Enterprise vehicle energy consumption and energy saving oriented charging path planning method and system

Abstract

The invention provides a charging path planning method and system for energy consumption and energy conservation of enterprise vehicles, and relates to the technical field of vehicle operation management.The method comprises the steps that firstly, a multi-source operation data set, including vehicle real-time state data, environment and map data and historical and cost data, of the enterprise vehicles is obtained; then unit power consumption prediction processing is executed based on the multi-source operation data set, predicted unit power consumption of the enterprise vehicle under the current operation condition is obtained, a path diagram is constructed based on the predicted unit power consumption, and the comprehensive cost of going to a nearby charging station and returning to a fixed station charging station is calculated; and comparing and determining a charging path optimization strategy with a better energy consumption and saving effect, calculating the energy consumption saving amount, generating a charging path instruction, associating the charging path instruction with the energy consumption saving amount information, and outputting the information to the enterprise vehicle management terminal, thereby effectively reducing the energy consumption and operation cost of the enterprise vehicle.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of vehicle operation management, in particular to a charging path planning method and system for energy saving of enterprise vehicle energy consumption. BACKGROUND

[0002] In the field of enterprise vehicle operation management, vehicle energy consumption control and charging planning are key links to improve operation efficiency and reduce cost. At present, enterprises mostly use a relatively simple way to plan the charging path of vehicles.

[0003] On the one hand, some enterprises only choose charging stations according to the current power of the vehicle and the principle of proximity, without considering the influence of real-time state of the vehicle, operating environment and historical data on energy consumption. For example, the unit power consumption is significantly different when the vehicle drives in different road conditions (such as congested urban roads or smooth highways), but the traditional planning method does not take such factors into account.

[0004] On the other hand, the existing planning method lacks comprehensive cost evaluation for going to different charging stations. In the process of operation, enterprise vehicles going to different charging stations not only involves power consumption caused by driving distance, but also may produce different comprehensive costs due to service fees, charging efficiency and other factors of charging stations. However, the traditional method fails to comprehensively calculate these factors, resulting in that the selected charging path may not be the optimal solution in terms of energy consumption, thereby increasing the operation cost of enterprises. SUMMARY

[0005] In view of the above-mentioned problems, in combination with the first aspect of the present application, the embodiments of the present application provide a charging path planning method for energy saving of enterprise vehicle energy consumption, which comprises: obtaining a multi-source operation data set of an enterprise vehicle, the multi-source operation data set comprising vehicle real-time state data, environment and map data, and historical and cost data; performing unit power consumption prediction processing based on the multi-source operation data set to obtain the predicted unit power consumption of the enterprise vehicle under the current operating conditions; constructing a path graph comprising candidate charging station nodes and the current position node of the vehicle, and calculating the first comprehensive cost of going to the nearest charging station and the second comprehensive cost of returning to the fixed station charging station based on the predicted unit power consumption; comparing the first comprehensive cost and the second comprehensive cost to determine the charging path optimization strategy with better energy saving effect, and calculating the energy saving amount information of the charging path optimization strategy relative to another strategy; based on the charging path optimization strategy and the energy saving amount information, generating a charging path instruction comprising a charging path node sequence, and outputting the charging path instruction and the energy saving amount information to an enterprise vehicle management terminal.

[0006] In still another aspect, the embodiments of the present application also provide a charging path planning system for energy saving of enterprise vehicle energy consumption, characterized in comprising: a processor; a machine readable storage medium for storing machine executable instructions of the processor; wherein the processor is configured to execute the charging path planning method for energy saving of enterprise vehicle energy consumption by executing the machine executable instructions.

[0007] Based on the above aspects, by acquiring a multi-source operation data set of the enterprise vehicle, covering vehicle real-time state data, environment and map data, and historical and cost data, performing unit power consumption prediction processing based on the multi-source operation data set, the predicted unit power consumption of the enterprise vehicle under the current operation condition can be accurately obtained, fully considering the influence of the vehicle real-time state and operation environment on energy consumption, then a path graph containing candidate charging station nodes and vehicle current position nodes is constructed, and the comprehensive cost of going to the nearest charging station and returning to the fixed charging station is calculated based on the predicted unit power consumption, the charging path optimization strategy with better energy saving effect is determined by comparing the comprehensive cost, and the energy saving amount information relative to another strategy is calculated, thereby selecting the optimal charging path for the enterprise vehicle, effectively reducing the vehicle energy consumption and reducing the enterprise operation cost. Finally, the charging path instruction containing the charging path node sequence is generated based on the charging path optimization strategy and the energy saving amount information, and is associated with the energy saving amount information and output to the enterprise vehicle management terminal, improving the intelligent level and efficiency of enterprise vehicle operation management. BRIEF DESCRIPTION OF DRAWINGS

[0008] Figure 1 is the execution flow diagram of the charging path planning method for energy saving of enterprise vehicle energy consumption provided by the embodiments of the present application.

[0009] Figure 2 is the schematic diagram of exemplary hardware and software components of the charging path planning system for energy saving of enterprise vehicle energy consumption provided by the embodiments of the present application. DETAILED DESCRIPTION

[0010] The present application will be described in detail below with reference to the accompanying drawings, Figure 1 is the flow diagram of the charging path planning method for energy saving of enterprise vehicle energy consumption provided by an embodiment of the present application, and the charging path planning method for energy saving of enterprise vehicle energy consumption will be described in detail below.

[0011] Step S110: acquiring a multi-source operation data set of the enterprise vehicle, the multi-source operation data set containing vehicle real-time state data, environment and map data, and historical and cost data.

[0012] In this embodiment, the specific data sources and contents are as follows: The vehicle real-time state data is collected by the vehicle-mounted sensor and the vehicle control system in real time, including the current position coordinate data (the latitude and longitude information output by the vehicle-mounted GPS module), the current battery state of charge data (the percentage of the remaining battery capacity in the total battery capacity output by the vehicle battery management system), the real-time load data (the current load weight output by the vehicle-mounted load sensor), the real-time speed data (the current driving speed output by the vehicle-mounted speed sensor), the current battery temperature data (the average temperature of the battery monomer output by the battery temperature sensor), the current battery voltage data (the total battery voltage output by the battery voltage sensor), the current battery current data (the battery charging and discharging current output by the battery current sensor), and the battery cycle number data (the number of completed charging and discharging cycles recorded by the battery management system).

[0013] The environment and map data are obtained by connecting the third-party weather platform and the map service interface, including the current temperature data (the air temperature at the current position of the vehicle output by the weather platform), the current weather condition data (the weather type at the current position of the vehicle output by the weather platform, such as sunny, cloudy, rainy, snowy, etc.), the current road segment feature data (the feature information of the road segment where the vehicle is currently located output by the map service interface, including the road segment slope data, the road segment congestion degree data, and the road segment pavement material data), the surrounding candidate charging station position coordinate data (the latitude and longitude information of all available charging stations within the preset range of the current position of the vehicle output by the map service interface), the fixed station charging station position coordinate data (the latitude and longitude information of the enterprise-owned charging station preset by the enterprise vehicle fleet management system), the road distance data corresponding to each connection edge (the road length between two nodes output by the map service interface), the estimated driving time data (the estimated time for the vehicle to pass through a certain road output by the map service interface), and the road speed limit data (the maximum allowed driving speed of a certain road output by the map service interface).

[0014] The historical and cost data is extracted from the historical database of the enterprise vehicle fleet management system, including historical charging electricity data (actual charging electricity of the enterprise vehicle in each past charging period), corresponding charging period mileage data (total mileage of the enterprise vehicle in each past charging period), historical charging unit price data (electricity price of the enterprise vehicle in each past charging), charging electricity data corresponding to each charging (charging electricity of the enterprise vehicle in each past charging), enterprise vehicle annual average driving mileage data (total driving mileage of the enterprise vehicle in the past year divided by 12 months to obtain the monthly average driving mileage), battery health benchmark cycle number data (upper limit of the cycle number of the battery under normal use provided by the battery manufacturer), mileage utilization rate data of the enterprise vehicle in multiple past charging periods (ratio of the actual driving mileage of the enterprise vehicle in each past charging period to the standard cruising range of the vehicle), health index data of the enterprise vehicle battery at multiple past time nodes (health condition score of the enterprise vehicle battery at each past time node), past monthly stability variance data of the corresponding driver of the enterprise vehicle (mileage utilization rate fluctuation degree of the corresponding driver of the enterprise vehicle in each past month), number of times that the corresponding driver of the enterprise vehicle obeys the charging path optimization strategy in the past (number of times that the corresponding driver of the enterprise vehicle obeys the recommended charging path of the system in the past), total charging times data (total charging times of the corresponding driver of the enterprise vehicle in the past), and historical comprehensive cost data of the enterprise vehicle under the same operating conditions in the past (comprehensive cost of the enterprise vehicle under the same operating conditions in the past to go to the fixed charging station of the station).

[0015] It is worth noting that the collection of the above data needs to comply with privacy protection specifications, and sensitive data such as vehicle location and driver identity need to be desensitized, such as fuzzy processing of location coordinates (accurate to two decimal places), and anonymization conversion of driver identity information (using driver number instead of real name), to ensure that data collection and use comply with relevant legal regulations.

[0016] Step S120: performing unit electricity consumption prediction processing based on the multi-source operation data set to obtain the predicted unit electricity consumption of the enterprise vehicle under the current operating conditions.

[0017] The embodiment integrates historical data and real-time data, and uses a pre-trained model to accurately predict unit electricity consumption. The specific process is as follows: Step S121: extracting historical charging electricity data and corresponding charging period mileage data of the enterprise vehicle from the historical and cost data of the multi-source operation data set, and calculating the historical average unit electricity consumption of the enterprise vehicle based on the historical charging electricity data and the charging period mileage data. The historical average unit electricity consumption is the ratio of the historical charging electricity data to the corresponding charging period mileage data.

[0018] In this embodiment, the historical charging power data is the actual charging power data of the enterprise vehicle in each past charging period, and the corresponding charging period mileage data is the total mileage data of the vehicle in the charging period. When calculating the historical average unit power consumption, the historical charging power data and the corresponding charging period mileage data of each charging period are matched one by one, and then the sum of all matched historical charging power data and the sum of all corresponding charging period mileage data are calculated. Finally, the sum of the historical charging power data is divided by the sum of the corresponding charging period mileage data to obtain the historical average unit power consumption.

[0019] For example, the historical charging power data of a certain enterprise logistics vehicle in the past 3 charging periods is Q1, Q2 and Q3, and the corresponding charging period mileage data is L1, L2 and L3. Then the historical average unit power consumption η_avg=(Q1+Q2+Q3) / (L1+L2+L3).

[0020] Step S122: Extract the vehicle real-time load data, vehicle real-time speed data and vehicle battery current state of charge data from the vehicle real-time state data of the multi-source operation data set, extract the current temperature data, current weather condition data and current road section characteristic data from the environment and map data of the multi-source operation data set. The current road section characteristic data includes road section slope data, road section congestion degree data and road section pavement material data.

[0021] In this embodiment, the vehicle real-time load data is collected by the vehicle-mounted load sensor, the vehicle real-time speed data is collected by the vehicle-mounted GPS and speed sensor, and the vehicle battery current state of charge data is obtained by the vehicle battery management system. The current temperature data and the current weather condition data are obtained by connecting the third-party meteorological platform interface, and the current road section characteristic data is obtained by connecting the map service interface. The road section slope data is the ratio of the change of the altitude of the road section to the length of the road section, the road section congestion degree data is the ratio of the current speed of the road section to the speed limit of the road section, and the road section pavement material data is the quantitative data corresponding to the material type of the road section pavement.

[0022] For example, the current real-time load data of the enterprise logistics vehicle is W (the weight of the goods detected by the vehicle-mounted load sensor), the real-time speed data is V (the current driving speed output by the vehicle-mounted GPS), the current state of charge data of the battery is SOC (the percentage of the remaining power output by the battery management system), the current temperature data is T (the temperature at the current location output by the weather platform), the current weather condition data is Wt (the current weather type output by the weather platform, such as rainy day), the road slope data in the current road section feature data is S (the ratio of the change in elevation of the current road section to the length of the road section output by the map service interface), the road congestion degree data is C (the ratio of the current road section speed to the speed limit output by the map service interface), and the road surface material data is M (the quantized value corresponding to the current road surface material output by the map service interface, such as 0.8 for asphalt pavement and 0.6 for cement pavement).

[0023] Step S123: inputting the historical average unit power consumption, the real-time load data of the vehicle, the real-time speed data of the vehicle, the current state of charge data of the battery of the vehicle, the current temperature data, the current weather condition data, and the current road section feature data into the pre-trained unit power consumption prediction model, wherein the unit power consumption prediction model comprises a feature fusion layer and a prediction output layer.

[0024] In this embodiment, the pre-trained unit power consumption prediction model is a model trained based on a machine learning algorithm, the feature fusion layer is used for fusion processing of the input multi-dimensional features, and the prediction output layer is used for converting the fused features into predicted unit power consumption. Each data input into the model needs to be standardized to convert data of different dimensions into the same dimension range, so as to ensure the consistency of the model input data.

