Electric power business environment index evaluation method based on artificial intelligence technology

Abstract

The invention relates to an electric power business environment index evaluation method based on an artificial intelligence technology. The method comprises the following steps: firstly, acquiring power supply related data corresponding to an electric power business environment in each electric power region from an electric power system; performing data cleaning on the power supply related data, and extracting data related to an electric power business environment in a classified manner; calculating scores of various indexes in combination with the data set obtained through cleaning; and finally, according to the score of each index and the corresponding weight relationship, determining the comprehensive score of the electric power environment of each electric power region. By constructing a quantitative index system and introducing an artificial intelligence technology, objective evaluation and accurate optimization of an electric power business environment are realized. Through dynamic weight calculation and real-time data processing, the method can adapt to dynamic changes of a power system, and prospective optimization is realized through a prediction model. The method has a wide application prospect, and the optimization efficiency and effect of the power business environment can be remarkably improved.

Description

Technical Field

[0001] A data processing method specifically used for administrative supervision, specifically involving an electric power business environment indicator evaluation method based on artificial intelligence technology. Background Art

[0002] Generally speaking, the business environment refers to the sum of various external environments throughout a company's lifecycle, from inception to deregistration. It is a systematic project encompassing numerous areas of economic and social reform and opening up to the outside world. The assessment of the electricity business environment involves power grid companies' electricity procedures, connection timelines, connection costs, power supply reliability, and other power supply services. Currently, assessments of the electricity business environment primarily rely on manual experience summarization and qualitative analysis. This approach has the following shortcomings in practical applications:

[0003] High subjectivity: A large amount of power supply-related data has accumulated in the power system, but this data is highly dependent on the experience and judgment of the evaluators, making it difficult to fully cover all aspects of the power business environment. It has not been fully explored and utilized, and it is easy to miss key issues, resulting in highly subjective and low-precision evaluation results.

[0004] Single evaluation indicator: People's experience and judgment often focus on a single indicator (such as electricity cost or power quality), and the evaluation indicators of the electricity business environment are mostly qualitative descriptions, lacking quantitative standards and comprehensive evaluation of multi-dimensional indicators, making it difficult to compare and verify the evaluation results.

[0005] Low optimization efficiency: The manual summary and analysis process is time-consuming and labor-intensive, and it is difficult to accurately locate problems in the power business environment and respond quickly to changes in the power business environment, resulting in difficulty in ensuring the optimization execution efficiency and accuracy of the power business environment.

[0006] Lack of dynamic feedback mechanism: Existing methods are mostly post-analysis, lacking the ability to predict and proactively optimize the electricity business environment, making it difficult to promptly identify problems and shortcomings in business links. Summary of the Invention

[0007] This invention proposes a method for evaluating power business environment indicators based on artificial intelligence technology. The specific technical solution is as follows:

[0008] A method for evaluating power business environment indicators based on artificial intelligence technology includes the following steps:

[0009] S100: Obtain power supply data related to the power business environment from the power system; specifically including:

[0010] Electricity consumption data: electricity consumption, electricity consumption time, electricity cost, and electricity consumption time distribution;

[0011] Power load data: voltage stability, peak load, average load, and number of faults;

[0012] Service quality data: service response time, customer satisfaction, and customer complaint rate;

[0013] Environmental data: carbon emissions, proportion of renewable energy use;

[0014] Fault records: number of faults, repair time;

[0015] S200: Cleaning the collected raw data; including:

[0016] S210: Desensitize sensitive data in the original data to protect user privacy;

[0017] S220: Remove duplicate values ​​and outliers from the data and fill in missing values;

[0018] S230: Perform data standardization processing on the data, normalizing it to the interval [0,1] to ensure data quality;

[0019] S300: Auxiliary classification of business environment indicators into the following three categories: comprehensive electricity consumption indicators, power quality indicators and service quality indicators, and power diagnosis and repair indicators;

[0020] The comprehensive electricity consumption index includes: electricity efficiency index, which is used to characterize the economic benefits of unit electricity consumption; electricity cost index, which is used to characterize the cost of unit electricity consumption;

[0021] Power quality indicators include: power load index, which represents the ratio of peak load to average load; power failure index, which represents the product of the average number of failures per year and the repair time; carbon emission index, which represents the carbon emissions per unit of electricity consumption;

[0022] The service quality indicators include: power supply strategy indicator, which represents the flexibility score of the power supply strategy;

[0023] Power diagnosis and repair indicators include: a comprehensive score of fault diagnosis accuracy and repair efficiency;

[0024] S400: Extracting the corresponding key features for each type of indicator; specifically including:

[0025] S410: Calculate the correlation between features and major indicator variables, and select features that are highly correlated with major indicator variables;

[0026] S420: then using recursive feature elimination (RFE) to calculate the features that contribute the least to the comprehensive score of the power business environment until the required number of features is reached;

[0027] S430: Score each feature. If there is a major fault, deduct points accordingly to obtain a score for each feature.

