A new energy vehicle charging pile operation data analysis system based on big data

Abstract

This invention relates to the field of data management technology for new energy vehicle charging piles, specifically a big data-based operational data analysis system for new energy vehicle charging piles. The invention first collects operational data from charging piles, and analyzes usage patterns, charging demand predictions, and charging behavior patterns through time-series modeling and feature extraction. Then, it extracts key operational features of the equipment and evaluates the operational status of charging piles under different conditions. Combining usage patterns, charging demand predictions, and charging behavior patterns, it analyzes future peak hours and high-demand areas. Simultaneously, based on operational status, it calculates the working priority of charging piles and identifies potential faults. This invention also combines the load and operational status of regional charging piles to generate real-time charging pile recommendation strategies. Through the collaborative work of multiple units, this invention provides comprehensive support for the efficient management of charging pile systems.

Description

Technical Field

[0001] The present invention relates to the technical field of new energy vehicle charging pile data management, and in particular to a new energy vehicle charging pile operation data analysis system based on big data. Background Art

[0002] New energy vehicle charging piles are devices that provide electric energy to electric vehicles and plug-in hybrid vehicles through cables. They are also equipped with identity recognition, payment systems, and monitoring functions to ensure safety and convenience during charging. At present, with the rapid development of the new energy vehicle industry, charging infrastructure has covered most cities and highways. Although charging speed and network coverage have improved, with the rapid growth of the number of new energy vehicles, problems such as chaotic charging management, insufficient utilization of charging piles, and imperfect scheduling management have become increasingly prominent. Therefore, it is urgent to conduct a comprehensive and accurate analysis of the operating data of new energy vehicle charging piles to provide theoretical and technical support for charging pile managers to upgrade equipment performance and optimize management and scheduling.

[0003] In the face of the above problems, existing technologies have also proposed a variety of solutions. For example, by collecting new energy vehicle operation data and conducting on-site surveys, the matching accuracy and charging safety of new energy vehicle charging piles can be improved; an online management system is used to build a network topology, and real-time analysis of charging pile operation data can be performed to achieve intelligent scheduling and optimized management of charging piles. Some scholars have proposed real-time analysis of temperature and power data under different charging conditions to identify abnormal conditions and issue warnings, thereby improving the safety and efficiency of charging pile use. In addition, there are related studies that record the usage data of charging piles and analyze their health status, and dispatch charging piles in good health to high-frequency usage locations to improve the use efficiency of charging piles.

[0004] The existing technology has improved the management effect of new energy vehicle charging piles to a certain extent, but there are still some shortcomings. Specifically, the existing technology does not analyze the usage patterns and rules of charging piles, nor does it take into account the changes in the operating status of charging piles under different working modes, thus failing to provide overall support for the efficient management and scheduling of charging pile systems.

[0005] Therefore, a new energy vehicle charging pile operation data analysis system based on big data is proposed. Summary of the invention

[0006] The purpose of the present invention is to provide a new energy vehicle charging pile operation data analysis system based on big data. First, the operation data of the charging pile is collected, and the usage mode, charging demand prediction and charging behavior law of the charging pile are analyzed through time series modeling and feature extraction; then, the key characteristics of the equipment operation are extracted to evaluate the operation status of the charging pile in different states; combined with the usage mode, charging demand prediction and charging behavior law, the future peak period and high demand area are analyzed, and at the same time, the working priority of the charging pile is calculated and potential faults are identified based on the operation status. The present invention also combines the load condition and operation status of the regional charging pile to generate a real-time charging pile recommendation strategy.

[0007] To achieve the above purpose, the present invention provides a new energy vehicle charging pile operation data analysis system based on big data, comprising:

[0008] Data collection unit: used to collect charging pile operation data and transmit the charging pile operation data to the cloud big data platform for storage and management; the charging pile operation data includes charging behavior data, equipment operation data, location information data and usage frequency data;

[0009] Usage mode analysis unit: including a time series modeling module and a feature extraction module; the time series modeling module is used to perform time series dependency modeling on the charging pile operation data; the feature extraction module is used to extract complex features from the time series dependency, analyze and obtain the charging pile usage mode, charging demand prediction and charging behavior rules;

[0010] Furthermore, the steps of the timing modeling module are:

[0011] Preprocessing the charging behavior data and the usage frequency data to construct them into time series data;

[0012] The time series data is input into the Transfomer encoder, and the self-attention mechanism is used to capture the time characteristics and long-distance dependencies of the charging behavior, so as to generate the time series feature representation of the charging pile operation data.