[0025] For example, after the historical average unit power consumption η avg, the real-time load data W, the real-time speed data V, the current state of charge data SOC of the battery, the current temperature data T, the current weather condition data Wt, and the current road section feature data (S, C, M) are standardized, they are input into the pre-trained unit power consumption prediction model. The feature fusion layer of the model integrates the above features, and the prediction output layer outputs the predicted unit power consumption η pred.

[0026] Step S124: performing feature correlation processing on the input historical average unit power consumption, the real-time load data of the vehicle, the real-time speed data of the vehicle, the current state of charge data of the battery of the vehicle, the current temperature data, the current weather condition data, and the current road section feature data through the feature fusion layer of the unit power consumption prediction model to generate an electric power consumption influence feature vector fused with multiple factors, wherein the feature correlation processing includes dimension unification of different data type features and modeling of the correlation relationship between features.

[0027] The embodiment generates an electricity consumption influence feature vector fused with the influence of multiple factors by multi-dimensional processing and integration of input features, and the specific process is as follows: For example, in step S1241, the input historical average unit electricity consumption, real-time vehicle load data, real-time vehicle speed data, current state of charge data of the vehicle battery, current temperature data, current weather condition data, and current road section characteristic data are received, the historical average unit electricity consumption is subjected to time decay processing, the value proportion of the historical average unit electricity consumption is adjusted according to the interval between the historical charging period and the current time, and the historical average unit electricity consumption after time decay is obtained.

[0028] In the embodiment, the interval between the historical charging period and the current time needs to be determined first, and the longer the interval, the lower the value proportion of the historical average unit electricity consumption. In specific processing, the decay coefficient can be set according to the interval time, the decay coefficient is k1 when the interval time is t1, the decay coefficient is k2 when the interval time is t2, wherein k1 is less than k2 when t1 is greater than t2, and then the historical average unit electricity consumption is multiplied by the corresponding decay coefficient to obtain the historical average unit electricity consumption after time decay.

[0029] For example, the historical average unit electricity consumption of a certain enterprise logistics vehicle in the past 3 charging periods is η_avg1, η_avg2, and η_avg3, the interval between the corresponding historical charging period and the current time is t1, t2, and t3 (t1>t2>t3), and the decay coefficient is k1, k2, and k3 (k1<k2<k3), respectively. The historical average unit electricity consumption after time decay is η_avg1*k1+η_avg2*k2+η_avg3*k3.

[0030] In step S1242, the real-time load data of the vehicle, the real-time speed data of the vehicle, and the current state of charge data of the vehicle battery are subjected to range calibration processing, the real-time load data of the vehicle is compared and adjusted with the rated load data of the vehicle, the real-time speed data of the vehicle is compared and adjusted with the road speed limit data, and the current state of charge data of the vehicle battery is compared and adjusted with the normal range data of the state of charge of the battery, to obtain the real-time state data of the vehicle after range calibration.

[0031] In the embodiment, the rated load data of the vehicle is the maximum load data designed for the vehicle, the road speed limit data is the maximum allowable driving speed data of the current road section, and the normal range data of the state of charge of the battery is the state of charge range data for normal operation of the battery. In range calibration processing, if the real-time load data of the vehicle exceeds the rated load data of the vehicle, it is adjusted to the rated load data of the vehicle; if the real-time speed data of the vehicle exceeds the road speed limit data, it is adjusted to the road speed limit data; and if the current state of charge data of the vehicle battery exceeds the normal range data of the state of charge of the battery, it is adjusted to the upper limit or lower limit of the normal range data of the state of charge of the battery.

[0032] For example, the rated load data of a certain enterprise logistics vehicle is W max, the current real-time load data is W (W > W max), the range calibrated real-time load data is W max; the current road section speed limit data is V limit, the real-time speed data is V (V > V limit), the range calibrated real-time speed data is V limit; the normal range data of the battery state of charge is SOC min to SOC max, the current battery state of charge data is SOC (SOC < SOC min), the range calibrated battery state of charge data is SOC min.

[0033] Step S1243: numerical regularization processing is performed on the current air temperature data, and the current air temperature data is converted into a value of the same order of magnitude as the historical average unit power consumption; category quantization processing is performed on the current weather condition data, and different weather conditions are converted into corresponding quantized values to obtain the regularized environmental meteorological data.

[0034] In this embodiment, the numerical regularization processing can be achieved by dividing the current air temperature data by a fixed value, and the selection of the fixed value needs to ensure that the processed current air temperature data is of the same order of magnitude as the historical average unit power consumption; the category quantization processing can correspond different weather conditions such as sunny, cloudy, rainy, snowy, etc. to different quantized values, such as a value a for sunny, a value b for cloudy, a value c for rainy, and a value d for snowy, where a, b, c, and d are different values.

[0035] For example, the current air temperature data is T (unit: ℃), the historical average unit power consumption is η avg (unit: kWh / km), the current air temperature data is divided by a fixed value k (such as k = 10) to obtain the regularized air temperature data T' = T / k; the current weather condition is rainy, and the corresponding quantized value is c, so the regularized environmental meteorological data is (T', c).

[0036] Step S1244: attribute disintegration processing is performed on the road section slope data, road section congestion degree data, and road section pavement material data in the current road section feature data, core feature values of each attribute are extracted, and standardization processing is performed on the core feature values to obtain the standardized road section feature data.

[0037] In this embodiment, the attribute disintegration processing needs to extract the core feature values of the positive and negative values of the road section slope data, the size of the road section congestion degree data, and the type of the road section pavement material data; the standardization processing can convert the core feature values into values between 0 and 1, and in the specific processing, the core feature values can be subtracted by the minimum value and divided by the difference between the maximum value and the minimum value, where the minimum value is the historical minimum value of the core feature value, and the maximum value is the historical maximum value of the core feature value.

[0038] For example, the current road slope data is S (unit: %), the core feature value is the positive and negative value of S (positive for uphill and negative for downhill); the road congestion degree data is C (unit: %), the core feature value is the size of C; the road surface material data is M (type: asphalt, cement, gravel), the core feature value is the quantitative value corresponding to the type of M. After standardizing the above core feature values, the standardized road feature data is (S', C', M'), wherein S', C', and M' are all numerical values between 0 and 1.

[0039] Step S1245: Integrating the time-decayed historical average unit power consumption, the range-calibrated vehicle real-time state data, the regularized environmental meteorological data, and the standardized road feature data to obtain a single-feature adaptation result.

[0040] In this embodiment, the integration processing needs to arrange the time-decayed historical average unit power consumption, the range-calibrated vehicle real-time state data (including vehicle real-time load data, vehicle real-time speed data, and vehicle battery current state of charge data), the regularized environmental meteorological data (including current temperature data and current weather condition data), and the standardized road feature data (including road slope data, road congestion degree data, and road surface material data) in a preset order to obtain a single-feature adaptation result.

[0041] For example, the time-decayed historical average unit power consumption is η_avg_decay, the range-calibrated vehicle real-time state data is (W_cal, V_cal, SOC_cal), the regularized environmental meteorological data is (T', Wt_q), and the standardized road feature data is (S', C', M'), and the single-feature adaptation result is [η_avg_decay, W_cal, V_cal, SOC_cal, T', Wt_q, S', C', M'].

[0042] Step S1246: Dividing the single-feature adaptation result into historical statistical features, vehicle state features, environmental features, and road features according to the data source type, performing self-stability analysis on the historical statistical features, calculating the fluctuation degree of the historical average unit power consumption in different historical periods, and generating a historical statistical aggregation feature.

[0043] In this embodiment, the self-stability analysis needs to calculate the difference between the maximum and minimum values of the historical average unit power consumption in different historical periods, and the smaller the difference, the lower the fluctuation degree. When generating the historical statistical aggregation feature, the historical average unit power consumption and the fluctuation degree can be combined to obtain the historical statistical aggregation feature.

[0044] For example, the historical statistics type feature is η avg decay, the maximum value η max and the minimum value η min of η avg decay in different historical periods are calculated, the fluctuation degree is Δη = η max - η min, and the historical statistics type aggregated feature is [η avg decay, Δη].

[0045] Step S1247: Interrelate the vehicle state type features, calculate the correlation degree of the vehicle real-time load data and the vehicle real-time speed data, and the correlation degree of the vehicle real-time speed data and the vehicle battery current state of charge data, and generate the vehicle state type aggregated feature.

[0046] In this embodiment, the correlation degree can be realized by calculating the correlation coefficient of the two data. The closer the correlation coefficient is to 1, the higher the correlation degree is. When generating the vehicle state type aggregated feature, the vehicle real-time load data, the vehicle real-time speed data, the vehicle battery current state of charge data, and the correlation coefficient can be combined to obtain the vehicle state type aggregated feature.

[0047] For example, the vehicle state type features are (W cal, V cal, SOC cal ), the correlation coefficient r1 of W cal and V cal, and the correlation coefficient r2 of V cal and SOC cal are calculated, and the vehicle state type aggregated feature is [W cal, V cal, SOC cal, r1, r2].

[0048] Step S1248: Preliminary distribution of influence weights is performed on the environment type features, the influence proportions of the current temperature data and the current weather condition data are determined according to the combination relationship, and the environment type aggregated feature is generated.

[0049] In this embodiment, the preliminary distribution of influence weights needs to be determined according to the influence degree of the combination of the current temperature data and the current weather condition data on the unit power consumption. The higher the influence degree is, the greater the influence proportion is. When generating the environment type aggregated feature, the current temperature data, the current weather condition data, and the respective influence proportions can be combined to obtain the environment type aggregated feature.

[0050] For example, the environment type features are (T', Wt q ), the influence proportion of T' is w T, and the influence proportion of Wt q is w Wt (w T + w Wt = 1) according to the combination relationship of T' and Wt q, and the environment type aggregated feature is [T', Wt q, w T, w Wt ].

[0051] Step S1249: Attribute coupling analysis is performed on the road segment type features, the coupling coefficient of the road segment slope data and the road segment pavement material data, and the coupling coefficient of the road segment congestion degree data and the road segment slope data are calculated, and the road segment type aggregated feature is generated.

[0052] In this embodiment, the coupling coefficient can be realized by calculating the product of two data, and the greater the product, the higher the coupling degree. When generating the road section class aggregation feature, the road section slope data, the road section congestion degree data, the road section pavement material data and the coupling coefficient can be combined to obtain the road section class aggregation feature.

[0053] For example, the road section class feature is (S', C', M'), the coupling coefficient c1=S'*M' of S' and M' is calculated, the coupling coefficient c2=C'*S' of C' and S' is calculated, and the road section class aggregation feature is [S', C', M', c1, c2].

[0054] Step S12410: integrating the historical statistics class aggregation feature, the vehicle state class aggregation feature, the environment class aggregation feature and the road section class aggregation feature to obtain a class-in aggregation feature set.

[0055] In this embodiment, the integration process needs to arrange the historical statistics class aggregation feature, the vehicle state class aggregation feature, the environment class aggregation feature and the road section class aggregation feature in a predetermined order to obtain the class-in aggregation feature set.

[0056] For example, the historical statistics class aggregation feature is [η_avg_decay, Δη], the vehicle state class aggregation feature is [W_cal, V_cal, SOC_cal, r1, r2], the environment class aggregation feature is [T', Wt_q, w_T, w_Wt], and the road section class aggregation feature is [S', C', M', c1, c2]. The class-in aggregation feature set is [η_avg_decay, Δη, W_cal, V_cal, SOC_cal, r1, r2, T', Wt_q, w_T, w_Wt, S', C', M', c1, c2].

[0057] Step S12411: analyzing the mutual influence relationship between the vehicle state class aggregation feature and the road section class aggregation feature, and generating a first cross-class association feature according to the correspondence between the vehicle real-time load data and the road section slope data and the correspondence between the vehicle real-time speed data and the road section congestion degree data.

[0058] In this embodiment, the mutual influence relationship analysis needs to determine the change of the influence of the road section slope data on the unit power consumption when the vehicle real-time load data increases, and the change of the influence of the road section congestion degree data on the unit power consumption when the vehicle real-time speed data increases. When generating the first cross-class association feature, the correspondence between the vehicle real-time load data and the road section slope data and the correspondence between the vehicle real-time speed data and the road section congestion degree data can be combined to obtain the first cross-class association feature.

[0059] For example, the vehicle real-time load data is W cal, the road slope data is S', the corresponding relationship is that when W cal increases, the influence coefficient of S' on unit power consumption is k WS; the vehicle real-time speed data is V cal, the road congestion degree data is C', the corresponding relationship is that when V cal increases, the influence coefficient of C' on unit power consumption is k VC, and the first cross-class associated feature is [W cal, S', k WS, V cal, C', k VC].

[0060] Step S12412: analyzing the mutual influence relationship between the environment class aggregated feature and the vehicle state class aggregated feature, generating a second cross-class associated feature according to the corresponding relationship between the current temperature data and the vehicle battery current state of charge data and the corresponding relationship between the current weather condition data and the vehicle real-time speed data.

[0061] In this embodiment, the mutual influence relationship analysis needs to determine the influence change of the vehicle battery current state of charge data on unit power consumption when the current temperature data decreases and the influence change of the vehicle real-time speed data on unit power consumption when the current weather condition data is rainy. When generating the second cross-class associated feature, the corresponding relationship between the current temperature data and the vehicle battery current state of charge data and the corresponding relationship between the current weather condition data and the vehicle real-time speed data can be combined to obtain the second cross-class associated feature.