[0028] S500: The entropy weight method is used to calculate the weight of each indicator; the calculation formula is:

[0029]

[0030] Among them, w j is the weight of the jth indicator, D j is the entropy value, f ij is the normalized index value, k is the number of key features mentioned above;

[0031] S600: Calculate the comprehensive score of the power business environment based on the scores and weights of each indicator; the comprehensive scoring model is:

[0032]

[0033] Among them, S j is the score of the jth indicator, w j is the weight of the j-th indicator.

[0034] Preferably, after S600, the method further includes:

[0035] S700: After extracting the above features and scores, a time series model is constructed through a long short-term memory network (LSTM) for time series prediction and real-time monitoring of indicator changes. The trained LSTM model is used to predict real-time data to obtain a predicted score. The predicted score is compared with the actual score, and a threshold is set. When the difference exceeds the threshold, the abnormal situation is handled in a timely manner.

[0036] Preferably, the number of LSTM layers of the LSTM model is usually set to 1 to 3 layers, and the fully connected layer: maps the output of the LSTM layer to the output space for predicting the score;

[0037] The model training selects the mean square error loss MSE as the loss function to measure the difference between the predicted value and the true value; multiple iterations of training are performed, and the loss is calculated and the model parameters are updated in each iteration; finally, the test set is used to evaluate the performance of the model and calculate the mean square error MSE and mean absolute error MAE indicators.

[0038] Preferably, the S210 includes the following process:

[0039] S211: Identify sensitive data, clarify the databases, tables, and fields where sensitive data such as user names, ID numbers, electricity addresses, and contact information are located, and classify and grade the sensitive data;

[0040] S212: Then, according to the data type and business scenario, select the appropriate desensitization algorithm and parameters, and configure specific desensitization rules based on the desensitization algorithm;

[0041] S213: Desensitize the sensitive data according to the configured desensitization rules.

[0042] Preferably, the S220 includes the following process:

[0043] S221: Use an exact matching algorithm to compare the various fields of the data records to find identical duplicate records; use a fuzzy matching algorithm, such as a basic nearest neighbor sorting algorithm, to determine the similarity by detecting records within a window and identify data with a similarity of more than 90%; after determining duplicate records, retain one record according to business rules and delete other duplicate records.

[0044] S222: Use the boxplot method to determine the range of outliers by calculating the quartiles of the data; use the Z-Score method within this range to calculate the distance between the data point and the mean, measured in units of standard deviation, and using the 3σ principle. Values ​​exceeding the mean ± 3 times the standard deviation are considered outliers; outliers are directly deleted or retained but marked;

[0045] S223: Fill in missing values ​​in the data, since the data is divided into different types, including numerical data, categorical data, and time series data.

[0046] Preferably, the S230 includes the following process:

[0047] Normalize the data to the interval [0,1], where the formula for data normalization is as follows:

[0048]

[0049] Among them, x min and x max are the minimum and maximum values ​​of the data, respectively.

[0050] Preferably, in said S400,

[0051] For the electricity efficiency index, the key features are extracted from the electricity consumption growth rate and peak-to-valley ratio;

[0052] For power quality indicators, key features are extracted, such as voltage fluctuation rate and fault repair time;

[0053] For service quality indicators, key features are extracted, such as customer complaint rate and service response time;

[0054] The electricity consumption growth rate is the rate of change of electricity consumption in adjacent time periods. The formula is: electricity consumption growth rate = (current period electricity consumption - previous period electricity consumption) / previous period electricity consumption × 100%;

[0055] Peak-valley ratio is the ratio of electricity consumption during peak hours to electricity consumption during off-peak hours. The formula is: Peak-valley ratio = electricity consumption during peak hours / electricity consumption during off-peak hours.