[0013] Furthermore, the steps of the feature extraction module include:

[0014] Feature engineering technology is used to extract complex features from time series feature representation, including: calculating the number of charging times and charging amount in each time period to obtain time period characteristics; analyzing the distribution of charging frequency and charging duration to obtain charging behavior characteristics;

[0015] Calculate the geographical distribution density of charging piles based on geographic information data to obtain location characteristics;

[0016] The complex features are fused to form a unified feature vector.

[0017] Furthermore, the usage pattern analysis unit also includes:

[0018] Perform cluster analysis on the time series feature representation and the unified feature vector to identify the usage frequency, regional distribution and time period distribution of the charging piles, and obtain the usage pattern of the charging piles;

[0019] Training a charging demand prediction model according to historical charging pile operation data, taking the time series feature representation and the unified feature vector as inputs of the charging demand prediction model, and obtaining the charging demand prediction;

[0020] The charging behavior rule is obtained according to the charging behavior feature extracted from the time series feature representation.

[0021] Operation status analysis unit: including a data processing module and an operation status evaluation module; the data processing module is used to extract key features from the device operation data; the operation status evaluation module is used to evaluate the operation status of the charging pile under different states and obtain the operation score and operation trend;

[0022] Furthermore, the steps of the data processing module include:

[0023] Performing data cleaning and data standardization on the equipment operation data to obtain equipment operation standard data; extracting working state operation characteristics and non-working state operation characteristics from the equipment operation standard data;

[0024] The operating characteristics of the working state include voltage fluctuation rate, current fluctuation rate, temperature change rate, fault frequency and power consumption; the operating characteristics of the non-working state include standby power consumption, voltage and current stability in the non-working state, ambient temperature and cumulative operating time;

[0025] The working state operation characteristics and the non-working state operation characteristics are respectively combined into a first operation characteristic vector and a second operation characteristic vector.

[0026] Furthermore, the steps of the operation status evaluation module include: according to the first operation feature vector pass Calculate the first run score ; According to the second running feature vector pass Calculate the second run score ;in, and are the number of features in the first running feature vector and the second running feature vector, respectively, and is the weight of the first running feature vector and the second running feature vector;

[0027] Collect the first operation scores at a specified time, define a sliding window to extract short-term time features; input the short-term time features into a multi-layer convolutional layer to capture local time series patterns in the first operation scores;

[0028] A gated recurrent unit is used to obtain the long-term dependency in the local time series pattern, and a first operation trend is obtained by calculating the operation score prediction through a fully connected layer;

[0029] The second operation scores at a specified time are collected, and a linear regression method is used to obtain a second operation trend.

[0030] Scheduling demand analysis unit: includes a demand analysis module and a potential fault identification module; the demand analysis module is used to analyze and obtain future peak periods and high-demand areas based on the charging pile usage pattern, the charging demand forecast and the charging behavior pattern; the potential fault identification module is used to calculate the working priority of the charging pile and identify potential faults based on the operation score and the operation trend.

[0031] Furthermore, the steps of the demand analysis module include:

[0032] Identify future peak hours and high-demand areas based on charging demand forecasts and charging behavior patterns; including:

[0033] All time periods are based on the predicted charging demand Sort in descending order, select the one with the highest demand The time period is taken as the future peak period;

[0034] Aggregate charging piles by region based on geographical location and calculate the total demand in each region ;according to Sort all regions in descending order; select the regions with the highest demand as the high demand area.

[0035] Furthermore, the steps of the potential fault identification module include:

[0036] Calculating a work priority score of the charging pile according to the operation score and the operation trend; ;

[0037] in, rate the priority of said work; and are the slopes of the first and second operating trends, respectively. and are the weights of the first run score and the second run score, respectively; and are the weights of the first operating trend and the second operating trend respectively;

[0038] A fault threshold is set, and the fault risk of the charging pile is identified according to the work priority score.

[0039] Furthermore, it also includes real-time charging pile recommendation strategy;

[0040] The real-time charging pile recommendation strategy includes regional load calculation and recommendation list generation;

[0041] Calculate the load rate of each charging pile in the real-time demand area based on the current power and maximum power of the charging pile;

[0042] When the charging pile is in an idle state, the current power is the current charging power, and the maximum power is the maximum charging power; when the charging pile is in an idle state, the current power is the power consumed by the charging pile itself, and the maximum power is the maximum carrying power;

[0043] According to calculating the load condition of the real-time demand area.