[0062] For example, the current temperature data is T', the vehicle battery current state of charge data is SOC cal, the corresponding relationship is that when T' decreases, the influence coefficient of SOC cal on unit power consumption is k TSOC; the current weather condition data is Wt q (rainy day), the vehicle real-time speed data is V cal, the corresponding relationship is that when Wt q is rainy, the influence coefficient of V cal on unit power consumption is k WtV, and the second cross-class associated feature is [T', SOC cal, k TSOC, Wt q, V cal, k WtV].

[0063] Step S12413: analyzing the mutual influence relationship between the historical statistical class aggregated feature and other class aggregated features, generating a third cross-class associated feature according to the corresponding relationship between the historical average unit power consumption and the vehicle real-time load data and the corresponding relationship between the historical average unit power consumption and the road feature data.

[0064] In this embodiment, the mutual influence relationship analysis needs to determine the influence change of the vehicle real-time load data on unit power consumption when the historical average unit power consumption increases and the influence change of the road feature data on unit power consumption when the historical average unit power consumption increases. When generating the third cross-class associated feature, the corresponding relationship between the historical average unit power consumption and the vehicle real-time load data and the corresponding relationship between the historical average unit power consumption and the road feature data can be combined to obtain the third cross-class associated feature.

[0065] For example, the historical average unit power consumption is η avg decay, the vehicle real-time load data is W cal, the corresponding relationship is that when η avg decay increases, the influence coefficient of W cal on the unit power consumption is k _ ηW; the historical average unit power consumption is η avg decay, the road section characteristic data is [S', C', M'], the corresponding relationship is that when η avg decay increases, the influence coefficient of [S', C', M'] on the unit power consumption is k _ ηS, k _ ηC, k _ ηM, and the third cross-class association characteristic is [η avg decay, W cal, k _ ηW, S', k _ ηS, C', k _ ηC, M', k _ ηM].

[0066] Step S12414: The first cross-class association characteristic, the second cross-class association characteristic and the third cross-class association characteristic are integrated to obtain a cross-class association characteristic set.

[0067] In this embodiment, the integration processing needs to arrange the first cross-class association characteristic, the second cross-class association characteristic and the third cross-class association characteristic according to a preset order to obtain the cross-class association characteristic set.

[0068] For example, the first cross-class association characteristic is [W cal, S', k_WS, V cal, C', k_VC], the second cross-class association characteristic is [T', SOC cal, k_TSOC, Wt_q, V cal, k_WtV], and the third cross-class association characteristic is [η avg decay, W cal, k _ ηW, S', k _ ηS, C', k _ ηC, M', k _ ηM], and the cross-class association characteristic set is [W cal, S', k_WS, V cal, C', k_VC, T', SOC cal, k_TSOC, Wt_q, V cal, k_WtV, η avg decay, W cal, k _ ηW, S', k _ ηS, C', k _ ηC, M', k _ ηM].

[0069] Step S12415: The dimensions of all cross-class association characteristics in the cross-class association characteristic set are detected to determine the current dimensions of the cross-class association characteristics, and the dimensions of the cross-class association characteristics are converted based on a preset target dimension to make the dimensions of all cross-class association characteristics consistent with the target dimension, thereby obtaining a dimension-unified characteristic set.

[0070] In this embodiment, the dimension detection needs to determine the current dimensions of the cross-class association characteristics, such as that the current dimension of the first cross-class association characteristic is d1, the current dimension of the second cross-class association characteristic is d2, and the current dimension of the third cross-class association characteristic is d3. The dimension conversion needs to convert the cross-class association characteristics into a preset target dimension d_target, so that the dimensions of all cross-class association characteristics are d_target.

[0071] For example, the preset target dimension is d_target, the current dimension of the first cross-class association feature is d1, which is converted to the dimension d_target through dimension conversion; the current dimension of the second cross-class association feature is d2, which is converted to the dimension d_target through dimension conversion; the current dimension of the third cross-class association feature is d3, which is converted to the dimension d_target through dimension conversion. The feature set after dimension conversion is the feature set after dimension unification.

[0072] Step S12416: Mutual dependence relationship analysis is performed on each feature in the feature set after dimension unification, the correlation degree between any two features is calculated, a corresponding association weight is assigned to each feature according to the correlation degree, and a weighted feature set is obtained.

[0073] In this embodiment, the mutual dependence relationship analysis needs to calculate the correlation coefficient between any two features. The closer the correlation coefficient is to 1, the higher the correlation degree is. According to the correlation degree, a corresponding association weight is assigned to each feature. The higher the correlation degree is, the greater the weight value is.

[0074] For example, the feature set after dimension unification is [F1, F2, F3,..., Fn], the correlation coefficient r_ij (i, j = 1, 2,..., n) between any two features is calculated, and the association weight w_i is assigned to each feature Fi according to r_ij. The weighted feature set is [F1*w_1, F2*w_2,..., Fn*w_n].

[0075] Step S12417: The weighted feature set is sequentially arranged according to the preset feature sorting rule, and a power consumption influence feature vector fused with the influence of multiple factors is generated.

[0076] In this embodiment, the preset feature sorting rule can be arranged according to the influence degree of the feature on unit power consumption from large to small. The feature with greater influence degree is arranged in front.

[0077] For example, the weighted feature set is [F1*w_1, F2*w_2,..., Fn*w_n], and the power consumption influence feature vector fused with the influence of multiple factors is [Fk*w_k, Fm*w_m,..., Fl*w_l] according to the influence degree of the feature on unit power consumption from large to small. Fk*w_k has the greatest influence on unit power consumption, and Fl*w_l has the least influence on unit power consumption.

[0078] Step S125: Based on the power consumption influence feature vector, power consumption prediction calculation is performed through the prediction output layer of the unit power consumption prediction model. The prediction output layer converts the power consumption influence feature vector into a corresponding power consumption prediction value through a preset mapping relationship, and obtains the predicted unit power consumption of the enterprise vehicle under the current operating condition.

[0079] For example, this embodiment obtains the predicted unit power consumption by processing and mapping the feature vector of power consumption influence. The specific process is as follows: Step S1251: Receive the power consumption influence feature vector, analyze each feature component in the power consumption influence feature vector one by one, extract the numerical magnitude and position information of each feature component, and obtain the feature component analysis result.

[0080] In this embodiment, feature component analysis needs to extract the numerical magnitude and position information of each feature component in the power consumption influence feature vector. For example, if the feature vector is [F1, F2, ..., Fn], then the feature component analysis result is [(F1, pos1), (F2, pos2), ..., (Fn, posn)], where pos is the position information of the feature component Fi.

[0081] Step S1252: Based on the location information in the feature component analysis results, divide the power consumption influence feature vector into multiple feature sub-vectors. Each feature sub-vector corresponds to a type of influencing factor, which includes historical statistical influencing factors, vehicle status influencing factors, environmental influencing factors, and road segment influencing factors, thus obtaining a set of classification feature sub-vectors.

[0082] In this embodiment, the power consumption impact feature vector is divided into multiple feature sub-vectors based on the location information, and each feature sub-vector corresponds to a type of influencing factor. For example, location information pos1 to pos3 corresponds to historical statistical influencing factors, location information pos4 to pos6 corresponds to vehicle status influencing factors, location information pos7 to pos9 corresponds to environmental influencing factors, and location information pos10 to pos12 corresponds to road segment influencing factors. Then the set of classification feature sub-vectors is [F1-F3, F4-F6, F7-F9, F10-F12].

[0083] Step S1253: Perform internal consistency verification on each feature subvector in the classification feature subvector set, check whether the numerical change trend of each feature component in the same feature subvector conforms to the preset rule, remove abnormal feature components that do not conform to the preset rule, and obtain the verified feature subvector set.

[0084] In this embodiment, the internal consistency check needs to check whether the numerical change trend of each feature component in the same feature sub-vector conforms to the preset rule. For example, the value of each feature component in the feature sub-vector of historical statistical influencing factors should show a decreasing trend over time. If the value of a certain feature component shows an increasing trend, it is an abnormal feature component and needs to be removed.

[0085] For example, the feature sub-vector of the historical statistical influencing factor is [F1, F2, F3], the preset rule is that the numerical value presents a decreasing trend over time, if the value of F2 is greater than the value of F1, F2 is an abnormal feature component, and the feature sub-vector after verification is [F1, F3] after removing F2.

[0086] Step S1254: performing information condensation processing on each feature sub-vector in the feature sub-vector set after verification, extracting core representation information of each feature sub-vector, converting the core representation information into a fixed-length sub-vector representation value, and obtaining a sub-vector representation value set.

[0087] In this embodiment, the information condensation processing needs to extract the core representation information of each feature sub-vector. For example, the feature sub-vector is [F1, F2, F3], the core representation information is F1+F2+F3, and the sub-vector representation value is F1+F2+F3. The core representation information is converted into a fixed-length sub-vector representation value to obtain a sub-vector representation value set.

[0088] For example, the feature sub-vector set after verification is [V1, V2, V3, V4], wherein V1 is the feature sub-vector of the historical statistical influencing factor, V2 is the feature sub-vector of the vehicle state influencing factor, V3 is the feature sub-vector of the environmental influencing factor, and V4 is the feature sub-vector of the road segment influencing factor. After performing information condensation processing on each feature sub-vector, the sub-vector representation value set is [V1_rep, V2_rep, V3_rep, V4_rep], wherein Vi_rep is the representation value of the feature sub-vector Vi.

[0089] Step S1255: based on a preset mapping rule, performing preliminary mapping processing on each sub-vector representation value, converting the sub-vector representation value into a corresponding preliminary power consumption contribution value, the mapping rule is determined based on the corresponding relationship between the feature sub-vector and the power consumption contribution in a large amount of historical data, and a preliminary power consumption contribution value set is obtained.

[0090] In this embodiment, the preset mapping rule is determined based on the corresponding relationship between the feature sub-vector and the power consumption contribution in a large amount of historical data, for example, when the feature sub-vector representation value is V_rep, the corresponding power consumption contribution value is C_rep. The preliminary mapping processing needs to convert each sub-vector representation value into a corresponding preliminary power consumption contribution value.

[0091] For example, the sub-vector representation value set is [V1_rep, V2_rep, V3_rep, V4_rep], the preset mapping rule is V1_rep→C1_rep, V2_rep→C2_rep, V3_rep→C3_rep, and V4_rep→C4_rep, and the preliminary power consumption contribution value set is [C1_rep, C2_rep, C3_rep, C4_rep].

[0092] Step S1256: mutual interaction analysis is performed on all the preliminary power consumption contribution values in the preliminary power consumption contribution value set, the synergistic influence degree between any two preliminary power consumption contribution values is calculated, each preliminary power consumption contribution value is adjusted according to the synergistic influence degree, and an adjusted power consumption contribution value set is obtained.

[0093] In this embodiment, the mutual interaction analysis needs to calculate the synergistic influence degree between any two preliminary power consumption contribution values, such as the synergistic influence degree between C1_rep and C2_rep is s_12, the greater s_12 is, the stronger the synergistic effect of the two is. Each preliminary power consumption contribution value is adjusted according to the synergistic influence degree, such as C1_rep is adjusted to C1_rep*(1+s_12), and C2_rep is adjusted to C2_rep*(1+s_12).

[0094] For example, the preliminary power consumption contribution value set is [C1_rep, C2_rep, C3_rep, C4_rep], the synergistic influence degree s_ij (i, j = 1, 2, 3, 4) between any two preliminary power consumption contribution values is calculated, each preliminary power consumption contribution value is adjusted according to s_ij, and the adjusted power consumption contribution value set is [C1_adj, C2_adj, C3_adj, C4_adj].

[0095] Step S1257: all the adjusted power consumption contribution values in the adjusted power consumption contribution value set are accumulated to obtain a preliminary power consumption prediction value.

[0096] In this embodiment, the accumulation processing needs to add all the adjusted power consumption contribution values in the adjusted power consumption contribution value set to obtain the preliminary power consumption prediction value.

[0097] For example, the adjusted power consumption contribution value set is [C1_adj, C2_adj, C3_adj, C4_adj], and the preliminary power consumption prediction value is C_pred_init = C1_adj + C2_adj + C3_adj + C4_adj.

[0098] Step S1258: from the historical and cost data of the multi-source operation data set, the power consumption prediction error data of the enterprise vehicle under the same feature combination in the past is extracted, the same feature combination includes the same historical statistical features, vehicle state features, environmental features and road segment features, and a historical prediction error set is obtained.

[0099] In this embodiment, the same feature combination needs to be consistent with the current feature combination, such as the current feature combination is (historical statistical feature H, vehicle state feature V, environmental feature E, road segment feature R), and the power consumption prediction error data under the same feature combination (H, V, E, R) in the past needs to be extracted.

[0100] For example, the past power consumption prediction error data under the same feature combination is [ε1, ε2,..., εm], and the historical prediction error set is [ε1, ε2,..., εm].

[0101] Step S1259: statistically analyzing the historical prediction error set, extracting the distribution characteristics of the historical prediction error, determining the error correction amount corresponding to the current power consumption prediction preliminary value based on the distribution characteristics, and obtaining the error correction amount data.

[0102] In this embodiment, the statistical analysis needs to extract the distribution characteristics of the historical prediction error, such as the mean μ and the variance σ². The error correction amount needs to be determined according to the distribution characteristics, such as the current power consumption prediction preliminary value C_pred_init and the error correction amount ε_corr = μ + k * σ (k is a constant).