[0056] The calculation formula for voltage fluctuation rate is:

[0057]

[0058] Among them, V i is the voltage value at the i-th time point, V avg is the average voltage.

[0059] Compared with the prior art, the present invention has the following advantages:

[0060] Systematic assessment: Through a multi-dimensional indicator system and artificial intelligence algorithms, a comprehensive and dynamic assessment of the power business environment is achieved.

[0061] High precision and real-time performance: Use machine learning models to analyze data in real time to improve assessment accuracy and response speed.

[0062] Targeted optimization: Identify problem areas through comprehensive scoring, and provide a clear direction for optimizing the electricity business environment.

[0063] Low carbon and environmental protection: Introduce carbon emission indicators to promote the development of green power services.

[0064] In a pilot application in a certain region, the method improved the comprehensive score of the electricity business environment by 15%, shortened the fault repair time by 30%, and reduced the customer complaint rate by 20%. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Figure 1 It is a flow chart of an electric power business environment index evaluation method based on artificial intelligence technology of the present invention. DETAILED DESCRIPTION

[0066] A method for evaluating power business environment indicators based on artificial intelligence technology includes the following steps:

[0067] S100: Acquire power supply data related to the power business environment from the power system, such as from the power company's database, smart meters, environmental monitoring equipment, and other systems. Specifically, this includes:

[0068] Electricity consumption data: electricity consumption, electricity consumption time, electricity cost, and electricity consumption time distribution;

[0069] Power load data: voltage stability, peak load, average load, and number of faults;

[0070] Service quality data: service response time, customer satisfaction, and customer complaint rate;

[0071] Environmental data: carbon emissions, proportion of renewable energy use;

[0072] Fault records: number of faults, repair time;

[0073] During the data collection process, smart meters and sensors collect power system data automatically in real time or on an hourly, daily, or weekly basis, depending on the indicator requirements. Data that cannot be collected automatically can be supplemented through manual entry.

[0074] S200: Cleaning the collected raw data; including:

[0075] S210: Desensitize sensitive data in the original data to protect user privacy. This process includes the following:

[0076] S211: Identify sensitive data, clarify the databases, tables, and fields where sensitive data such as user names, ID numbers, electricity addresses, and contact information are located, and classify and grade the sensitive data;

[0077] S212: Then, according to the data type and business scenario, select the appropriate desensitization algorithm and parameters, and configure specific desensitization rules according to the desensitization algorithm; specifically,

[0078] For ID card numbers, a masking algorithm is used to retain the first 10 digits, and the remaining digits are replaced with "X" or "*";

[0079] For names, replace them with randomly generated pseudonyms;

[0080] For electricity addresses, a replacement algorithm is used to replace the real address with fictitious address information;

[0081] For electricity consumption data, a random offset is added or subtracted through the offset algorithm to make the data vary within a certain range while maintaining the relative relationship of the data;

[0082] Contact data is encrypted using a symmetric encryption algorithm to ensure data security during transmission and storage;

[0083] S213: Desensitize sensitive data according to the configured desensitization rules. The desensitized data can maintain the original data characteristics, business rules, and data relevance, ensuring that it will not be affected in subsequent data analysis.

[0084] S220: Remove duplicate values ​​and outliers from the data and use mean / median / interpolation to fill in missing values; including the following process:

[0085] S221: Using an exact matching algorithm, compare each field of the data record to find identical duplicate records; using a fuzzy matching algorithm, such as a basic nearest neighbor sorting algorithm, to determine similarity by detecting records within a window, and identify data with a similarity of more than 90%; after determining duplicate records, retain one record and delete the other duplicate records according to business rules;

[0086] S222: Use the boxplot method to determine the range of outliers by calculating the quartiles of the data; use the Z-Score method within this range to calculate the distance between the data point and the mean, measured in units of standard deviation, and using the 3σ principle. Values ​​exceeding the mean ± 3 times the standard deviation are considered outliers; outliers are directly deleted or retained but marked;

[0087] S223: Fill in the missing values ​​in the data. Since the data is divided into different types, including numerical data, categorical data, and time series data; specifically,

[0088] For numerical data, the mean or median is used to fill missing values. This is because the mean is suitable for cases where the data distribution is relatively uniform, while the median is suitable for cases where the data distribution is skewed or has extreme values. For categorical data, the mode is used to fill missing values. For time series data, linear interpolation can be used to estimate missing values. When the missing rate in the data exceeds 30%, the data point is directly removed.

