health score.

[0080] 1. Secondary weight calculation Communication scores from the past few evaluation periods (e.g., the last 30 days) Functional rating Operational rating As the input matrix, the same entropy weighting method + CRITIC method combined weighting algorithm as in Stage 1 is used to calculate the quadratic weights of the three dimensions. , , ,satisfy .

[0081] 2. Integration of weighted and cross-correction First, calculate the preliminary fusion score:

[0082] Recalculate the cross correction term :

[0083] The final basic health score is:

[0084] The purpose of this cross-correction term is to add a positive correction term to the initial fusion score when the scores of the three dimensions differ significantly. This is done by multiplying the square of the lowest score by the product of the three weights, thereby highlighting the weight of the weakness and preventing the weakness from being masked by the weighted average.

[0085] Phase 3: Extract historical health score sequences, calculate trend health coefficient and fluctuation health coefficient, dynamically compensate the basic health score, obtain the final health score, and classify health levels according to the score range.

[0086] 1. Trend Health Index Take the historical baseline health score sequence over the past T=7 days. ,in t This indicates the current evaluation period. For ease of calculation, time points are numbered as follows: The corresponding score is recorded as .

[0087] Calculate the slope k of the first-order linear regression: , ,

[0088] The trend health coefficient is defined as:

[0089] in, (The trend correction factor can be selected between 0.1 and 0.3 depending on the actual scenario.) (Trend decay index, with a value less than 1, typically ranging from 0.5 to 0.8); For symbolic functions: when When +1 is taken, When -1 is taken, Take 0 at that time.

[0090] 2. Fluctuation Health Index Two auxiliary indicators sensitive to operational stability were selected: voltage error. (Pick and (average) and command response success rate Collect the values ​​of these two indicators over the most recent 5 assessment periods (i.e., 5 days) and calculate their short-term coefficients of variation.

[0091] For voltage error sequence Calculate the mean and standard deviation : ,

[0092] The coefficient of variation is:

[0093] Similarly, for the command response success rate sequence ,calculate .

[0094] The fluctuation health coefficient is:

[0095] The fluctuation health coefficient ranges from [0,1], with more drastic fluctuations (larger coefficient of variation). The smaller it is, it can even be reduced to 0.

[0096] 3. Final health score Basic health score Trend Health Coefficient Fluctuation health coefficient Fusion:

[0097] The exponential adjustment coefficient satisfies In this embodiment, we take... , In other words, the trend health coefficient has a slightly higher weight than the fluctuation health coefficient, reflecting that long-term degradation trends have a more fundamental impact on equipment health than short-term fluctuations. In practice, this can be flexibly adjusted according to the operation and maintenance objectives.

[0098] 4. Health Level Classification This embodiment is based on the final health score. Charging stations are classified into the following five levels:

[0099] The following is a data segment of a specific charging pile over five consecutive evaluation periods to illustrate the calculation process.

[0100] Raw data (example from the communication dimension)

[0101] The weights were calculated using the entropy weight method combined with the CRITIC method: online rate 0.42, packet loss rate 0.35, and response success rate 0.23. The standardized values ​​for period 5 (assuming historical extreme values) were 0.85, 0.70, and 0.80, respectively. Therefore, the communication score is: 0.42×0.85+0.35×0.70+0.23×0.80=0.771 Similarly, the functional score and operational score are calculated, assuming the final result is: , , Using the historical 30-day rating sequence to calculate the quadratic weights, we obtain: , , The basic health score is as follows:

[0102]

[0103]

[0104] 7 days of history sequence fitting slope ,but:

[0105] The coefficient of variation for voltage error over the past 5 days is 0.12, and the coefficient of variation for response success rate is 0.08. Therefore:

[0106] Pick , ,but:

[0107] The final health score is 0.752, which is considered "good". A routine check-up once a month is recommended.

[0108] To verify the effectiveness of the method in this embodiment, a comparative test was conducted over a period of 6 months on 50 charging piles operated by a certain operator. 25 of these piles used the traditional threshold alarm method (control group), while the remaining 25 used the method of this invention (experimental group). The results showed that the experimental group detected 17 potential faults in advance (including excessive metering deviation, insulation detection function degradation, and frequent packet loss), with an average early warning time of 8.5 days. The control group could only detect serious faults that had already led to charging failure. The mean time between failures (MTBF) of the charging piles in the experimental group increased by 32%, and maintenance costs (including manual inspection, emergency repairs, and spare parts replacement) decreased by 27%. The Pearson correlation coefficient between the health score and the actual remaining lifespan of the equipment in the experimental group reached 0.86, indicating that the score has good predictive ability.

