Intelligent charging management method and system for new energy commercial vehicle

Abstract

The application discloses an intelligent charging management method and system for new energy commercial vehicles, relates to the technical field of charging management, and calculates a multivariate joint offset index of each vehicle by analyzing multi-dimensional data uploaded by the vehicle, and evaluates the coupling deviation degree between variables; meanwhile, multi-dimensional data uploaded by each history is extracted, a linear abnormality index is calculated, and the abnormality degree of current data compared with historical trends is measured. A false index is calculated by combining the two, the data authenticity is judged: if the data is real, a charging strategy is formulated according to a preset rule; if the data is false, the strategy generation is stopped; the management platform can judge the authenticity of the data uploaded by the new energy commercial vehicle, when the uploaded data is real, the management platform can generate a corresponding charging strategy based on the real data of the new energy commercial vehicle, and ensure that the overall new energy commercial vehicle fleet can perfectly execute an operation plan, and ensure the scheduling efficiency of the overall fleet, and the overall charging balance resource mismatch.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of charge management, in particular to an intelligent charge management method and system for new energy commercial vehicles. BACKGROUND

[0002] New energy commercial vehicles refer to various commercial transport vehicles mainly driven by electricity, such as electric buses, logistics distribution vehicles, sanitation vehicles, engineering operation vehicles and other large commercial electric vehicles, which have the characteristics of long running mileage, large carrying capacity and high use frequency, and are widely used in urban public transportation, logistics distribution and industrial transportation scenes. Since this type of vehicle has high operation intensity, intensive scheduling and strong dependence on energy supply, the charge management of new energy commercial vehicle fleet is particularly important. If the charge scheduling is improper, it may directly lead to the vehicle being unable to start on time, the task being delayed, and thus affecting the stability of the entire logistics, transportation or public service chain. The traditional charge management of new energy commercial vehicles mainly relies on a rule-driven centralized scheduling strategy, which is mainly dependent on the battery state information (such as SOC) reported by the vehicle at regular intervals and preset charging rules, such as the rule logic of "arranging charging for vehicles with the lowest SOC first". The management platform usually plans the charging mode of each vehicle based on these reported data to ensure that the new energy commercial vehicle can perfectly execute the preset operation plan.

[0003] However, in the actual operation environment, this mechanism based on "vehicle self-reporting state + rule-driven scheduling" has a key hidden danger that the authenticity of the vehicle reporting data is difficult to guarantee. Especially in a platformized fleet with multiple vehicle types mixed, vehicles from different suppliers, and some vehicles accessed by outsourced operators, the BMS systems of different vehicles have significant differences in data accuracy, timeliness and consistency. More complexly, some vehicles may intentionally exaggerate the charging urgency, conceal the current SOC or other state information, in order to "trick" better scheduling resources. For example, a vehicle with sufficient power may report an abnormally low SOC, and the system may consider it as a priority target for energy supply, thereby disrupting the original fair and reasonable scheduling order and causing the delay of charging or even the interruption of the truly high-priority task vehicle. Such "state fraud" is often difficult to be discovered in a short time, and the management platform defaults to trust the reported information and lacks a mechanism to judge its credibility, which may eventually lead to a decrease in scheduling efficiency, overall resource mismatch and disruption of operation plans. SUMMARY

[0004] The purpose of the present application is to solve the above-mentioned problems and provide an intelligent charge management method and system for new energy commercial vehicles.

[0005] In the first aspect of the embodiment of the present application, an intelligent charging management method for new energy commercial vehicles is provided, which comprises:

[0006] The multi-dimensional data uploaded by each vehicle to the management platform is acquired, and the multi-dimensional data is analyzed to calculate the multi-variable joint offset index of each vehicle;

[0007] The multi-dimensional data uploaded by each vehicle in each history is extracted, and the multi-dimensional data uploaded by each vehicle in each history is analyzed to calculate the linear abnormality index of each vehicle;

[0008] The false index of the multi-dimensional data uploaded by each vehicle is calculated according to the multi-variable joint offset index and the linear abnormality index of each vehicle, and whether the multi-dimensional data uploaded by each vehicle is real is judged according to the false index;

[0009] If the multi-dimensional data uploaded by the vehicle is real, the charging strategy of the vehicle is determined according to the preset charging rule, and the vehicle is charged according to the corresponding charging strategy.

[0010] Optionally, the step of calculating the multi-variable joint offset index of each vehicle is:

[0011] The multi-dimensional data uploaded by each vehicle to the management platform in the latest preset time window is recorded as the multi-dimensional data uploaded by each vehicle to the management platform in this round, and the uploading times of the multi-dimensional data uploaded to the management platform in this round are acquired; and a plurality of variables are extracted from the multi-dimensional data, and a standardized first-order difference sequence of each variable is constructed according to the variable value of each variable uploaded each time;

[0012] For each time point , a pair of variables is formed by any two variables corresponding to the time point , the included angle cosine value of each pair of variables is calculated, and the formula is: , wherein is a minimum constant to prevent the denominator from being 0, and the value is ; represents the included angle cosine value of the th variable and the th variable at the time point , and represents the included angle between the change trends of the th variable and the th variable;

[0013] For each pair of variables, the included angle cosine value sequence in the entire preset time window is acquired, the included angle cosine value sequence is discretely divided into equal width intervals, the probability of each interval is counted, and the cooperative included angle entropy of the pair of variables is calculated, and the formula is: , wherein, represents the cosine of the angle between the first variable and the second variable, is a preset number of partition discrete intervals;

[0014] The standard deviation of the cosine of the angle between all variable pairs is calculated, and the standard deviation is taken as the multivariate joint deviation index of each vehicle.