[0044] Furthermore, the step of generating the recommendation list includes:

[0045] Obtain real-time charging requirements, including charging mode and charging time; compare the charging mode with the charging pile Perform pattern matching; combine the charging pile Priority score for work , the charging pile Load rate , and the load calculation of the real-time demand area , obtain the recommended scores of all charging piles in the real-time demand area ; ;

[0046] in, , and Score the priority of each task , the load rate And the load calculation situation Dynamic weight of is the result of the pattern matching, when the charging pile When the charging mode is supported, ; When the charging pile When the charging mode is not supported, ;

[0047] The recommendation scores of all charging piles in the real-time demand area are sorted in descending order to generate a real-time charging pile recommendation list.

[0048] Compared with the prior art, the present invention has the following beneficial effects:

[0049] 1. The present invention collects the operating data of new energy vehicle charging piles, and through time series modeling and complex feature extraction methods, deeply analyzes the usage patterns of charging piles, predicts charging needs, and the laws of charging behavior. The Transformer encoder and self-attention mechanism can accurately capture the temporal characteristics and long-distance dependencies of charging behavior, and generate high-quality time series feature representations. The feature vectors extracted using specific feature engineering methods can comprehensively reflect the operating status and charging needs of the charging piles. The clustering algorithm is used to analyze the time series features and unified feature vectors, accurately predict future peak hours and high-demand areas, and extract charging behavior laws. Through the above-mentioned multi-level data analysis, the present invention achieves a deep understanding of the usage patterns of charging piles and accurate prediction of charging needs, thereby providing data support for optimizing the scheduling and management strategies of charging piles.

[0050] 2. The present invention first performs data cleaning and standardization on the equipment operation data of the charging pile, extracts the characteristics of voltage fluctuation rate, temperature change rate, etc. in the working state, and the characteristics of standby power consumption, ambient temperature, etc. in the non-working state, and generates the corresponding operation feature vector. Through multi-layer convolutional layers and gated recurrent units (GRU), time series analysis is performed on the working state operation score to capture short-term and long-term dependencies and predict the working state operation trend; at the same time, the linear regression method is used to analyze the operation trend in the non-working state. This method can not only help identify potential faults of charging piles, but also balance the equipment loss of regional charging piles, providing theoretical and data support for improving the availability, system life and operation and maintenance efficiency of the charging pile system.

[0051] 3. The present invention identifies future peak hours and high-demand areas based on the usage patterns, charging demand forecasts, and charging behavior patterns of the charging piles; and calculates the work priority scores of the charging piles based on the operation scores and operation trends of the charging piles. The operating status of the charging piles is evaluated by analyzing the slope of the operation trend. Through precise demand analysis and potential fault identification, the present invention can provide a reference for optimizing the charging pile scheduling plan in advance and improve resource utilization; it can also timely detect potential faults based on operation scores and trends, reduce the failure rate, and improve the reliability of the system. This method combines charging demand forecasting with equipment operation management, realizes intelligent scheduling and efficient management of the charging pile system, and enhances the system's operational stability and resource allocation capabilities. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 It is a flow chart of a data analysis system for new energy vehicle charging pile operation based on big data;

[0053] Figure 2 It is a schematic diagram of the framework of a new energy vehicle charging pile operation data analysis system based on big data;

[0054] Figure 3 Schematic diagram of the process framework for real-time charging pile recommendation strategy. DETAILED DESCRIPTION

[0055] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0056] The present invention provides a new energy vehicle charging pile operation data analysis system based on big data. The specific system flow chart and system structure diagram refer to Figure 1 and Figure 2 The present invention will describe a specific implementation method based on the charging pile operation data of a large-scale new energy vehicle charging station A. In addition, a specific implementation method of a real-time charging pile recommendation strategy will also be described. Example

[0057] As an embodiment of the present invention, refer to Figure 1 S10 in the embodiment is applied to a data acquisition unit of a new energy vehicle charging pile operation data analysis system based on big data. The data acquisition unit is used to collect charging pile operation data and transmit it to a cloud big data platform for storage and management; the charging pile operation data includes charging behavior data, equipment operation data, location information data and usage frequency data.

[0058] Further, refer to Figure 1 S20 in the embodiment is applied to a usage mode analysis unit of a new energy vehicle charging pile operation data analysis system based on big data. The charging behavior data and usage frequency data in the charging pile operation data are preprocessed to construct time series data; specifically including:

[0059] Data of different dimensions are processed by minimum-maximum normalization.

[0060] The charging behavior data and usage frequency data are arranged in chronological order to construct a multi-dimensional time series: ;

[0061] in, Indicates a point in time, Represents the multidimensional data at that point in time, is the total number of time points.

[0062] Divide the data into segments and set the time step , each segment contains time steps, forming a sequence block: ;

[0063] For each time series block , through the Transformer embedding layer Map it into a high-dimensional vector;

[0064] The self-attention mechanism is used to calculate the data correlation between each time step to capture the temporal features: ;

[0065] in, , and They are the query vector, key vector and value vector in the self-attention mechanism respectively; is the dimension of the key vector.