[0109] Example 2 Based on the above method, this embodiment provides a charging pile health status assessment system. This system adopts a three-tier architecture (edge-cloud) and includes a data acquisition module, a data preprocessing and feature extraction module, a multi-dimensional assessment and dynamic compensation module, an information display module, and an operation and maintenance early warning and decision support module. Figure 2 As shown, the various modules in the system interact with each other through message queues and API interfaces to jointly achieve real-time monitoring, quantitative evaluation, and visualization of the health status of charging piles.

[0110] The data acquisition module is deployed on the charging pile terminal equipment, acquiring three types of raw data in real time: communication status, functional status, and operational status through an embedded intelligent acquisition unit. Regarding communication status, the module collects parameters such as device online time, heartbeat packet timeouts, successful command issuance, and total command issuance. For functional status, it collects the metering module's accuracy deviation, display status, card reader recognition success rate, and the effectiveness of insulation detection in the protection module. For operational status, it collects input voltage error, output voltage error, voltage regulation accuracy, current regulation accuracy, overall efficiency, power factor, output power, and rated power. All data is packaged and timestamped according to a preset frequency (key electrical parameters are collected once per second, and status parameters are collected once per minute), and sent to the edge computing node via an encrypted transmission protocol.

[0111] The data preprocessing and feature extraction module runs on the local edge computing gateway of the charging station, responsible for cleaning, imputing missing values, removing outliers, and standardizing the raw data. Specifically, this module identifies and filters out obviously unreasonable values ​​caused by momentary sensor failures or communication interference (e.g., sampling points where voltage fluctuations exceed 50% of the rated value), imputes short-term missing data using linear interpolation, and then maps each indicator to the range of 0 to 1 according to the standardization formula in this invention. Simultaneously, this module calculates the basic statistical characteristics for each evaluation period (default 24 hours), including the mean, maximum value, standard deviation, and coefficient of variation of each indicator, preparing the input data matrix for subsequent entropy weighting and CRITIC weighting calculations.

[0112] The multidimensional assessment and dynamic compensation module is the core computing engine of the system and can be deployed on a cloud platform or an edge server with strong computing power. This module first calls the standardized indicator sequence output by the data preprocessing module. Following the combined weighting process in the first stage of this invention, it automatically calculates the geometric mean weights of the entropy weight method and the CRITIC method for the online rate, packet loss rate, and response success rate of the communication dimension, and generates a communication score accordingly. The same combined weighting operation is performed on the measurement deviation, display status, card reader success rate, and insulation detection effectiveness of the functional dimension to obtain a functional score. For the operational dimension, it first calculates the load rate and the historical average load rate to detect whether the system is in a degraded operational state, and then substitutes these values ​​into the operational scoring formula to obtain the operational score. After completing the independent scoring of the three dimensions, the module enters the second stage, using these scores as input, and again applies the entropy weight method and the CRITIC method to calculate the secondary weights of the three dimensions of communication, functionality, and operation, and obtains a basic health score by introducing a cross-correction term. The third stage then begins, where the module extracts the basic health score sequence from the past 7 days, obtains the trend slope through linear regression fitting, and calculates the trend health coefficient. Simultaneously, it extracts the voltage error and command response success rate sequences from the past 5 days, calculates their respective short-term coefficients of variation, and derives the fluctuation health coefficient. Finally, it integrates the basic health score with the two dynamic coefficients in an exponentially weighted manner, outputs the final health score, and automatically determines the health level of the charging pile based on a preset threshold range.