[0015] Optionally, the step of calculating the linear anomaly index of each vehicle is:

[0016] The multidimensional data uploaded by each vehicle to the management platform in the last preset time window is recorded as the multidimensional data uploaded by each vehicle to the management platform in this round, and the number of uploads of the multidimensional data uploaded to the management platform in this round is obtained; a plurality of variables are extracted from the multidimensional data, and a variable value sequence of each variable changing over time is obtained according to the variable value of each variable uploaded each time;

[0017] For the variable value sequence of each variable, a trend line of each variable in this round is obtained by fitting the variable value sequence using the least squares method;

[0018] The slope and intercept of the trend line of each variable are obtained when each vehicle uploads multidimensional data in each round in the past, and the average of all slopes and intercepts is calculated as the historical slope center and the historical intercept center of each variable;

[0019] The difference degree between the trend in this round and the historical center trend is calculated according to the slope of the trend line in this round, the intercept of the trend line in this round, the historical slope center, and the historical intercept center.

[0020] The difference angle between the trend in this round and the historical center trend of each variable is subtracted from the value 1 to obtain the difference index of each variable, and the difference indexes of all variables are added to obtain the linear anomaly index.

[0021] Optionally, the step of calculating the false index of the multidimensional data uploaded by each vehicle according to the multivariate joint deviation index and the linear anomaly index of each vehicle is:

[0022] The multivariate joint deviation index and the linear anomaly index of each vehicle are normalized to remove units, and the value is mapped to the interval [0, 1], and the multivariate joint deviation index and the linear anomaly index of each vehicle after normalization are weighted and summed to obtain the false index.

[0023] Optionally, the step of judging whether the multidimensional data uploaded by each vehicle is real according to the false index is:

[0024] The false index of the multidimensional data uploaded by each vehicle is compared with the preset threshold. If the false index is less than the preset threshold, it means that the multidimensional data uploaded by the vehicle to the charging management platform is real. Then, according to the preset charging rules, the charging strategy of the vehicle is determined and the vehicle is charged according to the corresponding charging strategy.

[0025] If the false index is not less than the preset threshold, it means that the multidimensional data uploaded by the vehicle to the charging management platform is false. At this time, the charging management platform will issue an alarm to the corresponding vehicle and stop planning charging strategies for the corresponding vehicle.

[0026] In a second aspect of this invention, an intelligent charging management system for new energy commercial vehicles is proposed, the system comprising:

[0027] Variable offset module: acquires multidimensional data uploaded by each vehicle to the management platform in this round, analyzes the multidimensional data, and calculates the multivariate joint offset index for each vehicle;

[0028] Linear Anomaly Module: Extracts the historical multidimensional data uploaded by each vehicle each time, and analyzes and calculates the linear anomaly index of each vehicle based on the historical multidimensional data uploaded each time.

[0029] Judgment module: Calculates the false index of the multidimensional data uploaded by each vehicle based on the multivariate joint offset index and the linear anomaly index of each vehicle, and judges whether the multidimensional data uploaded by each vehicle is real based on the false index;

[0030] Charging management module: If the multi-dimensional data uploaded by the vehicle is true, the charging strategy of the vehicle is determined according to the preset charging rules, and the vehicle is charged according to the corresponding charging strategy.

[0031] Optionally, the variable offset module includes:

[0032] First-order difference module: Records the multidimensional data uploaded to the management platform by each vehicle in the most recent preset time window as the multidimensional data uploaded to the management platform by each vehicle in this round, and obtains the upload count of the multidimensional data uploaded to the management platform in this round; extracts multiple variables from the multidimensional data, and constructs a standardized first-order difference sequence for each variable based on the variable value uploaded each time;

[0033] Cosine angle module: for each time point At the time point Any two variables corresponding to a given time form a variable pair. Calculate the cosine of the angle between each variable pair using the following formula: In the formula, To prevent the denominator from being zero, the value is a local constant, taking the value of ; Indicates a point in time At that time, the first cosine of the angle between the first variable and the second variable, indicates the angle between the first variable and the second variable, indicates the angle between the first variable and the second variable;

[0034] The cooperative angle entropy module: for each variable pair, the cosine of the angle sequence in the entire preset time window is obtained, the cosine of the angle sequence is discretely divided into equal width intervals, the probability of each interval is counted , and the cooperative angle entropy of the variable pair is calculated , the formula is: , wherein indicates the cooperative angle entropy of the first variable and the second variable, is the preset number of discrete interval division;

[0035] The multivariate joint offset index module: the standard deviation of the cooperative angle entropy of all variable pairs is calculated, and the standard deviation is taken as the multivariate joint offset index of each vehicle.

[0036] Optionally, the linear anomaly module comprises:

[0037] The variable value sequence module: the multi-dimensional data uploaded by each vehicle to the management platform in the recent preset time window is recorded as the multi-dimensional data uploaded by each vehicle to the management platform in this round, and the number of uploads of the multi-dimensional data uploaded to the management platform in this round is obtained; and a plurality of variables are extracted from the multi-dimensional data, and the variable value sequence of each variable changing with time is obtained according to the variable value of each variable uploaded each time;

[0038] The trend line module: for the variable value sequence of each variable, the least square method is used to fit the variable value sequence to obtain the trend line of each variable in this time;

[0039] The history center module: the slope and intercept of the trend line of each variable when the multi-dimensional data of each vehicle is uploaded in each round in the history are obtained, and the mean value of all slopes and intercepts is calculated as the historical slope center and the historical intercept center of each variable;

[0040] The difference angle module: the difference angle between the trend line in this round and the historical center trend line is calculated according to the slope of the trend line in this round, the intercept of the trend line in this round, the historical slope center and the historical intercept center;

[0041] The linear anomaly index module: the difference value index of each variable is obtained by subtracting the difference angle between the trend line in this round and the historical center trend line from the value 1, and the linear anomaly index is obtained by adding the difference indexes of all variables.

[0042] ​​​​Optionally, the judging module is applied as:

[0043] The multivariate joint offset index and the linear anomaly index of each vehicle are normalized, units are removed, and the values are mapped to the 0-1 interval, and the multivariate joint offset index and the linear anomaly index of each vehicle after normalization are weighted and summed to obtain a false index.