[0066] The captured temporal features are mapped through the fully connected layer: ;

[0067] in, is the final time series feature representation; is the weight matrix of the fully connected layer, For bias.

[0068] This embodiment captures the time dependency and long-distance features in the charging pile operation data through the transformer encoder and self-attention mechanism. The self-attention mechanism can flexibly focus on the correlation of different time steps, and the transformer model can more accurately process large-scale and multi-dimensional data and generate high-quality time series feature representations; the combination of the two provides high-quality data support for deep mining of charging pile usage patterns, charging behaviors, and demand changes.

[0069] Furthermore, a specific feature engineering method is used to extract complex features from the time series feature representation, including:

[0070] Time period characteristics: calculate the number of charging times and charging amount in each time period; ;

[0071] ;

[0072] in, Indicates a charging event. For the The amount of charge per charge.

[0073] Charging behavior characteristics: calculate the charging frequency in each time period and average charging time ; ;

[0074] ;

[0075] in, For the The charging time for each charge.

[0076] Location characteristics: Based on the geographical location information of the charging piles, calculate their distribution density in the area to which they belong. There are Charging stations, and area The area is , then the density of charging piles in this area is ; The extracted time period features, charging behavior features and location features are integrated into a unified feature vector; ;

[0077] This embodiment extracts complex features from the time series feature representation for multi-dimensional understanding and analysis of the operation of charging piles. The time period feature can capture the charging demand in different time intervals; the location feature reflects the utilization rate of the spatial distribution of charging piles; and the charging behavior feature provides a distribution display of charging behavior. Fusion of these features into a unified feature vector can provide multi-dimensional data support for subsequent analysis, prediction and optimization.

[0078] Further, the time series feature is represented as With the unified eigenvector Splice to generate a multi-dimensional input feature matrix ;

[0079] Use K-means algorithm to input feature matrix Perform clustering to identify usage pattern categories of charging stations; ;

[0080] in, is the objective function, is the total number of samples; For the samples, indicating the Multi-dimensional feature vector of charging piles; is the number of clusters, For the The center of the cluster; is an indicator variable, when the sample Belongs to cluster hour, ,on the contrary .

[0081] Through the clustering algorithm, high / low frequency charging piles, high / low frequency areas, and high / low frequency time periods are obtained.

[0082] The charging demand prediction model is trained based on the historical charging pile operation data, and the time series feature table and the unified feature vector are used as inputs of the charging demand prediction model to obtain the charging demand prediction. Specifically:

[0083] Extract the number of charging times of each charging pile in the past period of time from the time series feature representation and the unified feature vector , Charging time and charging behavior characteristics , which constitutes historical data ;

[0084] A long short-term memory network is used to capture the temporal dependency of data and predict future charging demand; the recursive equation of LSTM is: ;

[0085] in, is the hidden state of LSTM, representing the time step Forecast of charging demand.

[0086] Mean square error As the loss function, it is used to evaluate the Predicted charging demand for each time step And the actual charging demand The difference between is the total number of time steps used for training.

[0087] Output the trained model to predict charging demand in the future time period .

[0088] According to the charging behavior features extracted from the time series feature representation and , analyze the charging behavior patterns.

[0089] This embodiment collects the charging behavior data and charging frequency data of the new energy vehicle charging pile operator A in the past week. Tables 1 and 2 give some data examples.

[0090] Table 1 Some examples of charging behavior data

[0091]

[0092] Table 2 Some examples of frequency of use data

[0093]

[0094] This embodiment also collects the location data of charging piles deployed by the new energy vehicle charging pile operator A in the urban area B. Table 3 gives some examples of the location data.

[0095] Table 3: Partial example of location data

[0096]

[0097] In Table 3, the specific location of the charging pile is represented in the form of position coding. Taking the A-001 charging pile as an example, its position code 01031221001 means charging pile No. 001 located in the 21st community, 12th grid, 03rd street, 01 district.

[0098] The usage pattern analysis unit of the new energy vehicle charging pile operation data analysis system based on big data proposed in this embodiment performs time series modeling and feature extraction, and clusters high / low frequency charging piles, high / low frequency usage areas, and high / low frequency usage time periods; some clustering results are shown in Table 4.