[0103] For example, the mean of the historical prediction error set is μ, the variance is σ², and the constant k is 2. Then the error correction amount is ε_corr = μ + 2 * σ.

[0104] Step S12510: fusing the power consumption prediction preliminary value and the error correction amount data to obtain the calibrated power consumption prediction value.

[0105] In this embodiment, the fusion processing needs to add the power consumption prediction preliminary value and the error correction amount data to obtain the calibrated power consumption prediction value.

[0106] For example, the power consumption prediction preliminary value is C_pred_init, and the error correction amount is ε_corr. Then the calibrated power consumption prediction value is C_pred_cal = C_pred_init + ε_corr.

[0107] Step S12511: extracting the current state of charge data of the vehicle battery, and performing matching check on the calibrated power consumption prediction value and the current state of charge data of the vehicle battery to determine whether the calibrated power consumption prediction value conforms to the power consumption rule under the current battery state.

[0108] In this embodiment, the matching check needs to determine whether the calibrated power consumption prediction value conforms to the power consumption rule under the current battery state, such as the current battery state of charge SOC and the power consumption rule that the power consumption increases with the decrease of SOC. Then it needs to determine whether the calibrated power consumption prediction value increases with the decrease of SOC.

[0109] For example, the current battery state of charge is SOC_low (low), and the calibrated power consumption prediction value is C_pred_cal. If C_pred_cal conforms to the rule that the power consumption increases with the decrease of SOC, the matching check passes; otherwise, the matching check fails.

[0110] Step S12512: If the power consumption rule under the current battery state is met, the calibrated power consumption prediction value is determined as the predicted unit power consumption of the enterprise vehicle under the current operating condition.

[0111] In this embodiment, if the matching check passes, the calibrated power consumption prediction value is the predicted unit power consumption.

[0112] For example, if the matching check passes, the predicted unit power consumption is η_pred = C_pred_cal.

[0113] Step S12513: If the power consumption rule under the current battery state is not met, the mapping rule is adjusted again to map the values of each sub-vector again, and the steps of the preliminary mapping to the matching check are repeated until the predicted unit power consumption that meets the power consumption rule under the current battery state is obtained.

[0114] In this embodiment, if the matching check does not pass, the mapping rule needs to be adjusted again, such as adjusting the constant k in the mapping rule, and then repeating the steps of the preliminary mapping to the matching check until the predicted unit power consumption that meets the power consumption rule under the current battery state is obtained.

[0115] For example, if the matching check does not pass, the constant k in the mapping rule is adjusted to 1, and then the steps S1255 to S12511 are repeated until the predicted unit power consumption that meets the power consumption rule under the current battery state is obtained.

[0116] Step S130: A path graph containing the candidate charging station nodes and the vehicle current position nodes is constructed, and the first comprehensive cost of going to the nearest charging station and the second comprehensive cost of returning to the fixed station charging station are calculated based on the predicted unit power consumption.

[0117] This embodiment realizes the preliminary screening of the charging path by constructing the path graph and calculating the comprehensive cost, and the specific process is as follows: Step S131: The enterprise vehicle current position coordinate data, the surrounding candidate charging station position coordinate data, and the fixed station charging station position coordinate data are extracted from the environmental and map data of the multi-source operating data set. The surrounding candidate charging station position coordinate data is the position coordinate data of all available charging stations within the preset range of the enterprise vehicle current position.

[0118] In this embodiment, the enterprise vehicle current position coordinate data is the latitude and longitude information of the current position of the vehicle, the surrounding candidate charging station position coordinate data is the latitude and longitude information of all available charging stations within the preset range (such as 5 kilometers) of the current position of the vehicle, and the fixed station charging station position coordinate data is the latitude and longitude information of the enterprise-owned charging station.

[0119] For example, the current position coordinate data of the enterprise vehicle is (Lon0, Lat0), the position coordinate data of the surrounding candidate charging station is (Lon1, Lat1), (Lon2, Lat2), …, (Lonn, Latn), and the position coordinate data of the fixed station charging station is (Lonf, Latf).

[0120] Step S132: Taking the position corresponding to the current position coordinate data of the enterprise vehicle as the starting node, taking the positions corresponding to the position coordinate data of the surrounding candidate charging stations as the nearby candidate nodes, taking the position corresponding to the position coordinate data of the fixed station charging station as the fixed station node, and taking the passable roads between the nodes as the connection edges, an enterprise vehicle charging path graph is constructed, wherein each connection edge in the enterprise vehicle charging path graph corresponds to a passable road.

[0121] In this embodiment, the nodes include the starting node, the nearby candidate nodes, and the fixed station node, and the connection edges are the passable roads between the nodes.

[0122] For example, the starting node is N0 (Lon0, Lat0), the nearby candidate nodes are N1 (Lon1, Lat1), N2 (Lon2, Lat2), …, Nn (Lonn, Latn), the fixed station node is Nf (Lonf, Latf), and the connection edges are the passable roads N0-N1, N0-N2, …, N0-Nn, N0-Nf, N1-N2, …, Nn-Nf, etc.

[0123] Step S133: The road distance data and the estimated travel time data corresponding to each connection edge are extracted from the environment and map data of the multi-source operation data set, the electric consumption cost data corresponding to each connection edge is calculated by combining the predicted unit electric consumption, the electric consumption cost data is the product of the road distance data corresponding to the connection edge and the predicted unit electric consumption, and the estimated travel time data is calculated based on the road distance data and the current road section average travel speed data.

[0124] In this embodiment, the road distance data is the road length corresponding to the connection edge, the estimated travel time data is the road distance data divided by the current road section average travel speed data, and the electric consumption cost data is the road distance data multiplied by the predicted unit electric consumption.

[0125] For example, the road distance data corresponding to the connection edge N0-N1 is D01, the current road section average travel speed data is V01, the estimated travel time data is T01=D01 / V01, the predicted unit electric consumption is η_pred, and the electric consumption cost data is C01=D01*η_pred.

[0126] Step S134: The road distance data, the estimated travel time data and the electricity consumption cost data are normalized respectively, and corresponding weight coefficients are set for each standardized value. The weight coefficients are used to weight and sum each standardized value to obtain the comprehensive cost weight of each connection edge. The weight coefficients are determined based on the importance of distance, time and electricity consumption in the operation of the enterprise vehicle.

[0127] In this embodiment, the normalization process needs to convert the road distance data, the estimated travel time data and the electricity consumption cost data into values between 0 and 1. The weight coefficients need to be determined based on the importance of distance, time and electricity consumption in the operation of the enterprise vehicle. For example, if the enterprise pays more attention to electricity consumption, the weight coefficient of the electricity consumption cost data will be larger.

[0128] For example, the normalized value of the road distance data D is D_norm, the normalized value of the estimated travel time data T is T_norm, and the normalized value of the electricity consumption cost data C is C_norm. The weight coefficients are α, β and γ respectively (α+β+γ=1), and the comprehensive cost weight is ω=α*D_norm+β*T_norm+γ*C_norm.

[0129] Step S135: Based on the comprehensive cost weight of each connection edge in the charging path graph, a path search algorithm is used to calculate the total cost of the path from the starting node to each nearby candidate node, and the smallest total cost is selected as the first comprehensive cost of the path to the nearby charging station. The total cost of the path from the starting node to the fixed station node is calculated as the second comprehensive cost of the path to the fixed station charging station. The path search algorithm is used to find the path with the smallest sum of comprehensive cost weights between nodes.

[0130] For example, this step calculates the comprehensive cost by using the path search algorithm to realize the screening of the charging path, and the specific process is as follows: Step S1351: Obtain the charging path graph and the comprehensive cost weight of each connection edge. Extract the starting node information, the nearby candidate node set information, the fixed station node information and the connection edge information between nodes from the charging path graph. Integrate the starting node information, the nearby candidate node set information, the fixed station node information, the connection edge information and the comprehensive cost weight of each connection edge to obtain the node weight set.

[0131] In this step, the charging path graph contains nodes and connection edges, and the node weight set needs to integrate the information of all nodes and connection edges and the corresponding comprehensive cost weight. For example, the starting node of the charging path graph is N0, the nearest candidate node set is {N1, N2, N3}, and the fixed station node is Nf; the connection edge information is N0-N1, N0-N2, N0-N3, N0-Nf, N1-N2, N1-N3, N2-N3, N2-Nf, and N3-Nf; the comprehensive cost weight of each connection edge is ω01, ω02, ω03, ω0f, ω12, ω13, ω23, ω2f, and ω3f. After integration, the node weight set is {node: {N0: [], N1: [], N2: [], N3: [], Nf: []}, connection edge: {N0-N1: ω01, N0-N2: ω02, N0-N3: ω03, N0-Nf: ω0f, N1-N2: ω12, N1-N3: ω13, N2-N3: ω23, N2-Nf: ω2f, N3-Nf: ω3f}}.

[0132] Step S1352: Extract the charging station available state data corresponding to each nearest candidate node from the environment and map data of the multi-source operation data set, the charging station available state data contains charging pile occupation data and charging service availability data, and the straight-line distance data from the starting node to each nearest candidate node is combined to prioritize all nearest candidate nodes in the nearest candidate node set to obtain the sorted nearest candidate node sequence.

[0133] In this step, the charging station available state data needs to reflect the actual availability of the candidate charging station, the charging pile occupation data is the ratio of the number of currently occupied charging piles to the total number of charging piles of the candidate charging station, and the charging service availability data is an identifier indicating whether the candidate charging station provides charging service (such as 1 indicating available and 0 indicating unavailable); the straight-line distance data is the straight-line distance from the starting node to each nearest candidate node. The priority sorting needs to consider the available state and the straight-line distance, and the candidate node with better available state and closer straight-line distance has higher priority.

[0134] For example, the set of nearby candidate nodes is {N1, N2, N3}, the charging pile occupancy data corresponding to each node is 0.2, 0.5, and 0.1, the charging service availability data is 1, 1, and 1, and the straight-line distance data from the starting node to each node is 2, 3, and 1.5. The priority calculation rule is set as: priority score = (1 - charging pile occupancy data) * 0.6 + (1 / straight-line distance data) * 0.4. The priority score of N1 is calculated as (1-0.2) * 0.6 + (1 / 2) * 0.4 = 0.48 + 0.2 = 0.68; the priority score of N2 is (1-0.5) * 0.6 + (1 / 3) * 0.4 = 0.3 + 0.13 = 0.43; and the priority score of N3 is (1-0.1) * 0.6 + (1 / 1.5) * 0.4 = 0.54 + 0.27 = 0.81. According to the priority score from high to low, the sorted nearby candidate node sequence is [N3, N1, N2].

[0135] Step S1353: Select the first nearby candidate node in the sorted nearby candidate node sequence as the current target node, and extract all connection edges and corresponding comprehensive cost weights between the starting node and the current target node from the node weight set to obtain a current node connection edge weight set.

[0136] In this step, the first node in the sorted nearby candidate node sequence is the current target node, and all connection edges and corresponding comprehensive cost weights between the starting node and the current target node need to be extracted. For example, the first node in the sorted nearby candidate node sequence is N3, and all connection edges between the starting node N0 and N3 are extracted from the node weight set, including the direct connection edge N0-N3 and the indirect connection edges N0-N1-N3 and N0-N2-N3; and the corresponding comprehensive cost weights are ω03, ω01+ω13, and ω02+ω23, respectively. After integration, the current node connection edge weight set is {N0-N3: ω03, N0-N1-N3: ω01+ω13, N0-N2-N3: ω02+ω23}.

[0137] Step S1354: Based on the current node connection edge weight set, generate all possible path schemes from the starting node to the current target node, each path scheme containing a sequence of connection edges and a sequence of corresponding comprehensive cost weights, to obtain a current target node path scheme set.

[0138] In this step, the path scheme needs to include the complete connection edge sequence and the corresponding comprehensive cost weight sequence. For example, the current node connection edge weight set is {N0-N3: ω03, N0-N1-N3: ω01+ω13, N0-N2-N3: ω02+ω23}, and the corresponding path schemes are: scheme 1 is N0→N3, the connection edge sequence is [N0-N3], and the comprehensive cost weight sequence is [ω03]; scheme 2 is N0→N1→N3, the connection edge sequence is [N0-N1, N1-N3], and the comprehensive cost weight sequence is [ω01, ω13]; and scheme 3 is N0→N2→N3, the connection edge sequence is [N0-N2, N2-N3], and the comprehensive cost weight sequence is [ω02, ω23]. After integration, the current target node path scheme set is {scheme 1: {connection edge sequence: [N0-N3], weight sequence: [ω03]}, scheme 2: {connection edge sequence: [N0-N1, N1-N3], weight sequence: [ω01, ω13]}, scheme 3: {connection edge sequence: [N0-N2, N2-N3], weight sequence: [ω02, ω23]}}.

[0139] Step S1355: Calculate the total cost of each path scheme in the current target node path scheme set, accumulate the comprehensive cost weights of the connection edges in the path scheme, obtain the path total cost corresponding to each path scheme, and generate the current target node path total cost set.