[0044] Optionally, the charging management module further comprises:

[0045] The first management module: compares the false index of the multi-dimensional data uploaded by each vehicle with the preset threshold value, if the false index is less than the preset threshold value, it indicates that the multi-dimensional data uploaded by the vehicle to the charging management platform is real, then according to the preset charging rule, the charging strategy of the vehicle is determined, and the vehicle is charged according to the corresponding charging strategy;

[0046] The second management module: if the false index is not less than the preset threshold value, it indicates that the multi-dimensional data uploaded by the vehicle to the charging management platform is false, at this time the charging management platform sends an alarm to the corresponding vehicle, and stops the charging strategy planning for the corresponding vehicle.

[0047] The beneficial effects of the present application are:

[0048] The present application provides an intelligent charging management method and system for new energy commercial vehicles, by obtaining the multi-dimensional data uploaded by each vehicle to the management platform, analyzing the multi-dimensional data, calculating the multivariate joint offset index of each vehicle, and evaluating the deviation coupling degree between the multi-dimensional uploaded data parameters; and extracting the multi-dimensional data uploaded by each vehicle in each history, and analyzing and calculating the linear anomaly index of each vehicle based on the multi-dimensional data uploaded in each history, and evaluating the abnormal degree of the multi-dimensional data uploaded by the vehicle to the management platform compared with the historical upload; according to the multivariate joint offset index and the linear anomaly index of each vehicle, the false index of the multi-dimensional data uploaded by each vehicle is calculated, and whether the multi-dimensional data uploaded by each vehicle is real is judged according to the false index; if the multi-dimensional data uploaded by the vehicle is real, the charging strategy of the vehicle is determined according to the preset charging rule, and the vehicle is charged according to the corresponding charging strategy; in this way, the management platform can judge the authenticity of the data uploaded by the new energy commercial vehicle, when the uploaded data is real, the management platform can generate the corresponding charging strategy based on the real data of the new energy commercial vehicle, and ensure that the whole new energy commercial vehicle fleet can perfectly execute the operation plan, and ensure the scheduling efficiency and overall charging balance of the whole fleet. BRIEF DESCRIPTION OF DRAWINGS

[0049] The present application will be further described below in conjunction with the drawings.

[0050] Figure 1 It is a flowchart of an intelligent charging management method for new energy commercial vehicles;

[0051] Figure 2 A framework diagram of an intelligent charging management system for new energy commercial vehicles. DETAILED DESCRIPTION

[0052] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative labor fall within the scope of protection of the present application.

[0053] The embodiments of the present application provide an intelligent charging management method for new energy commercial vehicles. Figure 1 , Figure 1 A flowchart of an intelligent charging management method for new energy commercial vehicles provided by the embodiments of the present application. The method comprises the following steps:

[0054] S1: Obtain multi-dimensional data uploaded by each vehicle to the management platform in this round, and analyze the multi-dimensional data to calculate a multi-variable joint offset index of each vehicle, which is used to evaluate the deviation coupling degree between the multi-dimensional uploaded data parameters;

[0055] S2: Extract multi-dimensional data uploaded by each vehicle in each round in the past, and analyze and calculate a linear abnormality index of each vehicle based on the multi-dimensional data uploaded in each round in the past, to evaluate the abnormality degree of the multi-dimensional data uploaded by the vehicle to the management platform in this round compared with the historical uploaded data;

[0056] S3: Calculate a false index of the multi-dimensional data uploaded by each vehicle according to the multi-variable joint offset index and the linear abnormality index of each vehicle, and determine whether the multi-dimensional data uploaded by each vehicle is real according to the false index;

[0057] S4: If the multi-dimensional data uploaded by the vehicle is real, determine a charging strategy of the vehicle according to a preset charging rule, and the vehicle charges according to the corresponding charging strategy.

[0058] Based on the intelligent charging management method for new energy commercial vehicles provided by the embodiments of the present application, through the above manner, the management platform can judge the authenticity of the data uploaded by the new energy commercial vehicles. When the uploaded data is real, the management platform can generate a corresponding charging strategy based on the real data of the new energy commercial vehicles, so as to ensure that the overall new energy commercial vehicle fleet can perfectly execute the operation plan, and ensure the scheduling efficiency of the overall fleet and the overall charging balance resource mismatch.

[0059] In one embodiment, S1: for each vehicle uploaded multi-dimensional data, and the multi-dimensional data is analyzed, the multivariate joint offset index of each vehicle is calculated, which is used to evaluate the deviation coupling degree between the parameters of the multi-dimensional uploaded data;

[0060] The multi-dimensional data uploaded by the new energy commercial vehicle to the charging management The multi-dimensional data of the platform includes but is not limited to battery SOC, voltage, current, temperature, remaining endurance and data related to preset charging rules. These multi-dimensional data jointly determine the battery state of the new energy commercial vehicle. The charging management platform extracts data related to the final charging sequence from these multi-dimensional data to determine the charging sequence of the new energy commercial vehicle. The specific multi-dimensional data is set according to the actual situation and is not limited or elaborated.