[0113] The information display module serves as the front-end interface for operations and maintenance personnel to interact with the system, offered in both web dashboard and mobile application formats. Its core function is to display detailed health analysis of individual charging pile devices. The interface layout is designed according to a logic from overall to specific. When operations and maintenance personnel click on any charging pile number from the device list, the page first displays a comprehensive score card for that device, presenting the current final health score and corresponding health level with prominent numbers and a large progress ring chart. Simultaneously, it displays independent scores for three dimensions: communication status, functional status, and operational status. Each dimension is represented by a horizontal progress bar and specific numerical values, differentiated by color (green for excellent, yellow for acceptable, orange for requiring attention, and red for serious). Clicking on the progress bar of any dimension expands the detailed data panel for that dimension. For example, expanding the communication status panel displays the current values ​​and historical comparison curves for online rate, packet loss rate, and response success rate. It also lists the weight values ​​of each indicator within that dimension, automatically calculated using the entropy weight method and the CRITIC method, allowing operations and maintenance personnel to understand which indicators have the greatest impact on the current score. The function status panel displays specific values ​​for metering deviation, display status, card reader recognition success rate, and insulation detection effectiveness, as well as reference values ​​for normal ranges for each item. If any indicator deviates from the normal range, the system will mark it with a warning icon and provide a brief explanation. In addition to displaying key electrical parameters such as overall efficiency, power factor, voltage error, voltage regulation accuracy, current regulation accuracy, and load rate, the operation status panel also specifically indicates whether the system is in a degraded operation state. If in a degraded state, it displays the current value of the degraded coefficient and the number of days it has lasted.

[0114] In addition to detailed data across the three dimensions, the information display module also features dedicated lifecycle data and historical data areas. The lifecycle data area summarizes key statistics about the charging station from its initial commissioning to the present, including cumulative runtime, cumulative charging volume, number of completed charging sessions, number of major malfunctions, maintenance record summaries, and the remaining service life range predicted based on the health score (e.g., "estimated remaining healthy operating time is 8 to 12 months"). This information is presented in the form of a timeline and indicator cards, helping maintenance personnel assess the depreciation status and replacement timing of the equipment from a full lifecycle perspective. The historical data area provides interactive charts, allowing users to select a time range (last week, month, three months, or a custom interval) to view the historical curve of the charging station's health score, the evolution trend of the three-dimensional scores, and the historical fluctuations of key raw parameters (such as efficiency, packet loss rate, and voltage error). The charts support zooming and hovering to display specific values, and any two indicators can be overlaid on the same coordinate system for comparative analysis; for example, the health score and voltage regulation accuracy curve can be compared side-by-side to observe their correlation. In addition, the history section also lists detailed reports for each health assessment in tabular form, including the start and end dates of the assessment period, the final score, the scores for each dimension, the dynamic coefficient values ​​used, and a brief diagnostic opinion automatically generated by the system (such as "The packet loss rate has increased for three consecutive days, and it is recommended to check the network connection").

[0115] The operation and maintenance early warning and decision support module works closely with the information display module to proactively generate early warning information based on health scores and trend changes. When the health score of a charging pile drops to the qualified level or below, or the trend health coefficient is less than 0.95 for two consecutive periods (i.e., a significant decline in health), this module will send an early warning notification to relevant operation and maintenance personnel through pop-ups in the information display module, in-site messages, emails, or WeChat robots. The notification includes the device number, current location, current score, health level, and a preliminary assessment of possible causes (e.g., "The degraded operation status has lasted for 3 days, and the overall efficiency has dropped from 92% to 87%"). Simultaneously, this module will recommend reasonable operation and maintenance actions based on historical data and the current score. For example, for a charging pile with a score of 0.65, the system will suggest "arrange maintenance within two weeks, focusing on checking the metering module and insulation detection circuit"; for charging piles with a score below 0.3, an emergency shutdown command will be generated, and the information display module will highlight "Immediate shutdown for maintenance" with a red highlight box.

[0116] Through the collaborative work of the above functional modules, this system not only achieves an objective quantitative assessment of the health status of charging piles, but also presents detailed data on communication, function, and operation in a clear and easy-to-understand visual way to maintenance personnel. At the same time, it provides a panoramic view of lifecycle data and historical records, truly realizing the transformation from passive post-event maintenance to proactive pre-event early warning maintenance mode, significantly improving the reliability and maintenance efficiency of charging pile equipment.

[0117] The above system can execute the charging pile health status assessment method described in Embodiment 1, and has the corresponding functional modules and beneficial effects of the method. For technical details not described in detail in this embodiment, please refer to the corresponding description in Embodiment 1 of this invention.

[0118] Those skilled in the art should understand that the above parameters (such as period, weight, threshold, coefficient, etc.) are all exemplary descriptions. In actual applications, they can be flexibly adjusted according to the equipment model, operating environment and historical fault data, but these adjustments do not depart from the protection scope of this invention.