[0140] In this step, the path total cost is the sum of the comprehensive cost weights of the connection edges in the path scheme. For example, the weight sequence of scheme 1 is [ω03], and the total cost is ω03; the weight sequence of scheme 2 is [ω01, ω13], and the total cost is ω01+ω13; and the weight sequence of scheme 3 is [ω02, ω23], and the total cost is ω02+ω23. The current target node path total cost set is {scheme 1: ω03, scheme 2: ω01+ω13, scheme 3: ω02+ω23}.

[0141] Step S1356: Extract the minimum path total cost from the current target node path total cost set, determine it as the path total cost from the starting node to the current target node, and obtain the current node minimum total cost.

[0142] In this step, the minimum path total cost is the minimum value in the current target node path total cost set. For example, the current target node path total cost set is {scheme 1: 5, scheme 2: 7, scheme 3: 6}, and the minimum path total cost is 5, i.e., the current node minimum total cost is 5.

[0143] Step S1357: Determine whether there is an unprocessed nearby candidate node in the sorted nearby candidate node sequence. If there is, take the next nearby candidate node as a new current target node, and repeat the above steps of extracting the connection edge weight set to determining the minimum total cost of the current node until all nearby candidate nodes are processed.

[0144] In this step, each node in the sorted nearby candidate node sequence needs to be processed in turn. For example, the sorted nearby candidate node sequence is [N3, N1, N2], N3 has been processed, and the next step is to process N1: extract all connection edges from the starting node to N1 and the corresponding comprehensive cost weight, generate a path scheme set, calculate the total cost of each scheme, and obtain the minimum total cost from the starting node to N1; then process N2, repeat the above steps until all nodes are processed.

[0145] Step S1358: Collect the minimum total cost of each nearby candidate node to obtain a nearby candidate node total cost set.

[0146] In this step, the minimum total cost corresponding to each nearby candidate node needs to be collected. For example, the minimum total cost of the nearby candidate node N3 is 5, the minimum total cost of N1 is 6, and the minimum total cost of N2 is 8, so the nearby candidate node total cost set is {N3: 5, N1: 6, N2: 8}.

[0147] Step S1359: Extract the minimum node minimum total cost from the nearby candidate node total cost set, and determine it as the first comprehensive cost to the nearby charging station In this step, the first comprehensive cost is the minimum value in the nearby candidate node total cost set. For example, the nearby candidate node total cost set is {N3: 5, N1: 6, N2: 8}, and the first comprehensive cost is 5.

[0148] Step S13510: Extract all connection edges and corresponding comprehensive cost weights from the starting node to the fixed station node from the node weight set to obtain a fixed station connection edge weight set.

[0149] In this step, the fixed station connection edge weight set is all connection edges from the start node to the fixed station node and the corresponding comprehensive cost weight. For example, the start node is N0, the fixed station node is Nf, the connection edge information is N0-Nf, N0-N1-Nf, N0-N2-Nf, N0-N3-Nf, and the corresponding comprehensive cost weight is ω0f, ω01+ω1f, ω02+ω2f, ω03+ω3f. Then the fixed station connection edge weight set is {N0-Nf: ω0f, N0-N1-Nf: ω01+ω1f, N0-N2-Nf: ω02+ω2f, N0-N3-Nf: ω03+ω3f}.

[0150] Step S13511: Based on the fixed station connection edge weight set, all possible path schemes from the start node to the fixed station node are generated, each path scheme contains a continuous connection edge sequence and a corresponding comprehensive cost weight sequence, and a fixed station path scheme set is obtained.

[0151] In this step, the path scheme needs to contain all possible paths from the start node to the fixed station node. For example, the fixed station connection edge weight set is {N0-Nf: 10, N0-N1-Nf: 12, N0-N2-Nf: 11, N0-N3-Nf: 9}, and the corresponding path schemes are: scheme A is N0→Nf, the connection edge sequence is [N0-Nf], and the weight sequence is

[10] ; scheme B is N0→N1→Nf, the connection edge sequence is [N0-N1, N1-Nf], and the weight sequence is [5, 7]; scheme C is N0→N2→Nf, the connection edge sequence is [N0-N2, N2-Nf], and the weight sequence is [4, 7]; and scheme D is N0→N3→Nf, the connection edge sequence is [N0-N3, N3-Nf], and the weight sequence is [3, 6]. The generated fixed station path scheme set is {scheme A: {connection edge sequence: [N0-Nf], weight sequence:

[10] }, scheme B: {connection edge sequence: [N0-N1, N1-Nf], weight sequence: [5, 7]}, scheme C: {connection edge sequence: [N0-N2, N2-Nf], weight sequence: [4, 7]}, scheme D: {connection edge sequence: [N0-N3, N3-Nf], weight sequence: [3, 6]}}.

[0152] Step S13512: The total cost of each path scheme in the fixed station path scheme set is calculated, the comprehensive cost weight of each connection edge in the path scheme is accumulated, the path total cost corresponding to each path scheme is obtained, and a fixed station path total cost set is generated.

[0153] In this step, the path total cost is the sum of the comprehensive cost weights of each connection edge in the path scheme. For example, the weight sequence of scheme A is

[10] , and the total cost is 10; the weight sequence of scheme B is [5, 7], and the total cost is 12; the weight sequence of scheme C is [4, 7], and the total cost is 11; and the weight sequence of scheme D is [3, 6], and the total cost is 9. The generated fixed station path total cost set is {scheme A: 10, scheme B: 12, scheme C: 11, scheme D: 9}.

[0154] Step S13513: Extract the minimum path total cost from the fixed station path total cost set, and determine it as the second comprehensive cost of the return fixed station charging station In this step, the minimum path total cost is the minimum value in the fixed station path total cost set. For example, the fixed station path total cost set is {scheme A: 10, scheme B: 12, scheme C: 11, scheme D: 9}, and the minimum path total cost is 9, i.e. the second comprehensive cost is 9.

[0155] Through the above steps, the construction of the charging path graph and the calculation of the comprehensive cost are completed.

[0156] Step S140: Compare the first comprehensive cost with the second comprehensive cost, determine the charging path optimization strategy with better energy saving effect, and calculate the energy saving amount information of the charging path optimization strategy relative to another strategy.

[0157] In this embodiment, the charging path optimization strategy is determined by comparing the comprehensive cost and the calculated energy saving amount, and the specific process is as follows: Step S141: Compare the first comprehensive cost with the second comprehensive cost. If the first comprehensive cost is less than the second comprehensive cost, the strategy of going to the surrounding candidate charging station for charging is determined as the charging path optimization strategy with better energy saving effect, the difference between the second comprehensive cost and the first comprehensive cost is calculated as the comprehensive cost saving amount, and the difference between the road distance under the corresponding strategy is calculated as the road distance saving amount, the difference between the driving time is calculated as the driving time saving amount, and the difference between the power consumption cost is calculated as the power consumption saving amount. The comprehensive cost saving amount, the road distance saving amount, the driving time saving amount and the power consumption saving amount are integrated into the first energy saving amount information.

[0158] In the embodiment, if the first comprehensive cost S_min_near is less than the second comprehensive cost S0f, the charging path optimization strategy is to charge at the candidate charging station in the surrounding area; the comprehensive cost saving amount is ΔS=S0f-S_min_near; the road distance saving amount is ΔD=D0f-D_min_near (D0f is the road distance from the starting node to the fixed station node, and D_min_near is the minimum road distance from the starting node to the nearest candidate node); the travel time saving amount is ΔT=T0f-T_min_near (T0f is the estimated travel time from the starting node to the fixed station node, and T_min_near is the minimum estimated travel time from the starting node to the nearest candidate node); and the power consumption saving amount is ΔC=C0f-C_min_near (C0f is the power consumption cost from the starting node to the fixed station node, and C_min_near is the minimum power consumption cost from the starting node to the nearest candidate node).

[0159] For example, the first comprehensive cost S_min_near is 50, the second comprehensive cost S0f is 80, the comprehensive cost saving amount ΔS=80-50=30; the road distance saving amount ΔD=10-5=5; the travel time saving amount ΔT=20-10=10; and the power consumption saving amount ΔC=15-7.5=7.5; and the first energy saving amount information is [ΔS=30, ΔD=5, ΔT=10, ΔC=7.5].

[0160] In step S142, if the first comprehensive cost is not less than the second comprehensive cost, the strategy of returning to the fixed station charging station for charging is determined as the charging path optimization strategy with the more optimal energy saving effect, the difference between the first comprehensive cost and the second comprehensive cost is calculated as the comprehensive cost saving amount, the difference in road distance corresponding to the strategy is calculated as the road distance saving amount, the difference in travel time is calculated as the travel time saving amount, and the difference in power consumption cost is calculated as the power consumption saving amount, and the comprehensive cost saving amount, the road distance saving amount, the travel time saving amount, and the power consumption saving amount are integrated as the second energy saving amount information.

[0161] In the embodiment, if the first comprehensive cost S_min_near is not less than the second comprehensive cost S0f, the charging path optimization strategy is to return to the fixed station charging station for charging; the comprehensive cost saving amount is ΔS=S_min_near-S0f; the road distance saving amount is ΔD=D_min_near-D0f; the travel time saving amount is ΔT=T_min_near-T0f; and the power consumption saving amount is ΔC=C_min_near-C0f.

[0162] For example, the first comprehensive cost S_min_near is 70, and the second comprehensive cost S0f is 60, so the comprehensive cost saving amount AS = 70-60 = 10; the road distance saving amount AD = 8-6 = 2; the travel time saving amount AT = 16-12 = 4; the power consumption saving amount AC = 12-10 = 2; and the second energy consumption saving amount information is [AS = 10, AD = 2, AT = 4, AC = 2].

[0163] Step S143: extracting, from the historical and cost data of the multi-source operation data set, historical comprehensive cost data of the enterprise vehicle performing the charging path planning under the same operation condition in the past, the same operation condition including the same vehicle load range, the same weather condition type, and the same road section feature category.

[0164] In the embodiment, the same operation condition needs to be consistent with the current operation condition, for example, the current operation condition is (vehicle load range W_range, weather condition type Wt_type, road section feature category R_type), and the historical comprehensive cost data under the same operation condition (W_range, Wt_type, R_type) in the past needs to be extracted.

[0165] For example, the historical comprehensive cost data under the same operation condition in the past is [H1, H2,..., Hk].

[0166] Step S144: comparing the comprehensive cost corresponding to the currently determined charging path optimization strategy with the historical comprehensive cost data under the same operation condition in the past, calculating the difference between the current comprehensive cost and the historical comprehensive cost data as a historical comparison saving amount, and supplementing the historical comparison saving amount to the first energy consumption saving amount information or the second energy consumption saving amount information to obtain complete energy consumption saving amount information.

[0167] In the embodiment, the current comprehensive cost is the comprehensive cost corresponding to the charging path optimization strategy, for example, the charging path optimization strategy is to charge at a candidate charging station in the surrounding area, and the current comprehensive cost is S_min_near; and the historical comparison saving amount is AH = H_avg-S_min_near (H_avg is the average value of the historical comprehensive cost data under the same operation condition in the past).

[0168] For example, the average value H_avg of the historical comprehensive cost data under the same operation condition in the past is 70, and the current comprehensive cost S_min_near is 50, so the historical comparison saving amount AH = 70-50 = 20; and the AH is supplemented to the first energy consumption saving amount information to obtain complete energy consumption saving amount information [AS = 30, AD = 5, AT = 10, AC = 7.5, AH = 20].

[0169] Step S150: Based on the charging path optimization strategy and the energy consumption saving amount information, a charging path instruction containing a charging path node sequence is generated, and the charging path instruction is associated with the energy consumption saving amount information and output to the enterprise vehicle management terminal.

[0170] In this embodiment, the charging path instruction and the associated output are generated to realize the display and management of the charging path. The specific process is as follows: For example, step S151: target charging station node information is extracted from the charging path optimization strategy, the target charging station node information containing location coordinate data and basic information of the target charging station, and the basic information of the charging station containing available charging pile quantity data and charging service type data.

[0171] In this embodiment, the target charging station node information is the charging station information corresponding to the charging path optimization strategy. For example, if the charging path optimization strategy is to charge at the surrounding candidate charging station N1, the target charging station node information is (Lon1, Lat1), the basic information of the charging station is the available charging pile quantity 3, and the charging service type is direct current fast charging.

[0172] Step S152: All passable road node information from the current position node of the enterprise vehicle to the target charging station node is extracted from the environment and map data of the multi-source operation data set, the passable road node information containing location coordinate data of road nodes, connection relationship data between road nodes, and road attribute data corresponding to the road nodes.

[0173] In this embodiment, the passable road node information is the node information of all passable roads from the current position node of the enterprise vehicle to the target charging station node. For example, the current position node is N0, the target charging station node is N1, and the passable road node information is N0→A→B→N1, wherein A and B are road nodes, the connection relationship data is N0-A, A-B, and B-N1, and the road attribute data is the road slope data corresponding to the A node and the road congestion degree data corresponding to the B node.

[0174] Step S153: Based on the road distance data, the estimated driving time data, and the electricity consumption cost data in each passable road node information, and in combination with a preset path screening rule, an optimal road node sequence is screened from all passable road node combinations, the optimal road node sequence being a road node combination that comprehensively satisfies the shortest road distance, the least driving time, and the lowest electricity consumption cost.

[0175] In this embodiment, the path screening rule needs to consider the road distance, the driving time, and the electricity consumption cost to screen the optimal road node sequence.