[0061] It should be noted that the multivariate joint deviation index of the vehicle refers to a numerical index for measuring whether the current data deviates from the "normal physical behavior coupling mode" based on the joint distribution relationship between the multi-dimensional battery state data (such as SOC, voltage, current, temperature, internal resistance, etc.) uploaded by the new energy commercial vehicle to the charging management platform. The core use of the index is to evaluate whether there is a non-coordinated, abnormal drift or suspicious pattern deliberately constructed between the current uploaded data in multiple dimensions, so as to judge its authenticity. Under normal circumstances, the SOC, voltage and current parameters of the vehicle should present a certain physical coupling relationship under certain temperature and internal resistance conditions, for example, when the SOC is low, the voltage should tend to decrease, and the increase of internal resistance is usually accompanied by the decrease of charging efficiency, the increase of battery temperature and other phenomena. However, if the charging management platform finds that the SOC uploaded by a vehicle is very low (such as 15%), but at the same time, the total voltage, current fluctuation, battery temperature and other parameters are in an irregular range (for example, high voltage, very small internal resistance, low temperature), this uncoordinated combination may cause the joint deviation index to rise. The larger the multivariate joint deviation index, the farther the current data combination deviates from the coupling relationship between the statistical structure or physical logic and the historical behavior of the vehicle itself and the regularity of the same type of vehicle group, thereby indicating that the data may have problems such as forgery, tampering, abnormal collection or upload delay. For example: a vehicle hopes to obtain priority fast charging resources, and uploads false "12%" low power information when the actual SOC is about 40%, but does not modify the coupled temperature, voltage and other data at the same time, causing the charging management platform to judge that the uploaded information has a high inconsistency, thereby significantly increasing the multivariate joint deviation index. If the charging management platform still blindly trusts the "false low power" information uploaded by the vehicle without identifying such deviation, it may mistakenly prioritize the vehicle in emergency charging scheduling, causing other vehicles that are truly low on power and about to leave to be delayed in scheduling, and even miss the operation window. This will directly undermine the fairness of the original scheduling priority, reduce the overall charging resource utilization efficiency, affect the vehicle task matching rhythm, and in severe cases, cause operation risks such as distribution interruption, bus off-duty, and key node disconnection. Therefore, the introduction of the multivariate joint deviation index not only helps to reveal the authenticity of the uploaded data, but also is an important support means for building the robustness and fairness of the intelligent scheduling charging management platform.

[0062] Specifically, in one implementation, the step of calculating the multivariate joint deviation index of each vehicle is:

[0063] The multi-dimensional data uploaded by each vehicle to the management platform in the latest preset time window is denoted as the multi-dimensional data uploaded by each vehicle to the management platform in the current round, and the number of uploads of the multi-dimensional data uploaded to the management platform in the current round is obtained; and a plurality of variables are extracted from the multi-dimensional data, and a standardized first-order difference sequence of each variable is constructed according to the variable value of each variable uploaded each time. The specific calculation formula is: In the formula, , represents the index value corresponding to the maximum difference. This indicates the total number of times vehicles upload multidimensional data within a preset time window. For the first The variables at time... The reported value, Such as SOC, voltage, current, temperature, etc.; For the first The variables at time... The standardized first-order difference; To prevent the denominator from being zero, it is generally a local constant. Values ;

[0064] For each time point At the time point Any two variables corresponding to a given time form a variable pair. Calculate the cosine of the angle between each variable pair using the following formula: In the formula, To prevent the denominator from being zero, it is generally a local constant. Values ; Indicates a point in time At that time, the first The first variable and the second The cosine value of the included angle between the variables, Indicates the first The first variable and the second The angle between the changing trends of the variables;

[0065] For each pair of variables, obtain the sequence of cosine values ​​of the included angle over the entire preset time window. Discretize the sequence of cosine values ​​of the included angle into Given several equal-width intervals, calculate the probability of each interval occurring. And calculate the cooperative angle entropy of the variable pairs. This reflects the degree of consistency and disorder in its changing trends; the calculation formula is: In the formula, Indicates the first The first variable and the second The cooperative angle entropy of each variable, This is the preset number of discrete intervals, typically set to 10. Represents variable pairs The cosine value of the included angle falls into the first The probability of each interval;

[0066] Calculate the standard deviation of the cooperative angle entropy for all variable pairs, and use the standard deviation as the multivariate joint offset index for each vehicle.

[0067] It should be noted that in the calculation process of the above multivariate joint offset index, all kinds of data involved are derived from the multi-dimensional data of new energy commercial vehicle uploaded to the charging management platform in real time or periodically during operation; these data are automatically collected by the vehicle terminal through the Internet of Vehicles communication module and uploaded at a set frequency; the data acquisition process is usually based on a unified data transmission protocol, and the platform receives the data and sorts and stores them according to vehicle ID, time stamp, etc. The charging management platform will retrieve the multi-dimensional data records corresponding to a vehicle in a recent preset time window from the database, and extract the fields according to the variable name (such as SOC, voltage, current, temperature, etc.) and align the time sequence to form a set of structured variable time sequence data for calculation. In addition, the preset time window is usually set to 5-30 minutes or the last 10-30 upload cycles, depending on the vehicle upload frequency and the real-time demand of scheduling decisions;

[0068] The original charging management platform needs to ensure

[0069] It should be noted that the multivariate joint offset index of each vehicle is calculated in the above manner because this method can accurately reflect the stability and consistency of the inherent coupling relationship between each physical variable in the vehicle upload data from the perspective of dynamic evolution. Under normal working conditions, such as SOC drop usually accompanied by current increase, power fluctuation, temperature change, etc., there is a strong co-evolution rule between these variables. The use of standardized first-order difference can eliminate the dimension effect, only focusing on the "directionality" of the variable change trend, and then capturing the co-variation direction of each pair of variables at a certain moment by constructing the cosine of the angle between the differences. Further entropy analysis of the cosine sequence at multiple moments can determine whether the co-trend of the variable pair is stable, whether there is chaos or human disturbance (such as false reports, data fraud, etc.) - the higher the co-entropy, the more unstable the co-trend, and the more abnormal information. Finally, by calculating the standard deviation of the co-entropy of all variable pairs, the joint offset index can accurately reflect whether the overall variable co-relationship has been destroyed. Compared with traditional mean, variance, simple deviation, etc., this method can more sensitively identify hidden data inconsistency.