[0099] Table 4 Clustering results

[0100]

[0101] Whether a charging pile is a high-frequency charging pile or a low-frequency charging pile is determined by the number of times it is charged and the cumulative charging time. If a charging pile is used less frequently but has a longer cumulative charging time, it is because the charging mode is slow charging. Here, it is still determined to be a high-frequency charging pile. High-frequency areas and low-frequency areas are mainly due to their regional functions. For example, QY-01 and QY-02 in Table 4 are urban commercial residential areas and science and technology industrial parks; QY-13 and QY-14 are two remote industrial parks. The high-frequency use time period and the low-frequency use time period are different in each area. Specifically, in the high-frequency use area, the high-frequency use time period is the late night period and the morning / afternoon non-working peak time period; in the low-frequency use area, there is no obvious distinction in the use time period.

[0102] Based on the collected data, a one-week charging demand forecast is made for each charging pile, and some examples of the forecast results are given in Table 5.

[0103] Table 5 Some examples of charging pile demand forecast

[0104]

[0105] As can be seen from Table 5, the estimated number of charging times for fast charging piles and slow charging piles is quite different, but the estimated cumulative duration is similar. Both are at a high level of charging demand. In other words, fast charging piles and slow charging piles are in a state of roughly balanced demand. Operators can reasonably arrange the service resources of charging piles according to the predicted demand for charging piles to ensure the normal operation of the charging pile system.

[0106] The charging behavior of each charging pile is analyzed based on the collected data, and some examples of the analysis results are given in Table 6.

[0107] Table 6 Some examples of charging behavior patterns

[0108]

[0109] As can be seen from Table 6, fast charging pile A-001 is used more frequently in the morning and afternoon, while slow charging pile A-005 is used more frequently late at night. Operators can reasonably allocate the number of charging points and operation and maintenance personnel according to the charging behavior rules to improve service efficiency.

[0110] Through cluster analysis, this embodiment can identify the usage frequency, regional distribution and usage habits of charging piles, and help understand the usage patterns of charging piles; through the charging demand prediction model trained with historical data, it can predict future charging demand in advance and provide data support for reasonable resource scheduling; by analyzing the patterns of charging behavior, it can help optimize charging services and provide a theoretical basis for charging pile managers to improve the distribution settings of charging piles.

[0111] Further, refer to Figure 1 S30 in the embodiment is applied to an operation status analysis unit of an operation data analysis system for a new energy vehicle charging pile based on big data. According to the operation data of the equipment, key features are extracted, the operation status of the charging pile under different states is evaluated, and the operation score and operation trend are obtained. Specifically, data processing and operation status evaluation are included;

[0112] The data processing step includes: performing data cleaning and data standardization on the equipment operation data to obtain equipment operation standard data;

[0113] Extracting working state operation characteristics and non-working state operation characteristics from the equipment operation standard data respectively;

[0114] The acquisition of working status operation characteristics includes: Calculate the voltage fluctuation rate; by Calculate the current fluctuation; by Calculate the rate of temperature change; by Calculate the power consumed by the device during charging; obtain the fault frequency by counting the number of faults over a period of time;

[0115] The acquisition of non-working state operation characteristics includes: measuring the power of the charging device in standby mode; voltage fluctuation rate and current fluctuation in the non-working state; average ambient temperature and cumulative operating time; examples of some working state operation characteristics and non-working state operation characteristics are given in Table 7.

[0116] Table 7 Example of some operating characteristics

[0117]

[0118] The working state operation characteristics and the non-working state operation characteristics are respectively combined into a first operation characteristic vector and a second operation characteristic vector.

[0119] This embodiment can ensure the accuracy and comparability of equipment operation data through data cleaning and standardization. By extracting the key features of the charging pile in the working state and the non-working state and combining them into feature vectors, it can help to comprehensively and accurately evaluate the operating conditions of the charging pile in different working modes. The characteristics of the working state help detect the possible fault risks during the charging process, while the characteristics of the non-working state can reflect the operating conditions and energy efficiency performance of the equipment in a static state.

[0120] Furthermore, the operation status evaluation step includes:

[0121] According to the first run feature vector pass Calculate the first run score ; According to the second running feature vector pass Calculate the second run score ;in and are the number of features in the first running feature vector and the second running feature vector, respectively, and is the weight of each feature.

[0122] Collect the first running score in the past period of time , and conduct time series analysis on it.

[0123] Define a sliding window Extract short-term features for each time period ; The sliding window formula is: ;

[0124] in, is the current time step, and the size of the sliding window is , extracting the past The first run score within time steps;

[0125] The short-term features extracted Input into the multi-layer convolutional neural network, through Extracting local correlations of short-term features ; Then pass Max pooling is used to reduce the dimensionality and obtain the local time series pattern in the first run score;

[0126] To capture the long-term time series dependencies of the first run scores, local time series patterns Input to the gated recurrent unit Calculate; where yes The hidden state at the current time step represents the long-term dependencies captured at that moment;

[0127] Through the fully connected layer, The hidden state of Mapping to a new first operation score prediction value; recording multiple first operation score prediction values ​​to obtain a first operation trend;

[0128] Collect the second running score in the past period of time , a linear regression method is used to obtain the second operating trend. The linear regression model is , For the The second run score prediction value at time, and is the predicted value of the regression coefficient obtained by fitting the historical second run score.