[0176] For example, the passable road node combination is N0→A→B→N1, N0→C→D→N1, N0→E→F→N1, wherein the road distance of N0→A→B→N1 is the shortest, the driving time is the least, and the electricity consumption cost is the lowest, and thus the optimal road node sequence is N0→A→B→N1.

[0177] Step S154: arranging the current position node of the enterprise vehicle, the optimal road node sequence, and the target charging station node in the driving order to generate a charging path instruction containing a charging path node sequence, and extracting the driving direction indication data and the driving matters needing attention data between nodes from the environmental and map data of the multi-source operation data set, and supplementing them to the charging path instruction.

[0178] In this embodiment, the charging path node sequence is the current position node→the optimal road node sequence→the target charging station node, such as N0→A→B→N1; the driving direction indication data is the driving direction between nodes, such as eastward driving from N0 to A, southward driving from A to B, and westward driving from B to N1; and the driving matters needing attention data is the road matters needing attention between nodes, such as slow driving due to a school at the A node and detouring due to construction at the B node.

[0179] For example, the charging path instruction is “start from the current position N0, drive eastward to the A node, drive southward to the B node, and drive westward to the target charging station N1; matters needing attention: slow down at the A node due to a school, and detour at the B node due to construction”.

[0180] Step S155: extracting the idle electricity consumption ratio, the mileage utilization rate, and the battery health index from the key business index set, extracting the annual electricity consumption saving cost and the estimated length of battery life extension from the economic value estimation result, extracting the route optimization potential from the route optimization potential calculation result, and extracting the driver stability score from the driver stability calculation result.

[0181] In this embodiment, the key business index set is the index set generated in step S160, the economic value estimation result is the result generated in step S170, the route optimization potential calculation result is the result generated in step S180, and the driver stability calculation result is the result generated in step S190.

[0182] For example, the idle electricity consumption ratio is 0.2, the mileage utilization rate is 0.8, and the battery health index is 85; the annual electricity consumption saving cost is 10000 yuan, and the estimated length of battery life extension is 2 years; the route optimization potential is 0.6; and the driver stability score is 90.

[0183] Step S156: The energy saving amount information, the empty consumption ratio, the mileage utilization rate, the battery health index, the annual electricity consumption saving cost, the battery life extension estimated duration, the route optimization potential, and the driver stability score are associated with the charging path instruction to generate an associated output data set. The association is achieved by establishing a corresponding identifier relationship between each data and the charging path instruction.

[0184] In this embodiment, the association needs to establish a corresponding identifier relationship between each data and the charging path instruction. For example, the identifier of the charging path instruction is ID1, the identifier of the energy saving amount information is ID1, and the identifier of the empty consumption ratio is ID1.

[0185] For example, the associated output data set is {ID1:{charging path instruction: "start from the current location N0...", energy saving amount information: [ΔS=30, ΔD=5, ΔT=10, ΔC=7.5, ΔH=20], empty consumption ratio: 0.2, mileage utilization rate: 0.8, battery health index: 85, annual electricity consumption saving cost: 10000 yuan, battery life extension estimated duration: 2 years, route optimization potential: 0.6, driver stability score: 90}}.

[0186] Step S157: According to the data display format supported by the enterprise vehicle management terminal, the associated output data set is converted into a format recognizable by the terminal. The charging path instruction is displayed in the form of map node annotation and route drawing, and the energy saving amount information and various indicators are displayed in the form of data list combined with statistical charts.

[0187] In this embodiment, the format recognizable by the terminal needs to be consistent with the format supported by the enterprise vehicle management terminal. For example, the charging path instruction is converted into a map annotation format, the energy saving amount information is converted into a data list format, and various indicators are converted into a statistical chart format.

[0188] For example, the charging path instruction is converted into a map annotation format, and the N0, A, B, and N1 nodes are annotated on the map, and the route N0→A→B→N1 is drawn. The energy saving amount information is converted into a data list format, and ΔS=30, ΔD=5, ΔT=10, ΔC=7.5, and ΔH=20 are displayed. Various indicators are converted into statistical chart formats, such as a bar chart of the empty consumption ratio and a line chart of the mileage utilization rate.

[0189] Step S158: The converted associated output data set is sent to the enterprise vehicle management terminal through a data transmission protocol, so that the enterprise vehicle management terminal receives the converted associated output data set and displays the charging path instruction and the associated energy saving amount information and various indicators on the terminal interface, realizing the associated output of the charging path instruction and the related information.

[0190] In this embodiment, the data transmission protocol can adopt the HTTP protocol to send the converted associated output data set to the enterprise vehicle management terminal. After receiving, the enterprise vehicle management terminal displays the charging path instruction and the associated energy saving amount information and various indicators on the terminal interface.

[0191] For example, the interface of the enterprise vehicle management terminal displays the charging path instruction marked on the map, the data list of the energy saving amount information below, and the statistical chart of various indicators on the right side.

[0192] Step S160: Calculate the key business indicators of the enterprise vehicle based on the energy saving amount information and the multi-source operation data set.

[0193] This embodiment realizes the evaluation of the enterprise vehicle operation state by calculating the key business indicators, and the specific process is as follows: Step S161: Extract the electricity saving amount from the energy saving amount information, and extract the total electricity consumption data of the enterprise vehicle in the corresponding charging period from the historical and cost data of the multi-source operation data set, the corresponding charging period being the operation period from the last charging completion to the current charging need.

[0194] In this embodiment, the electricity saving amount is ΔC in the energy saving amount information, and the total electricity consumption data is the total electricity consumption in the corresponding charging period.

[0195] For example, the electricity saving amount ΔC is 7.5, and the total electricity consumption data in the corresponding charging period is 37.5.

[0196] Step S162: Calculate the idle consumption electricity ratio based on the electricity saving amount and the total electricity consumption data, the idle consumption electricity ratio being the ratio of the electricity saving amount to the total electricity consumption data, and the ratio being used to quantify the proportion of the idle consumption electricity reduced by the enterprise vehicle due to the charging path optimization in the total electricity consumption.

[0197] In this embodiment, the idle consumption electricity ratio Ew=ΔC / total electricity consumption data.

[0198] For example, the electricity saving amount ΔC is 7.5, and the total electricity consumption data is 37.5, so the idle consumption electricity ratio Ew=7.5 / 37.5=0.2.

[0199] Step S163: Extract the current battery state of charge data and the battery full state of charge data of the enterprise vehicle from the vehicle real-time state data of the multi-source operation data set, and calculate the estimated drivable mileage data in the current charging period in combination with the standard endurance data of the enterprise vehicle, the estimated drivable mileage data being the product of the standard endurance data and the proportion of the current battery state of charge data in the battery full state of charge data.

[0200] In the embodiment, the current battery state of charge data is SOC, the full battery state of charge data is SOC full, and the standard cruising data is R std, so the estimated driving range data R pred = R std * (SOC / SOC full).

[0201] For example, the current battery state of charge data SOC is 0.5, the full battery state of charge data SOC full is 1, and the standard cruising data R std is 200, so the estimated driving range data R pred = 200 * (0.5 / 1) = 100.

[0202] Step S164: Calculate the mileage utilization rate based on the estimated driving range data and the enterprise vehicle standard cruising data. The mileage utilization rate is the ratio of the estimated driving range data to the enterprise vehicle standard cruising data, which is used to evaluate the utilization degree of the vehicle cruising capability in the current charging period.

[0203] In the embodiment, the mileage utilization rate U = R pred / R std.

[0204] For example, the estimated driving range data R pred is 100, and the standard cruising data R std is 200, so the mileage utilization rate U = 100 / 200 = 0.5.

[0205] Step S165: Extract the mileage utilization rate data of the past multiple charging periods of the enterprise vehicle from the historical and cost data of the multi-source operation data set, calculate the average value of the mileage utilization rate data of the past multiple charging periods, calculate the difference between the mileage utilization rate data of each past charging period and the average value, square all the obtained difference values, and then calculate the average value to obtain the monthly stability variance, which is used to reflect the fluctuation of the mileage utilization rate.

[0206] In the embodiment, the mileage utilization rate data of the past multiple charging periods is [U1, U2,..., Um], the average value U avg = (U1+U2+...+Um) / m, and the monthly stability variance σ 2 = [(U1-U avg) 2 +(U2-U avg) 2 +...+(Um-U avg) 2] / m.

[0207] For example, the mileage utilization rate data of the past 3 charging periods is [0.6, 0.5, 0.4], the average value U avg = (0.6+0.5+0.4) / 3 = 0.5, and the monthly stability variance σ 2 = [(0.6-0.5) 2 +(0.5-0.5) 2 +(0.4-0.5) 2] / 3 = (0.01+0+0.01) / 3 ≈ 0.0067.

[0208] Step S166: Extract the enterprise vehicle battery cycle number data and the battery health reference cycle number data from the vehicle real-time state data of the multi-source operation data set, combine the idle power consumption ratio, and calculate the battery health correlation parameter through a preset correlation formula. The battery health correlation parameter is used to correlate the influence of the idle power consumption ratio and the battery cycle state on the battery health.

[0209] In this embodiment, the battery cycle number data is Cycles, the battery health reference cycle number data is Cycles_ref, and the preset correlation formula is BHI_rel=(1-Cycles / Cycles_ref)*(1-Ew).

[0210] For example, the battery cycle number data Cycles is 500, the battery health reference cycle number data Cycles_ref is 2000, and the idle power consumption ratio Ew is 0.2. Then the battery health correlation parameter BHI_rel=(1-500 / 2000)*(1-0.2)=0.75*0.8=0.6.

[0211] Step S167: Integrate the idle power consumption ratio, the mileage utilization rate, the monthly stability variance, and the battery health correlation parameter to generate a set of key business indicators of the enterprise vehicle. The set of key business indicators is used to comprehensively evaluate the energy consumption and energy saving effect and the operation state of the enterprise vehicle from multiple dimensions.

[0212] In this embodiment, the set of key business indicators is [Ew, U, σ², BHI_rel].

[0213] For example, the set of key business indicators is [0.2, 0.5, 0.0067, 0.6].

[0214] Step S170: Perform economic value estimation processing based on the set of key business indicators and the multi-source operation data set.

[0215] This embodiment realizes the evaluation of the operation cost of the enterprise vehicle by calculating the economic value. The specific process is as follows: Step S171: Extract the idle power consumption ratio and the mileage utilization rate from the set of key business indicators, and extract the enterprise vehicle annual average driving mileage data, the historical charging unit price data of each charging, and the charging power data corresponding to each charging from the historical and cost data of the multi-source operation data set.

[0216] In this embodiment, the idle power consumption ratio Ew is 0.2, the mileage utilization rate U is 0.5, the annual average driving mileage data is 10000, the historical charging unit price data of each charging is [1.5, 1.6, 1.4], and the charging power data corresponding to each charging is [50, 60, 40].

[0217] Step S172: Calculate a weighted average price based on the historical charging unit price data and the corresponding charging electricity quantity data, the weighted average price being a sum of products of each charging electricity quantity data and corresponding historical charging unit price data divided by a sum of each charging electricity quantity data, the weighted average price being used to reflect an actual electricity price level of charging of the enterprise vehicle.

[0218] In this embodiment, the weighted average price P=(Q1*C1+Q2*C2+...+Qn*Cn) / (Q1+Q2+...+Qn), wherein Qi is the charging electricity quantity data of the i-th charging, and Ci is the historical charging unit price data of the i-th charging.

[0219] For example, the historical charging unit price data of each charging is [1.5, 1.6, 1.4], and the corresponding charging electricity quantity data of each charging is [50, 60, 40], then the weighted average price P=(50*1.5+60*1.6+40*1.4) / (50+60+40)=(75+96+56) / 150=227 / 150≈1.51.

[0220] Step S173: Calculate an annual electricity consumption saving cost of the enterprise vehicle based on the idle consumption electricity consumption ratio, the annual average driving mileage data, the predicted unit electricity consumption, and the weighted average price, the annual electricity consumption saving cost being a product of the idle consumption electricity consumption ratio, the annual average driving mileage data, the predicted unit electricity consumption, and the weighted average price multiplied by an annual average charging cycle number of the enterprise vehicle, the annual average charging cycle number being extracted from the historical and cost data of the multi-source operation data set.

[0221] In this embodiment, the idle consumption electricity consumption ratio Ew is 0.2, the annual average driving mileage data is 10000, the predicted unit electricity consumption η_pred is 0.15, and the weighted average price P is 1.51; and the annual average charging cycle number is 200.

[0222] The annual electricity consumption saving cost=Ew*annual average driving mileage data*η_pred*P*annual average charging cycle number.

[0223] For example, the annual electricity consumption saving cost=0.2*10000*0.15*1.51*200=0.2*10000=2000; 2000*0.15=300; 300*1.51=453; 453*200=90600.

[0224] Step S174: Extracting the battery health related parameter from the key business indicator set, extracting the enterprise vehicle battery current capacity data and the battery initial capacity data from the vehicle real-time state data of the multi-source operation data set, calculating the battery capacity attenuation rate, which is the difference between the battery initial capacity data and the battery current capacity data divided by the battery initial capacity data, which reflects the degree of loss of the battery capacity.

[0225] In this embodiment, the battery health related parameter BHI_rel is 0.6; the battery current capacity data is 80, and the battery initial capacity data is 100; the battery capacity attenuation rate Dr=(C_initial-C_current) / C_initial.