[0070] In one embodiment, S2: extract the historical multi-dimensional data uploaded by each vehicle each time, and analyze and calculate the linear abnormality index of each vehicle based on the historical multi-dimensional data uploaded each time to evaluate the abnormality degree of the multi-dimensional data uploaded by the vehicle to the management platform this time compared with the historical upload;

[0071] In one implementation, the step of calculating the linear abnormality index of each vehicle is:

[0072] The multidimensional data uploaded to the management platform by each vehicle in the most recent preset time window is recorded as the multidimensional data uploaded to the management platform by each vehicle in this round, and the upload count of the multidimensional data uploaded to the management platform in this round is obtained; multiple variables are extracted from the multidimensional data, and the variable value sequence of each variable changing over time is obtained based on the variable value uploaded each time;

[0073] For the variable value sequence of each variable, the least squares method is used to fit the variable value sequence to obtain the trend line of each variable in this case; In the formula, It is the first The trend line represents the current variables. It is the first The slope of the current trend line is one variable. No. The intercept of the current trend line for each variable;

[0074] Obtain the slope and intercept of the trend line for each variable during each historical upload of multidimensional data for each vehicle, and calculate the mean of all slopes and intercepts as the historical slope center for each variable. and historical intercept center ;

[0075] Based on the slope of the trend line for each variable The intercept of this trend line Historical slope center and historical intercept center The formula for calculating the difference between the current trend and the historical central trend is as follows: In the formula, For variables The difference between the current trend and the historical central trend; It is to prevent division by zero for extremely small positive numbers, generally Values ;generally The smaller the value (closer to 1), the more consistent the trend of the variable in this cycle is with the historical trend;

[0076] Subtract the variable from the value 1 By analyzing the difference between the current trend and the historical central trend, we obtain the difference index of each variable. By summing the difference indices of all variables, we obtain the linear anomaly index.

[0077] It should be noted that the data acquisition method involved in the above calculation process mainly depends on the real-time communication mechanism between the vehicle and the charging management platform; each vehicle will continuously upload multi-dimensional state data to the platform at a fixed frequency within a preset time window during the time period approaching the need for charging. These multi-dimensional data usually include: SOC (state of charge), current, voltage, power, battery temperature, charging status code and other key operating parameters. The platform will cache these data in real time to form the data sequence of "this upload". At the same time, the platform will also retrieve similar upload data records corresponding to each charging of the vehicle from the database in the past, and construct the variable change curve corresponding to each upload to extract the trend line (slope, intercept) of each variable in the "history upload". These data do not depend on additional sensors, but are based on the existing vehicle upload record database of the platform for backtracking and extraction, and have integrity.

[0078] It should be noted that the linear anomaly index of the vehicle is a consistency index for measuring whether the multi-dimensional time series data uploaded by the vehicle this time conforms to the historical uploading behavior trend of the vehicle. It is based on modeling the linear change trend (i.e. slope and intercept) of each variable value uploaded continuously by the vehicle before each charging, and comparing the trend of the current uploading with the "trend center model" formed in the past uploading behavior of the vehicle. If the consistency of the current variable trend and the historical trend center in direction (slope) and position (intercept) is high, it means that the data uploaded this time continues the consistent behavior pattern of the vehicle, and is most likely real and naturally generated data. On the contrary, if the difference is significant, it means that the data uploaded by the vehicle this time is obviously inconsistent with the historical uploading behavior, and there is a possibility of data tampering, fabrication, splicing, tampering of uploading time sequence, device anomaly, etc. For example, the SOC (state of charge of battery) of a vehicle is a linear curve (such as the slope is between 0.02~0.03) that slowly rises or falls each time the data is uploaded in the past, but in this uploading, the SOC suddenly "jumps" up or down, or fluctuates up and down sharply in a short time, and the fitting trend line slope may deviate from the historical center trend a lot, indicating that this group of data no longer presents the characteristics of continuous and natural evolution. Similarly, if the current and voltage variables were originally stable, but the current uploading shows an abnormal straight line rise / fall, or an abnormal flatness (i.e. the slope is close to 0), it also indicates that the uploading data may have an abnormal generation method. The larger the linear anomaly index, the greater the deviation of the current uploading data from the historical trend, that is, the more "unreliable" the data. This is crucial for the new energy commercial vehicle charging management platform. Because the platform often needs to predict whether the vehicle needs to be charged, whether the battery is healthy, whether the current running state is normal, etc. based on these data, and then make key operation strategies such as charging sorting, queuing scheduling, and power resource allocation. Once the uploaded data is falsified or distorted, for example, a vehicle deliberately uploads a "strong fast charging demand" state (such as a low SOC and an abnormal current state), the platform may mistakenly prioritize the charging resources for it, while ignoring vehicles with real low battery levels.

[0079] It should be noted that the core advantage of calculating the linear anomaly index of each vehicle in the above manner is that by fitting a trend line (slope and intercept) of each variable changing over time through the least square method, and then comparing it with the trend center of the historical behavior of the vehicle (the average of the historical slope and intercept), the degree of abnormality of the current upload data is accurately evaluated from the deviation of the linear change trend. Compared with the traditional volatility analysis or single-point outlier detection method, this method does not depend on the specific variable value, but focuses on the consistency of the variable evolution direction, and has stronger interpretability, universality and stability. Based on the similarity judgment of the trend angle, the influence of individual noise points on the judgment result can be resisted, and the sensitivity to the overall behavior pattern change is improved; and the method structure is simple and clear, and does not depend on weight or model training, which is suitable for large-scale evaluation of multiple vehicles in multiple scenes, and effectively identifies the authenticity deviation of the upload data.