[0089] S230: Perform data standardization to the range [0,1] to ensure data quality. This includes the following steps:

[0090] After data cleaning is completed, in order to make the data more consistent and comparable, it is necessary to standardize the data, convert it into a unified scale and range, and normalize the data to the [0,1] interval. The formula for data standardization is as follows:

[0091]

[0092] Among them, x min and x max are the minimum and maximum values ​​of the data respectively;

[0093] S300: After obtaining the above pre-processed data, since power supply data is multi-dimensional data, auxiliary classification of business environment indicators is performed, and they are divided into the following three categories of indicators: comprehensive power consumption indicators, power quality indicators and service quality indicators, and power diagnosis and repair indicators;

[0094] The comprehensive electricity consumption index includes: electricity efficiency index, which is used to characterize the economic benefits of unit electricity consumption; electricity cost index, which is used to characterize the cost of unit electricity consumption;

[0095] Power quality indicators include: power load index, which represents the ratio of peak load to average load; power failure index, which represents the product of the average number of failures per year and the repair time; carbon emission index, which represents the carbon emissions per unit of electricity consumption;

[0096] The service quality indicators include: power supply strategy indicator, which represents the flexibility score of the power supply strategy;

[0097] Power diagnosis and repair indicators include: a comprehensive score of fault diagnosis accuracy and repair efficiency;

[0098] Indicator classification can be adjusted according to business needs and data characteristics to ensure the scientificity and rationality of the classification standards;

[0099] S400: Extract the corresponding key features for each type of indicator to reduce the dimension of the data. Specifically, it includes:

[0100] S410: Calculate the correlation between features and major indicator variables, and select features that are highly correlated with major indicator variables;

[0101] S420: then using recursive feature elimination (RFE) to calculate the features that contribute the least to the comprehensive score of the power business environment until the required number of features is reached;

[0102] S430: Score each feature. If there is a major fault, deduct points accordingly to obtain a score for each feature.

[0103] For each type of indicator, corresponding key features are extracted. For example, for electricity efficiency indicators, key feature data such as electricity consumption growth rate and peak-to-valley ratio can be extracted; for power quality indicators, key feature data such as voltage fluctuation rate and fault repair time can be extracted; for service quality indicators, key feature data such as customer complaint rate and service response time can be extracted;

[0104] The electricity consumption growth rate is the rate of change of electricity consumption in adjacent time periods. The formula is: electricity consumption growth rate = (current period electricity consumption - previous period electricity consumption) / previous period electricity consumption × 100%;

[0105] Peak-valley ratio is the ratio of electricity consumption during peak hours to electricity consumption during off-peak hours. The formula is: Peak-valley ratio = electricity consumption during peak hours / electricity consumption during off-peak hours.

[0106] The calculation formula for voltage fluctuation rate is:

[0107]

[0108] Among them, V i is the voltage value at the i-th time point, V avg is the average voltage;

[0109] For the growth rate of electricity consumption, a 30-day or 90-day moving average can be used;

[0110] For fault repair time, the time interval from fault occurrence to repair completion can be calculated, and statistics such as average and median can be calculated. If it exceeds 24 hours, it can be considered a major fault;

[0111] For the customer complaint rate, it is calculated as the ratio of the number of customer complaints to the total number of service times. The formula is: customer complaint rate = number of complaints / total number of service times × 100%;

[0112] S500: The entropy weight method is used to calculate the weight of each indicator; the calculation formula is:

[0113]

[0114] Among them, w j is the weight of the jth indicator, D j is the entropy value, f ij is the normalized index value, k is the number of key features mentioned above;

[0115] S600: Calculate the comprehensive score of the power business environment based on the scores and weights of each indicator; the comprehensive scoring model is:

[0116]

[0117] Among them, S j is the score of the jth indicator, w j is the weight of the j-th indicator;

[0118] The comprehensive score range can be set from 0 to 100 points, with scores of 85 or above being excellent, 60 to 85 being good, and scores below 60 being in need of improvement. The higher the score, the better the business environment.