[0226] For example, the battery initial capacity data C_initial is 100, and the battery current capacity data C_current is 80, then the battery capacity attenuation rate Dr=(100-80) / 100=0.2.

[0227] Step S175: Based on the battery health related parameter and the battery capacity attenuation rate, calculating the battery life extension estimated duration by a preset life estimation formula, which is constructed based on the slowing effect of the battery health related parameter on the battery attenuation and the correlation between the battery capacity attenuation rate and the battery life.

[0228] In this embodiment, the preset life estimation formula is ΔT=(1-Dr)*BHI_rel*T_ref, where T_ref is the battery reference life.

[0229] For example, the battery reference life T_ref is 5 years, the battery health related parameter BHI_rel is 0.6, and the battery capacity attenuation rate Dr is 0.2, then the battery life extension estimated duration ΔT=(1-0.2)*0.6*5=0.8*0.6*5=2.4 years.

[0230] Step S176: Extracting the enterprise vehicle battery replacement cost data and the battery maintenance cost data from the historical and cost data of the multi-source operation data set, calculating the indirect cost saving data corresponding to the battery life extension based on the battery life extension estimated duration, which is the saving amount allocated by the life extension proportion of the battery replacement cost data and the battery maintenance cost data.

[0231] In this embodiment, the battery replacement cost data is 20000, and the battery maintenance cost data is 5000; the battery life extension estimated duration ΔT is 2.4 years, and the battery reference life T_ref is 5 years; the indirect cost saving data=(battery replacement cost data+battery maintenance cost data)*(ΔT / T_ref).

[0232] For example, the battery replacement cost data is 20000, the battery maintenance cost data is 5000; the battery life extension estimated duration ΔT is 2.4 years, the battery reference life T_ref is 5 years; the indirect cost saving data is (20000+5000)*(2.4 / 5)=25000*0.48=12000.

[0233] Step S177: integrating the annual electricity consumption saving cost and the indirect cost saving data corresponding to the battery life extension, generating the economic value estimation result of the enterprise vehicle, which is used to quantify the economic benefits brought by the charging path optimization strategy.

[0234] In this embodiment, the economic value estimation result is [annual electricity consumption saving cost, indirect cost saving data].

[0235] For example, the economic value estimation result is [90600, 12000].

[0236] Step S180: calculating the route optimization potential based on the economic value estimation result and the multi-source operation data set.

[0237] This embodiment realizes the evaluation of the enterprise vehicle route optimization by calculating the route optimization potential, and the specific process is as follows: Step S181: extracting the indirect cost saving data corresponding to the annual electricity consumption saving cost and the battery life extension from the economic value estimation result, and adding the annual electricity consumption saving cost and the indirect cost saving data to obtain the annual total economic benefit data of the enterprise vehicle.

[0238] In this embodiment, the annual electricity consumption saving cost is 90600, and the indirect cost saving data is 12000; the annual total economic benefit data=90600+12000=102600.

[0239] Step S182: extracting the enterprise vehicle annual average charging path optimization frequency data from the historical and cost data of the multi-source operation data set, and dividing the annual total economic benefit data by the annual average charging path optimization frequency data to obtain the average economic benefit of single charging path optimization of the enterprise vehicle.

[0240] In this embodiment, the annual average charging path optimization frequency data is 200; the average economic benefit of single charging path optimization=102600 / 200=513.

[0241] Step S183: extracting vehicle total number data of the enterprise fleet and each vehicle annual average charging path optimization times data from the historical and cost data of the multi-source operation data set, and calculating enterprise fleet annual total charging path optimization times data based on the vehicle total number data and the each vehicle annual average charging path optimization times data, the enterprise fleet annual total charging path optimization times data being the sum of the each vehicle annual average charging path optimization times data.

[0242] In this embodiment, the vehicle total number data of the enterprise fleet is 100; the each vehicle annual average charging path optimization times data is [200, 180, …, 220] (a total of 100 data); and the enterprise fleet annual total charging path optimization times data = 200 + 180 + … + 220 = 20000.

[0243] Step S184: multiplying the enterprise fleet annual total charging path optimization times data and the average economic benefit of single charging path optimization to obtain annual total economic benefit estimation data of the enterprise fleet, which is used to estimate the annual economic benefit that the entire fleet can obtain through charging path optimization.

[0244] In this embodiment, the enterprise fleet annual total charging path optimization times data is 20000; the average economic benefit of single charging path optimization is 513; and the annual total economic benefit estimation data = 20000 * 513 = 10260000.

[0245] Step S185: extracting candidate charging station distribution density data and fixed site charging station distribution density data in the operation area of the enterprise fleet from the environment and map data of the multi-source operation data set, and calculating a charging station distribution difference coefficient, which is the ratio of the candidate charging station distribution density data and the fixed site charging station distribution density data, and is used to reflect the difference degree of the distribution of the two types of charging stations.

[0246] In this embodiment, the candidate charging station distribution density data is 5, and the fixed site charging station distribution density data is 1; and the charging station distribution difference coefficient = 5 / 1 = 5.

[0247] Step S186: calculating a route optimization potential base value based on the annual total economic benefit estimation data and the charging station distribution difference coefficient through a preset potential calculation formula, which is constructed based on the influence relationship of the annual total economic benefit estimation data and the charging station distribution difference on the route optimization potential.

[0248] In this embodiment, the preset potential calculation formula is ROI_base = annual total economic benefit estimation data * (charging station distribution difference coefficient / (charging station distribution difference coefficient + 1)).

[0249] For example, the total economic benefit estimation data is 10260000, and the charging station distribution difference coefficient is 5; ROI base = 10260000 * (5 / (5 + 1)) = 10260000 * 5 / 6 = 8550000.

[0250] Step S187: Extracting past route optimization implementation effect data of the enterprise vehicle fleet from the historical and cost data of the multi-source operation data set, the past route optimization implementation effect data including a ratio of actual economic benefits obtained by the past route optimization to estimated benefits.

[0251] In this embodiment, the past route optimization implementation effect data is 0.8 (ratio of actual benefits to estimated benefits).

[0252] Step S188: Adjusting the route optimization potential base value based on the past route optimization implementation effect data, multiplying the route optimization potential base value by the benefit ratio in the past route optimization implementation effect data to obtain the final route optimization potential, which is used to quantify the potential economic benefit scale that can be achieved by the enterprise vehicle fleet through charging path optimization.

[0253] In this embodiment, the final route optimization potential = 8550000 * 0.8 = 6840000.

[0254] Step S190: Calculating the driver stability based on the set of key business indicators and the multi-source operation data set.

[0255] This embodiment realizes the evaluation of the behavior of the enterprise vehicle driver by calculating the driver stability, and the specific process is as follows: Step S191: Extracting the monthly stability variance from the set of key business indicators, extracting past multiple monthly stability variance data of the corresponding driver of the enterprise vehicle from the historical and cost data of the multi-source operation data set, calculating the average value of the past multiple monthly stability variance data, and determining the average value as the stability reference value.

[0256] In this embodiment, the monthly stability variance σ² is 0.0067, the past multiple monthly stability variance data of the corresponding driver is [0.005, 0.007, 0.006], and the stability reference value is (0.005 + 0.007 + 0.006) / 3 = 0.006.

[0257] Step S192: Calculating the difference between the monthly stability variance and the stability reference value to obtain a stability deviation value, which is used to reflect the deviation degree of the current monthly stability from the average stability in the past.

[0258] In this embodiment, the stability deviation value = 0.0067 - 0.006 = 0.0007.

[0259] Step S193: Extracting vehicle speed fluctuation data, acceleration and deceleration frequency data and route deviation frequency data in the driving process of the corresponding driver from the vehicle real-time state data of the multi-source operation data set, the vehicle speed fluctuation data is the amplitude data of the speed change in the driving process, the acceleration and deceleration frequency data is the number of acceleration and deceleration operations in unit time, and the route deviation frequency data is the number of deviations from the planned route in the driving process.

[0260] In this embodiment, the vehicle speed fluctuation data is 5, the acceleration and deceleration frequency data is 10, and the route deviation frequency data is 2.

[0261] Step S194: Setting corresponding fluctuation weight coefficients for the vehicle speed fluctuation data, the acceleration and deceleration frequency data and the route deviation frequency data, and performing weighted sum processing on the vehicle speed fluctuation data, the acceleration and deceleration frequency data and the route deviation frequency data based on the fluctuation weight coefficients to obtain a driving behavior fluctuation coefficient, which is used to comprehensively reflect the fluctuation of the driving behavior of the driver.

[0262] In this embodiment, the fluctuation weight coefficients are 0.4, 0.3 and 0.3 respectively; the driving behavior fluctuation coefficient = 5*0.4 + 10*0.3 + 2*0.3 = 2 + 3 + 0.6 = 5.6.

[0263] Step S195: Setting a preset reference score of the driver's stability, calculating a corresponding deduction score based on the stability deviation value and the driving behavior fluctuation coefficient, and subtracting the deduction score from the preset reference score to obtain an initial value of the driver's stability, the deduction score is positively correlated with the stability deviation value and the driving behavior fluctuation coefficient.

[0264] In this embodiment, the preset reference score is 100; the deduction score = stability deviation value*10000 + driving behavior fluctuation coefficient*1; the deduction score = 0.0007*10000 + 5.6*1 = 7 + 5.6 = 12.6; the initial value of the driver's stability = 100 - 12.6 = 87.4.

[0265] Step S196: Extracting the number of times that the corresponding driver has followed the charging path optimization strategy and the total number of times that the driver has charged from the historical and cost data of the multi-source operation data set, and calculating a strategy compliance rate, which is the ratio of the number of times that the driver has followed the charging path optimization strategy to the total number of times that the driver has charged.

[0266] In this embodiment, the number of times that the corresponding driver has followed the charging path optimization strategy is 180, and the total number of times that the driver has charged is 200; the strategy compliance rate = 180 / 200 = 0.9.

[0267] Step S197: Determine a corresponding bonus score based on the policy compliance rate, the bonus score being positively correlated with the policy compliance rate, and add the bonus score to the initial driver stability value to obtain a final driver stability score, which is used to evaluate the performance of the enterprise vehicle driver in terms of driving behavior stability and charging path strategy execution stability.

[0268] In this embodiment, the bonus score = policy compliance rate * 10; the bonus score = 0.9 * 10 = 9; and the final driver stability score = 87.4 + 9 = 96.4.

[0269] Step S200: Calculate the battery health index based on the set of key business indicators and the set of multi-source operation data.

[0270] The embodiment calculates the battery health index to evaluate the health status of the enterprise vehicle battery, and the specific process is as follows: Step S201: Extract the battery health related parameter and the specific energy consumption ratio from the set of key business indicators, and extract the current temperature data, the current voltage data and the current current data of the enterprise vehicle battery from the real-time state data of the vehicle in the set of multi-source operation data, the current temperature data, the current voltage data and the current current data of the battery being real-time collected battery operation state data.

[0271] In this embodiment, the battery health related parameter BHI_rel is 0.6, and the specific energy consumption ratio Ew is 0.2; the current temperature data of the battery is 25℃, the current voltage data of the battery is 380V, and the current current data of the battery is 100A.

[0272] Step S202: Input the current temperature data, the current voltage data, the current current data of the battery, the battery health related parameter and the specific energy consumption ratio into the pre-trained battery health evaluation model, the battery health evaluation model being a machine learning model trained based on a large amount of battery historical health data, including a feature preprocessing layer, a feature weighting layer and an output layer.

[0273] In this embodiment, the pre-trained battery health evaluation model is a model trained based on a neural network, the feature preprocessing layer is used for preprocessing the input features, the feature weighting layer is used for weighting the preprocessed features, and the output layer is used for outputting the battery health index.

[0274] Step S203: Process the input current temperature data, the current voltage data, the current current data of the battery, the battery health related parameter and the specific energy consumption ratio through the feature preprocessing layer of the battery health evaluation model, convert different types of data into features of a unified dimension, generate a battery state feature vector, and the feature preprocessing includes data normalization and feature mapping.

[0275] In this embodiment, the feature preprocessing layer normalizes the current battery temperature data, the current battery voltage data, the current battery current data, the battery health-related parameter and the specific energy consumption ratio, and converts them into a unified dimension feature vector.

[0276] For example, the current battery temperature data of 25℃ is normalized to 0.5, the current battery voltage data of 380V is normalized to 0.8, the current battery current data of 100A is normalized to 0.6, the battery health-related parameter of 0.6 is normalized to 0.6, and the specific energy consumption ratio of 0.2 is normalized to 0.2; the battery state feature vector is [0.5, 0.8, 0.6, 0.6, 0.2].

[0277] Step S204: Based on the battery state feature vector, the feature weighting layer of the battery health assessment model gives each feature in the vector a corresponding health impact weight, which is determined by training based on the actual influence degree of each feature on the battery health state, and the feature weight value is large when the influence degree is large.

[0278] In this embodiment, the feature weighting layer gives each feature in the battery state feature vector a weight, such as the weight of the current battery temperature data is 0.3, the weight of the current battery voltage data is 0.2, the weight of the current battery current data is 0.2, the weight of the battery health-related parameter is 0.2, and the weight of the specific energy consumption ratio is 0.1.