[0080] In one embodiment, S3: calculating a false index of the multi-dimensional data uploaded by each vehicle according to the multivariate joint offset index and the linear anomaly index of each vehicle, and judging whether the multi-dimensional data uploaded by each vehicle is real according to the false index;

[0081] In one implementation, the step of calculating the false index of the multi-dimensional data uploaded by each vehicle according to the multivariate joint offset index and the linear anomaly index of each vehicle is:

[0082] The multivariate joint offset index and the linear anomaly index of each vehicle are normalized to remove the unit, and the values are mapped to the 0-1 interval, and the multivariate joint offset index and the linear anomaly index of each vehicle after normalization are weighted and summed to obtain the false index; the formula for calculation is: , in the formula is the false index, are the multivariate joint offset index and the linear anomaly index after normalization, respectively, respectively represent the preset weight coefficients of the multivariate joint offset index and the linear anomaly index of each vehicle after normalization, and are both greater than 0;

[0083] It should be noted that the commonly used dimension removal methods include Min-Max normalization, Z-Score standardization, etc., which will not be repeated here; According to the actual situation, generally are equal and the sum is 1, for example, both can be 0.5, 0.5.

[0084] In one embodiment, the step of judging whether the multi-dimensional data uploaded by each vehicle is real according to the false index is:

[0085] The false index of the multi-dimensional data uploaded by each vehicle is compared with the preset threshold value, if the false index is less than the preset threshold value, it indicates that the multi-dimensional data uploaded by the vehicle to the charging management platform is real, and the charging strategy of the vehicle is determined according to the preset charging rule, and the vehicle charges according to the corresponding charging strategy;

[0086] If the false index is not less than the preset threshold value, it indicates that the multi-dimensional data uploaded by the vehicle to the charging management platform is false, at this time, the charging management platform issues an alarm to the corresponding vehicle, and stops planning the charging strategy for the corresponding vehicle.

[0087] It should be noted that when the false index of the vehicle is less than the preset threshold value, it means that the uploaded data is consistent with the historical uploaded data in trend, structure or behavior mode, and the charging management platform will determine that the data is real and reliable, so that the platform can formulate a specific charging strategy based on the current data, combined with the task scheduling rule (such as charging priority, remaining power, etc.), for the vehicle to execute. When the false index of a vehicle is greater than or equal to the threshold value, it indicates that the data may have problems such as forgery, splicing, non-real collection (such as uploading SOC, voltage, etc. unrelated to the actual state), and the platform will trigger the early warning mechanism to issue a data authenticity risk warning to the vehicle, and suspend the charging strategy calculation and resource allocation for it, in order to avoid resource waste, scheduling confusion, and even operational risks. For example, a vehicle A continuously uploads 30 times of SOC, voltage, current, temperature and other data within a preset time window, and the system calculates its linear abnormal index as 0.82, which exceeds the platform set threshold value of 0.75, indicating that the variable change trend of the vehicle is inconsistent with the historical charging period, and the data may be fake. The platform immediately issues an alarm and removes it from the current charging queue. While another vehicle B has a false index of 0.21, which is far below the threshold value, the platform will arrange it in the priority queue, and arrange appropriate charging time and strategy according to the remaining power, location distance and other factors. This mechanism not only improves the scheduling accuracy, but also strengthens the platform's governance ability of abnormal behavior.

[0088] Based on the same inventive concept, the embodiments of the present application also provide an intelligent charging management system for new energy commercial vehicles. Referring to Figure 2 , Figure 2 The embodiments of the present application provide a framework diagram of an intelligent charging management system for new energy commercial vehicles, which comprises:

[0089] Variable offset module: obtaining the multi-dimensional data uploaded by each vehicle to the management platform in this round, and analyzing the multi-dimensional data to calculate the multi-variable joint offset index of each vehicle;

[0090] Linear abnormality module: extracting the multi-dimensional data uploaded by each vehicle in each historical round, and analyzing and calculating the linear abnormality index of each vehicle based on the multi-dimensional data uploaded in each historical round;

[0091] The judging module calculates a false index of the multi-dimensional data uploaded by each vehicle according to the multivariate joint deviation index and the linear anomaly index of each vehicle, and judges whether the multi-dimensional data uploaded by each vehicle is real according to the false index;

[0092] The charging management module determines a charging strategy of the vehicle according to a preset charging rule if the multi-dimensional data uploaded by the vehicle is real, and the vehicle charges according to the corresponding charging strategy.

[0093] According to the intelligent charging management system for new energy commercial vehicles provided by the embodiment of the application, the management platform can judge the authenticity of the data uploaded by the new energy commercial vehicles, and when the uploaded data is real, the management platform can generate a corresponding charging strategy based on the real data of the new energy commercial vehicles, so that the overall new energy commercial vehicle fleet can perfectly implement the operation plan, and the scheduling efficiency of the overall fleet and the overall charging balance resource mismatch can be ensured.

[0094] In one embodiment, the variable deviation module comprises:

[0095] The first-order difference module records the multi-dimensional data uploaded by each vehicle to the management platform in the latest preset time window as the multi-dimensional data uploaded by each vehicle to the management platform in this round, and obtains the uploading times of the multi-dimensional data uploaded to the management platform in this round; and extracts a plurality of variables from the multi-dimensional data, and constructs a standardized first-order difference sequence of each variable according to the variable value uploaded each time.

[0096] The cosine value module calculates the cosine value of each variable pair for each time point , and the formula is: , wherein is a minimum constant value to prevent the denominator from being 0, and the value is ; represents the cosine value of the first variable and the second variable at time point , and the formula is: , wherein represents the cosine value of the first variable and the second variable at time point , and the formula is: , wherein represents the cosine value of the first variable and the second variable at time point , and the formula is:

[0097] The cooperative cosine entropy module obtains the cosine value sequence of each variable pair in the entire preset time window, discretely divides the cosine value sequence into equal-width intervals, calculates the probability of each interval, and calculates the cooperative cosine entropy of the variable pair, and the formula is: , wherein, represents the co-angle entropy of the jth variable and the kth variable, is a preset number of partition discrete intervals;

[0098] Multivariate joint offset index module: calculate the standard deviation of the co-angle entropy of all variable pairs, and take the standard deviation as the multivariate joint offset index of each vehicle.