[0119] S700: After extracting the above features and scores, a long short-term memory (LSTM) network is used to construct a time series model for time series prediction and real-time monitoring of indicator changes. The trained LSTM model is used to predict real-time data to obtain a predicted score. The difference between the predicted score and the actual score is compared, and a threshold is set. When the difference exceeds the threshold, abnormal situations are promptly handled. By constructing an LSTM-based time series model, it is used to monitor changes in indicators related to the power business environment in real time, providing data support and decision-making basis for optimizing the power business environment.

[0120] The number of LSTM layers in the LSTM model is usually set to 1 to 3. The more layers, the stronger the nonlinear expression ability of the model, but it may also lead to overfitting. Fully connected layer: maps the output of the LSTM layer to the output space for prediction score; output size: determined by the prediction score. For example, if the value of the next time step is predicted, the output size is 1;

[0121] The model training uses the mean squared error (MSE) as the loss function to measure the difference between the predicted value and the true value. Multiple iterations of training are performed, and the loss is calculated and the model parameters are updated in each iteration. Finally, the test set is used to evaluate the performance of the model and calculate the mean squared error (MSE) and mean absolute error (MAE).

[0122] S800: Generates targeted optimization recommendations based on the comprehensive score and the performance of each indicator. If the electricity cost indicator is low, it recommends optimizing the electricity price strategy or promoting energy-saving equipment. If the power failure indicator is high, it recommends strengthening grid maintenance or introducing an intelligent fault prediction system.

Claims

1. A method for evaluating power business environment indicators based on artificial intelligence technology, characterized in that: The process includes the following: S100: Obtain power supply data related to the power business environment from the power system; specifically including: Electricity consumption data: electricity consumption, electricity consumption time, electricity cost, and electricity consumption time distribution; Power load data: voltage stability, peak load, average load, and number of faults; Service quality data: service response time, customer satisfaction, and customer complaint rate; Environmental data: carbon emissions, proportion of renewable energy use; Fault records: number of faults, repair time; S200: Cleaning the collected raw data; including: S210: Desensitize sensitive data in the original data to protect user privacy; S220: Remove duplicate values ​​and outliers from the data and fill in missing values; S230: Perform data standardization processing on the data, normalizing it to the interval [0,1] to ensure data quality; S300: Auxiliary classification of business environment indicators into the following three categories: comprehensive electricity consumption indicators, power quality indicators and service quality indicators, and power diagnosis and repair indicators; The comprehensive electricity consumption index includes: electricity efficiency index, which is used to characterize the economic benefits of unit electricity consumption; electricity cost index, which is used to characterize the cost of unit electricity consumption; Power quality indicators include: power load index, which represents the ratio of peak load to average load; power failure index, which represents the product of the average number of failures per year and the repair time; carbon emission index, which represents the carbon emissions per unit of electricity consumption; The service quality indicators include: power supply strategy indicator, which represents the flexibility score of the power supply strategy; Power diagnosis and repair indicators include: a comprehensive score of fault diagnosis accuracy and repair efficiency; S400: Extracting the corresponding key features for each type of indicator; specifically including: S410: Calculate the correlation between features and major indicator variables, and select features that are highly correlated with major indicator variables; S420: then using recursive feature elimination (RFE) to calculate the features that contribute the least to the comprehensive score of the power business environment until the required number of features is reached; S430: Score each feature. If there is a major fault, deduct points accordingly to obtain a score for each feature. S500: The entropy weight method is used to calculate the weight of each indicator; the calculation formula is: Among them, w j is the weight of the jth indicator, D j is the entropy value, f ij is the normalized index value, k is the number of key features mentioned above; S600: Calculate the comprehensive score of the power business environment based on the scores and weights of each indicator; the comprehensive scoring model is: Among them, S j is the score of the jth indicator, w j is the weight of the j-th indicator.

2. The method for evaluating power business environment indicators based on artificial intelligence technology according to claim 1, characterized in that: After S600, it also includes: S700: After extracting the above features and scores, a time series model is constructed through a long short-term memory network (LSTM) for time series prediction and real-time monitoring of indicator changes. The trained LSTM model is used to predict real-time data to obtain a predicted score. The predicted score is compared with the actual score, and a threshold is set. When the difference exceeds the threshold, the abnormal situation is handled in a timely manner.