[0119] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented using software plus a general-purpose hardware platform, or of course, using hardware. Based on this understanding, the above technical solutions, in essence or the parts that contribute to the related technology, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0120] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; under the concept of the present invention, the technical features of the above embodiments or different embodiments can also be combined, the steps can be implemented in any order, and there are many other variations of different aspects of the present invention as described above, which are not provided in detail for the sake of brevity; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

Claims

1. A method for assessing the health status of charging piles based on multi-dimensional data-driven approaches, characterized in that, Including the following steps: S1. Collect raw indicators of the communication status, functional status and operational status of the charging pile during the evaluation period. S2. Standardize the raw indicators in the three dimensions of communication status, functional status and operation status respectively, and automatically calculate the weight of the indicators in each dimension. The weighted sum is used to obtain the raw scores of communication dimension, functional dimension and operation dimension. S3. Using the original scores of the communication dimension, the function dimension, and the operation dimension as input, calculate the secondary weights of the three dimensions again, obtain the preliminary fusion score by weighted summation, and introduce a cross correction term to obtain the basic health score. S4. Extract the historical baseline health score sequence and calculate the trend health coefficient; Simultaneously, short-term data of auxiliary indicators are collected to calculate the volatility health coefficient; The basic health score is weighted and integrated with the trend health coefficient and the fluctuation health coefficient to output the final health score.

2. The method for assessing the health status of charging piles as described in claim 1, characterized in that, In step S2: The weights of the internal indicators of each dimension are calculated using a combination of entropy weighting and CRITIC geometric mean weighting strategies. First, calculate the information entropy of the standardized indicator data: in, Indicates the first j Information entropy of each indicator m The number of sampling points. For the first i The sampling point at the th sampling point j Standardized values ​​for each indicator; Then calculate the entropy weight method weights: in, Indicates the first j Entropy weighting of each indicator n The total number of indicators within the dimension; Next, calculate the first... j Standard deviation of each indicator: in, Indicates the first j The standard deviation of each indicator For the first time during the evaluation period j The average of the indicators; Calculate the next j Correlation coefficient between each indicator and other indicators within the dimension ; Then calculate the information content: in, Indicates the first j The amount of information in each indicator; Finally, calculate the CRITIC method weights: in, Indicates the first j CRITIC weighting of each indicator; The entropy weight method weights and the CRITIC method weights are then geometrically averaged and fused. in, Indicates the first j The combined weights of each indicator; For each dimension, the original score for each dimension is obtained by weighted summation based on the combined weights of each indicator within the dimension.

3. The method for assessing the health status of charging piles as described in claim 1, characterized in that, In step S3: The past M The original scores of the communication dimension, the original scores of the function dimension, and the original scores of the operation dimension within each evaluation period are used as the input matrix. The score sequence of each dimension is regarded as an indicator. The entropy weight method weight and the CRITIC method weight are calculated respectively, and then the geometric mean is fused to obtain the secondary weight of each dimension. The entropy weight method calculates the weights as follows: for the th j First, calculate the information entropy in each dimension: in, For the first i Within the evaluation period, the first j The original scores for each dimension; Then calculate the entropy weight method's second-order weights: ; The CRITIC method calculates the weights as follows: first calculate the weight of the first weight... j Standard deviation of the score series for each dimension: Calculate the next j Correlation coefficients between one dimension and the other dimensions ; Then calculate the information content: ; Obtain the CRITIC method's quadratic weights: ; By fusing the entropy weight method's quadratic weights and the CRITIC method's quadratic weights, we obtain the quadratic weights for each dimension: The preliminary fusion score is as follows: in, , and The original scores are for the communication dimension, the functionality dimension, and the operation dimension, respectively. , and These are the corresponding secondary weights; Then, a cross-correction term is introduced: The basic health score is then: .

4. The method for assessing the health status of charging piles as described in claim 1, characterized in that, The trend health coefficient mentioned in step S4 is calculated as follows: Extract the past T The historical baseline health score sequence for each assessment period is denoted as . ,in, t Indicates the current evaluation period. T The preset number of historical backtracking cycles; the time points are numbered as follows. The corresponding basic health score is recorded as Calculate the slope of the first-order linear regression: ， ， in, The slope of the first-order linear regression. It is the average value at any given time point. The mean of the basic health score; The trend health coefficient is defined as: in, The preset trend correction magnitude coefficient is greater than 0. The preset trend decay index is greater than 0 and less than 1; For symbolic functions: when When +1 is taken, When -1 is taken, Take 0 at that time.