[0129] Record the predicted values ​​of the second operation scores of multiple points and obtain the second operation trend. Some operation scores and operation trends are given as examples in Table 8.

[0130] Table 8 Some examples of operation scores and operation trends

[0131]

[0132] Table 8 gives specific values ​​of some operation scores and some operation trends. In the actual analysis process, they can be combined with hardware equipment to give an intuitive display in the form of, but not limited to, line graphs.

[0133] This embodiment comprehensively evaluates the operating status of the charging pile by combining the multi-dimensional operating characteristics in the working state and the non-working state. For the working state, the long-term and short-term time series patterns are captured by sliding windows, convolutional networks and gated recurrent units to achieve the operation trend prediction in the working state; for the non-working state, linear regression is used to simplify the operation evaluation process; this method provides reliable data support and decision-making basis for early detection of potential equipment problems, extension of equipment service life, optimization of management and preventive maintenance.

[0134] Further, refer to Figure 1 S40 in which S40 is applied to the scheduling demand analysis unit and includes a demand analysis phase; the steps of the demand analysis phase are: analyzing and obtaining future peak hours and high demand areas according to the charging pile usage mode, the charging demand forecast, and the charging behavior law;

[0135] Based on the charging demand forecast and charging behavior rules, identify the future peak hours and high demand areas of the charging piles; including:

[0136] Get Charging demand forecast data for each time period , sort them in descending order according to the predicted charging demand;

[0137] ;

[0138] From the sorted Select the demand in the front The time period is taken as the future peak period;

[0139] In two-dimensional coordinate form Collect the geographical location of each charging pile, aggregate the charging piles by region according to the geographical location, and calculate the total demand in each region ;in, Indicates each area The total predicted demand for charging piles in Indicates Charging stations;

[0140] Total demand for each region Sort in descending order;

[0141] ;

[0142] From the sorted The selected area demand is in the front as the high demand area.

[0143] This embodiment analyzes charging demand forecasts and charging behavior patterns to identify future charging peak hours and high-demand areas. This process can not only help optimize resource allocation, but also improve the utilization rate of charging piles, alleviate the pressure during charging peak hours, and provide assistance to the operation and maintenance of charging pile managers.

[0144] Further, refer to Figure 1 S40 in the above method is applied to the potential fault identification stage of the scheduling demand analysis unit; the step of potential fault identification includes:

[0145] quantifying the first operation trend and the second operation trend by slope, and calculating the work priority score of the charging pile according to the operation score and the operation trend; ;

[0146] in, is the working priority of the charging pile; and are the slopes of the first and second operating trends, respectively. and The first run score and the second run score The weight of and are the weights of the first operating trend and the second operating trend respectively;

[0147] A fault threshold is set, and the fault risk of the charging pile is identified according to the work priority score. The fault threshold is set based on historical data or industry experience, and its influencing factors include but are not limited to environment and equipment type.

[0148] In order to verify the effect of the fault identification method proposed in the present invention, this embodiment selects three currently commonly used charging pile fault identification methods based on charging pile operation data, conducts comparative experiments on the data collected in this embodiment, and records the comparative experimental results in Table 9.

[0149] Method 1: Sensor-based detection method; setting safety thresholds and detecting anomalies through voltage and current sensors;

[0150] Method 2: Detection algorithm based on pattern recognition; collect voltage and current waveform data at high frequency, and identify faults through signal processing;

[0151] Method 3: Rule-based method: historical experience and cases are summarized to form a rule base, and real-time monitoring data is matched with the rule base to trigger response fault diagnosis;

[0152] Method 4: The method proposed in the present invention is based on charging pile operation score and operation trend.

[0153] Table 9 Comparison results of fault identification methods

[0154]

[0155] This embodiment combines the operation score of the charging pile with the operation trend to evaluate the working priority of the charging pile and identify potential fault risks in advance. The working priority not only depends on the current equipment operation status, but also takes into account the future operation change trend of the equipment; and the operation status of the equipment when it is working and the operation status of the equipment when it is not working are also taken into consideration. This method makes fault identification more timely and accurate, reducing unnecessary losses caused by equipment failure.