[0099] In one embodiment, the linear anomaly module comprises:

[0100] Variable value sequence module: record the multi-dimensional data uploaded by each vehicle in the current round to the management platform as the multi-dimensional data uploaded by each vehicle in the current round to the management platform, and obtain the number of uploads of the multi-dimensional data uploaded in the current round to the management platform; and extract a plurality of variables from the multi-dimensional data, and obtain the variable value sequence of each variable with respect to time according to the variable value of each variable uploaded each time;

[0101] Trend line module: for the variable value sequence of each variable, fit the variable value sequence by the least square method to obtain the trend line of each variable in the current round;

[0102] History center module: obtain the slope and intercept of the trend line of each variable when each vehicle uploads the multi-dimensional data in each round in the past, and calculate the average of all slopes and intercepts as the historical slope center and the historical intercept center of each variable;

[0103] Difference angle module: calculate the difference angle between the trend line in the current round and the historical center trend line according to the slope of the trend line in the current round, the intercept of the trend line in the current round, the historical slope center and the historical intercept center;

[0104] Linear anomaly index module: subtract the difference angle between the trend line in the current round and the historical center trend line from the value 1 to obtain the difference index of each variable, and add the difference indexes of all variables to obtain the linear anomaly index.

[0105] In one embodiment, the judgment module is applied as:

[0106] Normalize the multivariate joint offset index and the linear anomaly index of each vehicle to remove the unit, map the value to the interval of 0-1, and weightedly sum the normalized multivariate joint offset index and the linear anomaly index of each vehicle to obtain the false index.

[0107] In one embodiment, the charging management module further comprises:

[0108] ​​The first management module: compare the false index of the multi-dimensional data uploaded by each vehicle with the preset threshold value, if the false index is less than the preset threshold value, it indicates that the multi-dimensional data uploaded by the vehicle to the charging management platform is real, then according to the preset charging rule, the charging strategy of the vehicle is determined, and the vehicle charges according to the corresponding charging strategy;

[0109] The second management module: if the false index is not less than the preset threshold value, it indicates that the multi-dimensional data uploaded by the vehicle to the charging management platform is false, at this time the charging management platform issues an alarm to the corresponding vehicle and stops the charging strategy planning for the corresponding vehicle.

[0110] The above describes one embodiment of the present application in detail, but the content is only the preferred embodiment of the present application, and cannot be considered as limiting the scope of the present application. Any equivalent changes and improvements made according to the scope of the present application should still belong to the scope of the present application.

Claims

1. An intelligent charging management method for new energy commercial vehicles, characterized in that, The method comprises the following steps: Obtain the multi-dimensional data uploaded by each vehicle to the management platform in this round, and analyze the multi-dimensional data to calculate the multi-variable joint deviation index of each vehicle; Extract the multi-dimensional data uploaded by each vehicle in each round in the past, and analyze and calculate the linear abnormality index of each vehicle; Calculate the false index of the multi-dimensional data uploaded by each vehicle according to the multi-variable joint deviation index and the linear abnormality index of each vehicle, and judge whether the multi-dimensional data uploaded by each vehicle is real according to the false index; If the multi-dimensional data uploaded by the vehicle is real, determine the charging strategy of the vehicle according to the preset charging rule, and the vehicle charges according to the corresponding charging strategy; The step of calculating the multi-variable joint deviation index of each vehicle is: Record the multi-dimensional data uploaded by each vehicle to the management platform in the latest preset time window as the multi-dimensional data uploaded by each vehicle to the management platform in this round, and obtain the upload times of the multi-dimensional data uploaded to the management platform in this round; Extract a plurality of variables from the multi-dimensional data, and construct the standardized first-order difference sequence of each variable according to the variable value uploaded by each variable each time; For each time point At the time point Any two variables corresponding to a given time form a variable pair. Calculate the cosine of the angle between each variable pair using the following formula: In the formula, To prevent the denominator from being zero, the value is a local constant, taking the value of ; For the first The variables at time... The standardized first-order difference; For the first The variables at time... The standardized first-order difference; Indicates a point in time At that time, the first The first variable and the second The cosine value of the included angle between the variables, Indicates the first The first variable and the second The angle between the changing trends of the variables; For each pair of variables, the sequence of the cosine of the angle is obtained in the whole preset time window, the sequence of the cosine of the angle is discretely divided into equal-width intervals, the probability of occurrence of each interval is counted , and the collaborative cosine entropy of the pair of variables is calculated , and the formula is: , wherein represents the collaborative cosine entropy of the th variable and the th variable, is the preset number of discrete interval division; Calculate the standard deviation of the synergistic included angle entropy of all variable pairs, and take the standard deviation as the multi-variable joint deviation index of each vehicle. 2.The intelligent charging management method for new energy commercial vehicle of claim 1, wherein, The step of calculating the linear abnormality index of each vehicle is: Record the multi-dimensional data uploaded by each vehicle to the management platform in the latest preset time window as the multi-dimensional data uploaded by each vehicle to the management platform in this round, and obtain the upload times of the multi-dimensional data uploaded to the management platform in this round; Extract a plurality of variables from the multi-dimensional data, and obtain the variable value sequence of each variable changing with time according to the variable value uploaded by each variable each time; For the variable value sequence of each variable, fit the variable value sequence by the least square method to obtain the trend line of each variable in this time; Obtain the slope and intercept of the trend line of each variable when each vehicle uploads the multi-dimensional data in each round in the past, and calculate the mean of all slopes and intercepts as the historical slope center and historical intercept center of each variable; Calculate the difference degree of the trend in this round and the historical center trend according to the slope of the trend line in this round, the intercept of the trend line in this round, the historical slope center and the historical intercept center; Subtract the difference angle of the trend in this round and the historical center trend from the value 1 to obtain the difference index of each variable, and add the difference indexes of all variables to obtain the linear abnormality index. 3.The intelligent charging management method for new energy commercial vehicle of claim 1, wherein, The step of calculating the false index of the multi-dimensional data uploaded by each vehicle according to the multi-variable joint deviation index and the linear abnormality index of each vehicle is: Normalize the multi-variable joint deviation index and the linear abnormality index of each vehicle to remove the unit, map the value to the interval of 0-1, and weightedly sum the normalized multi-variable joint deviation index and the linear abnormality index of each vehicle to obtain the false index. 4.The intelligent charging management method for new energy commercial vehicle of claim 1, wherein, The step of judging whether the multi-dimensional data uploaded by each vehicle is real according to the false index is: Compare the false index of the multi-dimensional data uploaded by each vehicle with the preset threshold value, if the false index is less than the preset threshold value, it indicates that the multi-dimensional data uploaded by the vehicle to the charging management platform is real, then determine the charging strategy of the vehicle according to the preset charging rule, and the vehicle charges according to the corresponding charging strategy; If the false index is not less than the preset threshold, it indicates that the multi-dimensional data uploaded by the vehicle to the charging management platform is false, at this time the charging management platform sends an alarm to the corresponding vehicle and stops the charging strategy planning for the corresponding vehicle.