5. The method for assessing the health status of charging piles as described in claim 1, characterized in that, The calculation method for the fluctuation health coefficient mentioned in step S4 is as follows: Several auxiliary indicators sensitive to operational stability were selected, and the numerical sequences of each auxiliary indicator were collected over the most recent L evaluation periods; for the nth auxiliary indicator, its numerical sequence is denoted as... Calculate the mean of the sequence. and standard deviation This leads to the coefficient of variation of the auxiliary indicator. ; Let N be the total number of auxiliary indicators selected, then the fluctuation health coefficient is: Fluctuation Health Index The value range is [0,1], and the more drastic the fluctuation, the smaller the value.

6. The method for assessing the health status of charging piles as described in claim 1, characterized in that, The final health score in step S4 is calculated as follows: in, For the final health score, Basic health score, As the trend health coefficient, The fluctuation health coefficient; and The preset exponential adjustment coefficient and satisfying .

7. The method for assessing the health status of charging piles as described in claim 1, characterized in that, The original score for the operational dimension is corrected by introducing a degradation coefficient based on the operational degradation status detection results: If the load rate of a charging pile exceeds a preset load rate threshold for multiple consecutive evaluation periods, and the actual output power is lower than a preset percentage of the rated power, it is determined to be in a degraded operation state, and a degrade factor is set. The value must be less than 1 within the preset range; otherwise... The original score for the operational dimension is corrected under the following formula when the rating is downgraded: in, The overall efficiency of a charging pile is equal to the ratio of output electrical energy to input electrical energy. The power factor of a charging pile is equal to the ratio of active power to apparent power. The input voltage error of the charging pile is equal to the absolute value of the relative error between the actual input voltage and the rated input voltage. The output voltage error of the charging pile is equal to the absolute value of the relative error between the actual output voltage and the target output voltage. For voltage regulation accuracy, For stable flow accuracy, This is the historical average load rate. This represents the current load rate.

8. The method for assessing the health status of charging piles as described in claim 7, characterized in that: The voltage regulation accuracy The calculation method is as follows: In constant voltage charging mode, the instantaneous voltage value at the output terminal is continuously collected at a frequency of not less than 1 Hz. Take all sampling points The maximum value divided by the target voltage value Then multiply by 100%; The current stabilization accuracy The calculation method is as follows: In constant current charging mode, the instantaneous current value at the output terminal is continuously collected at a frequency of not less than 1 Hz. Take all sampling points The maximum value divided by the target current value Then multiply by 100%.

9. A charging pile health status assessment system based on multi-dimensional data-driven approach, characterized in that, include: The data acquisition module is deployed on the charging pile terminal equipment to collect raw indicator data in three dimensions: communication status, functional status, and operational status in real time. The data preprocessing and feature extraction module runs on the edge computing node and is used to clean the raw data, fill in missing values, remove outliers and standardize the data, and calculate the statistical characteristics of each indicator. A multidimensional assessment and dynamic compensation module is used to perform steps S2 to S4 of the method described in any one of claims 1 to 8 and output the final health score. The information display module presents a detailed health analysis of a single charging pile device through a graphical interface, including at least a comprehensive score display area, a detailed communication status data area, a detailed functional status data area, a detailed operational status data area, a lifecycle data area, and a historical record area. The operation and maintenance early warning and decision support module is used to proactively generate early warning information based on health scores and trend changes, and recommend operation and maintenance action plans.

10. The charging pile health status assessment system as described in claim 9, characterized in that, In the information display module, the comprehensive score display area presents the final health score and health level in the form of a progress ring chart and numbers, and simultaneously displays independent scores for three dimensions: communication status, functional status, and operational status. The detailed data area for each dimension allows users to expand and view the current values, historical comparison curves, and automatically calculated weights of each indicator within that dimension. The lifecycle data area displays the cumulative running time, cumulative charging volume, number of charging times, number of major failures, maintenance record summary, and remaining service life range predicted based on the health score since the charging pile was put into operation. The historical record area provides interactive time range selection, historical health score curves, evolution trends of scores for each dimension, fluctuation curves of key raw parameters, and detailed reports for each assessment.

Images