[0156] Embodiment 2

[0157] As an embodiment of the present invention, refer to Figure 3 , Figure 3 A real-time charging pile recommendation strategy is applied to a new energy vehicle charging pile operation data analysis system based on big data. Figure 3 A description of the specific implementation method for implementing the recommended charging pile strategy.

[0158] The real-time charging pile recommendation strategy includes regional load calculation and recommendation list generation;

[0159] When charging pile area A receives a demand for charging of new energy vehicles, it collects the real-time status of all charging piles in the area;

[0160] according to Calculate the load rate of each charging pile in the real-time demand area.

[0161] When the charging pile is in working state, is the charging power, is the maximum charging power; when the charging pile is in an idle state, the is the power consumed by the charging pile itself, is the maximum carrying power. Calculate the load condition of the real-time demand area; wherein, For charging pile The power, For the charging pile Maximum power; is the total number of charging piles in the real-time demand area.

[0162] This embodiment calculates the load rate of each charging pile in the real-time demand area to accurately obtain the current status and usage of the charging piles, and also calculates the overall load situation in the real-time demand area. This provides key data support for the subsequent generation of recommendation lists, ensuring that the recommendation strategy can be made based on the actual load situation and the balance of regional supply and demand.

[0163] Furthermore, the step of generating the recommendation list includes:

[0164] Get the charging mode and compare it with the charging pile Perform pattern matching; combine the charging pile Priority score for work , the charging pile Load rate , and the load calculation of the real-time demand area , get the recommended score of each charging pile ; ;

[0165] in, , and Respectively , and stated Dynamic weight of As a result of pattern matching, when the charging pile When the charging mode is supported, ; When the charging pile When the charging mode is not supported, .

[0166] The charging piles in the real-time demand area are sorted in descending order according to the recommendation scores to generate a real-time charging pile recommendation list.

[0167] Based on the load calculation results of the previous stage, this embodiment provides users with accurate charging pile recommendations by comprehensively analyzing the load conditions of each charging pile in the real-time demand area. This method not only takes into account the availability of charging piles, but also takes into account the overall supply and demand balance in the real-time demand area, ensuring that charging pile recommendations can not only meet user needs, but also help charging pile managers extend the service life of the equipment.

[0168] The present invention integrates the operating data of new energy vehicle charging piles, makes full use of big data technology and advanced deep learning algorithms, and deeply understands and analyzes the charging behavior patterns and laws of charging piles. The design of the present invention improves the utilization rate and operating efficiency of charging pile resources, which is conducive to alleviating the charging pressure during peak hours. At the same time, the present invention also analyzes the operating status of the charging pile to identify potential faults, which ensures the reliability of the charging pile and further improves the overall operating efficiency of the charging pile. In addition, the present invention provides users with accurate charging pile recommendations by analyzing the charging demand in real time, integrating the operating status, load and regional load of the charging pile.

[0169] In summary, the present invention comprehensively analyzes the operating data of new energy vehicle charging piles through big data technology, providing data and theoretical support for charging pile managers to understand the usage of charging piles, ensure charging pile supply management, and optimize regional coordinated scheduling.

[0170] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A new energy vehicle charging pile operation data analysis system based on big data, characterized in that: include: Data collection unit: used to collect charging pile operation data and transmit the charging pile operation data to the cloud big data platform for storage and management; The charging pile operation data includes charging behavior data, equipment operation data, location information data and usage frequency data; Usage mode analysis unit: including a time series modeling module and a feature extraction module; the time series modeling module is used to perform time series dependency modeling on the charging pile operation data; The feature extraction module is used to extract complex features from time series dependencies, analyze and obtain charging pile usage patterns, charging demand predictions, and charging behavior rules; Operation status analysis unit: comprising a data processing module and an operation status evaluation module; the data processing module is used to extract key features from the equipment operation data; The operating status evaluation module is used to evaluate the operating status of the charging pile in different states and obtain the operating score and operating trend; The steps of the data processing module include: Performing data cleaning and data standardization on the equipment operation data to obtain equipment operation standard data; extracting working state operation characteristics and non-working state operation characteristics from the equipment operation standard data; The operating characteristics of the working state include voltage fluctuation rate, current fluctuation rate, temperature change rate, fault frequency and power consumption; the operating characteristics of the non-working state include standby power consumption, voltage and current stability in the non-working state, ambient temperature and cumulative operating time; Combining the working state operation characteristics and the non-working state operation characteristics into a first operation characteristic vector and a second operation characteristic vector respectively; The steps of the operation status assessment module include: According to the first run feature vector pass Calculate the first run score ; According to the second running feature vector pass Calculate the second run score ;in, and are the number of features in the first running feature vector and the second running feature vector, respectively, and is the weight of the first running feature vector and the second running feature vector; Collect the first operation scores at a specified time, define a sliding window to extract short-term time features; input the short-term time features into a multi-layer convolutional layer to capture local time series patterns in the first operation scores; A gated recurrent unit is used to obtain the long-term dependency in the local time series pattern, an operation score prediction is calculated through a fully connected layer, and a first operation score prediction value of multiple points is recorded to obtain a first operation trend; Collect the second operation scores at the specified time, use a linear regression method to record the multi-point second operation score prediction values, and obtain the second operation trend; The dispatch demand analysis unit includes a demand analysis module and a potential fault identification module; the demand analysis module is used to analyze and obtain future peak hours and high-demand areas according to the charging pile usage pattern, the charging demand forecast and the charging behavior law; the potential fault identification module is used to calculate the working priority of the charging pile according to the operation score and the operation trend, and identify potential faults; The steps of the potential fault identification module include: Calculating a work priority score of the charging pile according to the operation score and the operation trend; ； in, rate the priority of said work; and are the slopes of the first and second operating trends, respectively. and are weights of the first running score and the second running score respectively; and are weights of the first operating trend and the second operating trend respectively; A fault threshold is set, and the fault risk of the charging pile is identified according to the work priority score.