3. The method for evaluating power business environment indicators based on artificial intelligence technology according to claim 2 is characterized in that: The number of LSTM layers in the LSTM model is usually set to 1 to 3. The fully connected layer maps the output of the LSTM layer to the output space for predicting scores. The mean squared error (MSE) loss function is selected as the loss function for model training to measure the difference between the predicted value and the true value. Multiple iterations of training are performed, and the loss is calculated and the model parameters are updated in each iteration. Finally, the test set is used to evaluate the performance of the model and calculate the mean squared error (MSE) and mean absolute error (MAE) indicators.

4. The method for evaluating power business environment indicators based on artificial intelligence technology according to any one of claims 1 to 3, characterized in that: The S210 includes the following process: S211: Identify sensitive data, clarify the databases, tables, and fields where sensitive data such as user names, ID numbers, electricity addresses, and contact information are located, and classify and grade the sensitive data; S212: Then, according to the data type and business scenario, select the appropriate desensitization algorithm and parameters, and configure specific desensitization rules based on the desensitization algorithm; S213: Desensitize the sensitive data according to the configured desensitization rules.

5. The method for evaluating power business environment indicators based on artificial intelligence technology according to claim 4 is characterized in that: In the above S212, For ID card numbers, a masking algorithm is used to retain the first 10 digits, and the remaining digits are replaced with "X" or "*"; For names, replace them with randomly generated pseudonyms; For electricity addresses, a replacement algorithm is used to replace the real address with fictitious address information; For electricity consumption data, a random offset is added or subtracted through the offset algorithm to make the data vary within a certain range while maintaining the relative relationship of the data; Contact data is encrypted using a symmetric encryption algorithm to ensure data security during transmission and storage.

6. The method for evaluating power business environment indicators based on artificial intelligence technology according to any one of claims 1 to 3, characterized in that: The S220 includes the following process: S221: Use an exact matching algorithm to compare the various fields of the data records to find identical duplicate records; use a fuzzy matching algorithm, such as a basic nearest neighbor sorting algorithm, to determine the similarity by detecting records within a window and identify data with a similarity of more than 90%; after determining duplicate records, retain one record according to business rules and delete other duplicate records. S222: Use the boxplot method to determine the range of outliers by calculating the quartiles of the data; use the Z-Score method within this range to calculate the distance between the data point and the mean, measured in units of standard deviation, and using the 3σ principle. Values ​​exceeding the mean ± 3 times the standard deviation are considered outliers; outliers are directly deleted or retained but marked; S223: Fill in missing values ​​in the data, since the data is divided into different types, including numerical data, categorical data, and time series data.

7. The method for evaluating power business environment indicators based on artificial intelligence technology according to claim 6 is characterized in that: The S223 includes the following process: For numerical data, use the mean or median to fill missing values; For categorical data, the mode is used to fill missing values. For time series data, linear interpolation can be used to estimate missing values. When the missing rate in the data exceeds 30%, the data point is directly removed.

8. The method for evaluating power business environment indicators based on artificial intelligence technology according to any one of claims 1 to 3, characterized in that: The S230 includes the following process: Normalize the data to the interval [0,1], where the formula for data normalization is as follows: Among them, x min and x max are the minimum and maximum values ​​of the data, respectively.

9. The method for evaluating power business environment indicators based on artificial intelligence technology according to any one of claims 1 to 3, characterized in that: In the S400, For the electricity efficiency index, the key features are extracted from the electricity consumption growth rate and peak-to-valley ratio; For power quality indicators, key features are extracted, such as voltage fluctuation rate and fault repair time; For service quality indicators, key features are extracted, such as customer complaint rate and service response time; The electricity consumption growth rate is the rate of change of electricity consumption in adjacent time periods. The formula is: electricity consumption growth rate = (current period electricity consumption - previous period electricity consumption) / previous period electricity consumption × 100%; Peak-valley ratio is the ratio of electricity consumption during peak hours to electricity consumption during off-peak hours. The formula is: Peak-valley ratio = electricity consumption during peak hours / electricity consumption during off-peak hours. The calculation formula for voltage fluctuation rate is: Among them, V i is the voltage value at the i-th time point, V avg is the average voltage.

10. The method for evaluating power business environment indicators based on artificial intelligence technology according to claim 9, characterized in that: In the S400, For electricity consumption growth rate, use a 30-day or 90-day moving average; For fault repair time, the time interval from fault occurrence to repair completion is counted, and the mean and median statistics are calculated; For the customer complaint rate, calculate the ratio of the number of customer complaints to the total number of services. The formula is: Customer complaint rate = number of complaints / total number of services × 100%.

Images