5. An intelligent charging management system for new energy commercial vehicles, characterized in that, The system comprises: a variable offset module: obtaining the multi-dimensional data uploaded by each vehicle to the management platform in this round, and analyzing the multi-dimensional data to calculate the multi-variable joint offset index of each vehicle; a linear anomaly module: extracting the multi-dimensional data uploaded by each vehicle in each historical round, and analyzing and calculating the linear anomaly index of each vehicle; a judgment module: calculating the false index of the multi-dimensional data uploaded by each vehicle according to the multi-variable joint offset index and the linear anomaly index of each vehicle, and judging whether the multi-dimensional data uploaded by each vehicle is real according to the false index; a charging management module: if the multi-dimensional data uploaded by the vehicle is real, determining the charging strategy of the vehicle according to the preset charging rule, and the vehicle charges according to the corresponding charging strategy; The variable offset module comprises: a first-order difference module: recording the multi-dimensional data uploaded by each vehicle to the management platform in the recent preset time window as the multi-dimensional data uploaded by each vehicle to the management platform in this round, and obtaining the upload times of the multi-dimensional data uploaded to the management platform in this round; and extracting a plurality of variables from the multi-dimensional data, and constructing a standardized first-order difference sequence of each variable according to the variable value of each variable uploaded each time; Cosine angle module: for each time point At the time point Any two variables corresponding to a given time form a variable pair. Calculate the cosine of the angle between each variable pair using the following formula: In the formula, To prevent the denominator from being zero, the value is a local constant, taking the value of ; For the first The variables at time... The standardized first-order difference; For the first The variables at time... The standardized first-order difference; Indicates a point in time At that time, the first The first variable and the second The cosine value of the included angle between the variables, Indicates the first The first variable and the second The angle between the changing trends of the variables; The cooperative included angle entropy module: for each variable pair, the included angle cosine value sequence in the whole preset time window is obtained, the included angle cosine value sequence is discretely divided into equal-width intervals, the probability of occurrence of each interval is counted , and the cooperative included angle entropy of the variable pair is calculated , and the formula is: , wherein, represents the cooperative included angle entropy of the th variable and the th variable, is the preset number of divided discrete intervals; a multi-variable joint offset index module: calculating the standard deviation of the cooperative included angle entropy of all variable pairs, and taking the standard deviation as the multi-variable joint offset index of each vehicle. 6.The intelligent charging management system for new energy commercial vehicle of claim 5, wherein, The linear anomaly module comprises: a variable value sequence module: recording the multi-dimensional data uploaded by each vehicle to the management platform in the recent preset time window as the multi-dimensional data uploaded by each vehicle to the management platform in this round, and obtaining the upload times of the multi-dimensional data uploaded to the management platform in this round; and extracting a plurality of variables from the multi-dimensional data, and obtaining the variable value sequence of each variable changing with time according to the variable value of each variable uploaded each time; a trend line module: for the variable value sequence of each variable, fitting the variable value sequence by the least square method to obtain the trend line of each variable in this time; a historical center module: obtaining the slope and intercept of the trend line of each variable when each vehicle uploads the multi-dimensional data in each historical round, and calculating the mean of all slopes and intercepts as the historical slope center and historical intercept center of each variable; a difference angle module: calculating the difference angle between the trend line in this round and the historical center trend line according to the slope of the trend line in this round, the intercept of the trend line in this round, the historical slope center and the historical intercept center; a linear anomaly index module: subtracting the difference angle between the trend line in this round and the historical center trend line from 1 to obtain the difference index of each variable, and adding the difference indexes of all variables to obtain the linear anomaly index. 7.The intelligent charging management system for new energy commercial vehicle of claim 5, wherein, The judgment module is applied as: normalizing the multi-variable joint offset index and the linear anomaly index of each vehicle to remove the unit, mapping the value to the interval of 0-1, and weightedly summing the normalized multi-variable joint offset index and the linear anomaly index of each vehicle to obtain the false index. 8.The intelligent charging management system for new energy commercial vehicle of claim 5, wherein, The charging management module further comprises: The first management module compares the false index of the multi-dimensional data uploaded by each vehicle with a preset threshold value, if the false index is less than the preset threshold value, it indicates that the multi-dimensional data uploaded by the vehicle to the charging management platform is real, then according to the preset charging rule, the charging strategy of the vehicle is determined, and the vehicle charges according to the corresponding charging strategy; the second management module: if the false index is not less than the preset threshold value, it indicates that the multi-dimensional data uploaded by the vehicle to the charging management platform is false, at this time the charging management platform sends an alarm to the corresponding vehicle, and stops the charging strategy planning for the corresponding vehicle.

Images