2. According to the big data-based new energy vehicle charging pile operation data analysis system of claim 1, it is characterized in that: The steps of the timing modeling module are: Preprocessing the charging behavior data and the usage frequency data to construct them into time series data; The time series data is input into the Transfomer encoder, and the self-attention mechanism is used to capture the time characteristics and long-distance dependencies of the charging behavior, so as to generate the time series feature representation of the charging pile operation data.

3. According to the big data-based new energy vehicle charging pile operation data analysis system of claim 1, it is characterized in that: The steps of the feature extraction module include: Feature engineering technology is used to extract complex features from time series feature representation, including: calculating the number of charging times and charging amount in each time period to obtain time period characteristics; analyzing the distribution of charging frequency and charging duration to obtain charging behavior characteristics; Calculate the geographical distribution density of charging piles based on geographic information data to obtain location characteristics; The complex features are fused to form a unified feature vector.

4. According to the big data-based new energy vehicle charging pile operation data analysis system of claim 1, it is characterized in that: The usage pattern analysis unit also includes: Perform cluster analysis on the time series feature representation and the unified feature vector to identify the usage frequency, regional distribution and time period distribution of the charging piles, and obtain the usage pattern of the charging piles; Training a charging demand prediction model according to historical charging pile operation data, taking the time series feature representation and the unified feature vector as inputs of the charging demand prediction model, and obtaining the charging demand prediction; The charging behavior rule is obtained according to the charging behavior feature extracted from the time series feature representation.

5. According to the big data-based new energy vehicle charging pile operation data analysis system of claim 1, it is characterized in that: The steps of the demand analysis module include: Identify future peak hours and high-demand areas based on charging demand forecasts and charging behavior patterns; including: All time periods are based on the predicted charging demand Sort in descending order, select the one with the highest demand The time period is taken as the future peak period; Aggregate charging piles by region based on geographical location and calculate the total demand in each region ;according to Sort all regions in descending order; select the regions with the highest demand as the high demand area.

6. The big data-based new energy vehicle charging pile operation data analysis system according to claim 1 is characterized in that: It also includes real-time charging station recommendation strategy; The real-time charging pile recommendation strategy includes regional load calculation and recommendation list generation; Calculate the load rate of each charging pile in the real-time demand area based on the current power and maximum power of the charging pile; When the charging pile is in working state, the current power is the current charging power, and the maximum power is the maximum charging power; when the charging pile is in idle state, the current power is the power consumed by the charging pile itself, and the maximum power is the maximum carrying power; The load condition of the real-time demand area is calculated according to the current power and the maximum power of all the charging piles.

7. A new energy vehicle charging pile operation data analysis system based on big data according to claim 6, characterized in that: The step of generating the recommendation list comprises: Obtain real-time charging requirements, including charging mode and charging time; compare the charging mode with the charging pile Perform pattern matching; combine the charging pile Priority score for work , the charging pile Load rate , and the load calculation of the real-time demand area , obtain the recommended scores of all charging piles in the real-time demand area ; ； in, , and Score the priority of each task , the load rate And the load calculation situation Dynamic weight of is the result of the pattern matching, when the charging pile When the charging mode is supported, ; When the charging pile When the charging mode is not supported, ; The recommendation scores of all charging piles in the real-time demand area are sorted in descending order to generate a real-time charging pile recommendation list.

Images