[0279] Step S205: The output layer of the battery health assessment model performs comprehensive calculation on the weighted battery state feature vector, maps the weighted feature vector to the corresponding health score, and obtains the initial value of the battery health index, and the output layer uses a preset activation function to realize the conversion of the feature vector to the health score.

[0280] In this embodiment, the output layer uses a Sigmoid activation function to map the weighted battery state feature vector to a health score between 0 and 100.

[0281] For example, the weighted battery state feature vector is [0.5*0.3, 0.8*0.2, 0.6*0.2, 0.6*0.2, 0.2*0.1] = [0.15, 0.16, 0.12, 0.12, 0.02]; after comprehensive calculation, the initial value of the battery health index is 80.

[0282] Step S206: Extract the health index data of the enterprise vehicle battery at multiple time nodes from the historical and cost data of the multi-source operation data set, analyze the change trend of the past health index data, and check whether the initial value of the current battery health index conforms to the change trend.

[0283] In this embodiment, the health index data of the past multiple time nodes is [85, 83, 81], and the change trend is gradually decreasing; the initial value of the current battery health index is 80, which is consistent with the change trend of gradually decreasing.

[0284] Step S207: If the initial value of the current battery health index is consistent with the past change trend, the initial value is determined as the final battery health index; if not, the initial value of the battery health index is corrected based on the past change trend to obtain the final battery health index, which is used to intuitively quantify and evaluate the current health degree of the battery of the enterprise vehicle.

[0285] In this embodiment, the initial value of the current battery health index is 80, which is consistent with the past change trend, so the final battery health index is 80.

[0286] Based on the same inventive concept, please refer to Figure 2 , which shows the structural schematic block diagram of the charging path planning system 100 for energy consumption and energy saving of enterprise vehicles provided by the embodiment of the present application for executing the charging path planning method for energy consumption and energy saving of enterprise vehicles described above. The charging path planning system 100 for energy consumption and energy saving of enterprise vehicles can include a communication unit 110, a machine-readable storage medium 120, and a processor 130.

[0287] In this embodiment, the machine-readable storage medium 120 can also be integrated into the processor 130 and can communicate and interact with external systems through the communication unit 110. The machine-readable storage medium 120 is used to store machine-executable instructions for executing the scheme of the present application, and the processor 130 is used to execute the machine-executable instructions stored in the machine-readable storage medium 120 to implement the charging path planning method for energy consumption and energy saving of enterprise vehicles provided by the foregoing method embodiment.

[0288] It should be noted that, in order to simplify the description of the present application and help understand one or more embodiments of the present application, in the foregoing description of the embodiments of the present application, various features are sometimes combined into one embodiment, figure or description thereof.

Claims

1. A charging path planning method for energy conservation in enterprise vehicles, characterized in that, The method includes: Acquire a multi-source operational data set of the company's vehicles, the multi-source operational data set including real-time vehicle status data, environmental and map data, and historical and cost data; Based on the multi-source operational data set, unit power consumption prediction processing is performed to obtain the predicted unit power consumption of the company's vehicles under the current operating conditions; Construct a path graph that includes candidate charging station nodes and the vehicle's current location node, and calculate the first comprehensive cost of going to the nearest charging station and the second comprehensive cost of returning to the fixed charging station based on the predicted unit power consumption. By comparing the first comprehensive cost with the second comprehensive cost, a charging path optimization strategy with better energy saving effect is determined, and the energy saving information of the charging path optimization strategy relative to the other strategy is calculated. Based on the charging path optimization strategy and the energy saving information, a charging path instruction containing a sequence of charging path nodes is generated, and the charging path instruction is associated with the energy saving information and output to the enterprise vehicle management terminal.

2. The charging path planning method for energy conservation of enterprise vehicles according to claim 1, characterized in that, The step of performing unit power consumption prediction processing based on the multi-source operational data set to obtain the predicted unit power consumption of the company's vehicles under current operating conditions includes: Historical charging power data and corresponding charging cycle mileage data of the enterprise vehicles are extracted from the historical and cost data of the multi-source operation data set. Based on the historical charging power data and the charging cycle mileage data, the historical average unit power consumption of the enterprise vehicles is calculated. The historical average unit power consumption is the ratio of historical charging power data to corresponding charging cycle mileage data. Real-time vehicle load data, real-time vehicle speed data, and current battery charge status data are extracted from the vehicle real-time status data of the multi-source operation data set. Current temperature data, current weather condition data, and current road segment feature data are extracted from the environment and map data of the multi-source operation data set. The current road segment feature data includes road segment slope data, road segment congestion level data, and road surface material data. The historical average unit power consumption, real-time vehicle load data, real-time vehicle speed data, current battery charge status data, current temperature data, current weather condition data, and current road segment feature data are input into the pre-trained unit power consumption prediction model, which includes a feature fusion layer and a prediction output layer. The feature fusion layer of the unit power consumption prediction model performs feature association processing on the input historical average unit power consumption, real-time vehicle load data, real-time vehicle speed data, current battery charge status data, current temperature data, current weather condition data, and current road segment feature data to generate a power consumption influence feature vector that integrates the influence of multiple factors. The feature association processing includes dimensional unification of features of different data types and modeling of the relationship between features. Based on the power consumption impact feature vector, power consumption prediction calculation is performed through the prediction output layer of the unit power consumption prediction model. The prediction output layer converts the power consumption impact feature vector into the corresponding power consumption prediction value through a preset mapping relationship, thereby obtaining the predicted unit power consumption of the enterprise's vehicles under the current operating conditions.

3. The charging path planning method for energy conservation of enterprise vehicles according to claim 1, characterized in that, The construction of the path graph, which includes candidate charging station nodes and the vehicle's current location node, and the calculation of the first comprehensive cost of traveling to the nearest charging station and the second comprehensive cost of returning to the fixed charging station based on the predicted unit power consumption, includes: Extract the current location coordinates of the enterprise vehicle, the location coordinates of surrounding candidate charging stations, and the location coordinates of fixed charging stations from the environmental and map data of the multi-source operation data set. The location coordinates of surrounding candidate charging stations are the location coordinates of all available charging stations within a preset range of the current location of the enterprise vehicle. Using the current location coordinates of the enterprise vehicle as the starting node, the location coordinates of the surrounding candidate charging stations as the nearest candidate node, the location coordinates of the fixed charging station as the fixed station node, and the passable roads between each node as connecting edges, a charging path map for the enterprise vehicle is constructed. Each connecting edge in the charging path map for the enterprise vehicle corresponds to a passable road. The road distance data and estimated travel time data corresponding to each connecting edge are extracted from the environmental and map data of the multi-source operation data set. Combined with the predicted unit power consumption, the power consumption cost data corresponding to each connecting edge is calculated. The power consumption cost data is the product of the road distance data of the corresponding connecting edge and the predicted unit power consumption. The estimated travel time data is calculated based on the road distance data and the average travel speed data of the current road segment. The road distance data, estimated travel time data, and electricity consumption cost data are normalized respectively, and a corresponding weight coefficient is set for each standardized value. The standardized values ​​are then weighted and summed using the weight coefficients to obtain the comprehensive cost weight of each connecting edge. The weight coefficients are determined based on the importance that enterprises attach to distance, time, and electricity consumption during vehicle operation. Based on the comprehensive cost weight of each connecting edge in the charging path graph, a path search algorithm is used to calculate the total path cost from the starting node to each nearest candidate node. The minimum total path cost is selected as the first comprehensive cost to reach the nearest charging station. At the same time, the total path cost from the starting node to the fixed station node is calculated as the second comprehensive cost to return to the fixed station charging station. The path search algorithm is used to find the path with the minimum sum of comprehensive cost weights between nodes.

4. The charging path planning method for energy conservation of enterprise vehicles according to claim 3, characterized in that, Based on the comprehensive cost weights of each connecting edge in the charging path graph, a path search algorithm is used to calculate the total path cost from the starting node to each nearest candidate node, and the minimum total path cost is selected as the first comprehensive cost to reach the nearest charging station. Simultaneously, the total path cost from the starting node to the fixed charging station node is calculated as the second comprehensive cost to return to the fixed charging station, including: Obtain the charging path graph and the comprehensive cost weight of each connecting edge. Extract the starting node information, the nearest candidate node set information, the fixed station node information, and the connecting edge information between each node from the charging path graph. Integrate the starting node information, the nearest candidate node set information, the fixed station node information, the connecting edge information, and the comprehensive cost weight of each connecting edge to obtain the node weight set. The available status data of charging stations corresponding to each nearest candidate node is extracted from the environment and map data of the multi-source operation data set. The available status data of charging stations includes charging pile occupancy data and charging service availability data. Combined with the straight-line distance data from the starting node to each nearest candidate node, all the nearest candidate nodes in the nearest candidate node set are prioritized to obtain the sorted sequence of nearest candidate nodes. Select the first nearest candidate node from the sorted sequence of nearest candidate nodes as the current target node, and extract all the connecting edges between the starting node and the current target node and their corresponding comprehensive cost weights from the node weight set to obtain the current node connecting edge weight set. Based on the set of edge weights of the current node, generate all possible path schemes from the starting node to the current target node. Each path scheme contains a continuous sequence of edge connections and a corresponding sequence of comprehensive cost weights, thus obtaining the set of path schemes for the current target node. Calculate the total cost for each path scheme in the current target node path scheme set, sum the comprehensive cost weights of each connecting edge in the path scheme to obtain the total path cost corresponding to each path scheme, and generate the current target node path total cost set. Extract the minimum total path cost from the set of total path costs for the current target node, and determine it as the total path cost from the starting node to the current target node, thus obtaining the minimum total cost for the current node; Determine whether there are any unprocessed nearest candidate nodes in the sorted nearest candidate node sequence. If so, take the next nearest candidate node as the new current target node and repeat the above steps of extracting the set of connection edge weights to determine the minimum total cost of the current node until all nearest candidate nodes have been processed. Collect the minimum total cost of all nearby candidate nodes to obtain the set of total costs of nearby candidate nodes; Extract the minimum total cost of the node from the set of total costs of the nearest candidate nodes, and determine it as the first comprehensive cost to reach the nearest charging station. Extract all connection edges and their corresponding comprehensive cost weights between the starting node and the fixed station node from the node weight set to obtain the fixed station connection edge weight set. Based on the set of fixed station connection edge weights, all possible path schemes from the starting node to the fixed station node are generated. Each path scheme contains a continuous sequence of connection edges and a corresponding sequence of comprehensive cost weights, thus obtaining a set of fixed station path schemes. The total cost is calculated for each path scheme in the set of fixed station path schemes. The comprehensive cost weights of each connecting edge in the path scheme are accumulated to obtain the total path cost corresponding to each path scheme, and a set of total path costs for fixed stations is generated. Extract the minimum total path cost from the set of total path costs for fixed stations and determine it as the second comprehensive cost for returning to the fixed station charging station.

5. The charging path planning method for energy conservation of enterprise vehicles according to claim 1, characterized in that, The step of comparing the first comprehensive cost with the second comprehensive cost to determine the charging path optimization strategy with better energy-saving effect, and calculating the energy saving information of the charging path optimization strategy relative to the other strategy, includes: The first comprehensive cost is compared with the second comprehensive cost. If the first comprehensive cost is less than the second comprehensive cost, the strategy of going to a nearby candidate charging station for charging is determined as the charging path optimization strategy with better energy saving effect. The difference between the second comprehensive cost and the first comprehensive cost is calculated as the comprehensive cost saving amount. At the same time, the difference in road distance under the corresponding strategy is calculated as the road distance saving amount, the difference in travel time is calculated as the travel time saving amount, and the difference in electricity cost is calculated as the electricity saving amount. The comprehensive cost saving amount, road distance saving amount, travel time saving amount and electricity saving amount are integrated into the first energy saving amount information. If the first comprehensive cost is not less than the second comprehensive cost, then the strategy of returning to a fixed charging station for charging is determined as the charging path optimization strategy with better energy saving effect. The difference between the first comprehensive cost and the second comprehensive cost is calculated as the comprehensive cost saving amount. At the same time, the difference in road distance under the corresponding strategy is calculated as the road distance saving amount, the difference in travel time is calculated as the travel time saving amount, and the difference in electricity cost is calculated as the electricity saving amount. The comprehensive cost saving amount, road distance saving amount, travel time saving amount and electricity saving amount are integrated into the second energy saving amount information. Extract historical comprehensive cost data of enterprise vehicles performing charging route planning under the same past operating conditions from the historical and cost data of the multi-source operation data set. The same operating conditions include the same vehicle load range, the same weather condition type, and the same road segment feature category. The comprehensive cost corresponding to the current determined charging path optimization strategy is compared with the historical comprehensive cost data under the same operating conditions. The difference between the current comprehensive cost and the historical comprehensive cost data is calculated as the historical comparison saving amount. The historical comparison saving amount is added to the first energy saving amount information or the second energy saving amount information to obtain complete energy saving amount information.

6. The charging path planning method for energy conservation of enterprise vehicles according to claim 5, characterized in that, The method further includes calculating key business indicators for the company's vehicles based on the energy savings information and multi-source operational data sets. This step includes: The energy savings are extracted from the energy savings information, and the total energy consumption data of the enterprise vehicles in the corresponding charging cycle is extracted from the historical and cost data of the multi-source operation data set. The corresponding charging cycle is the operation cycle from the completion of the last charging to the current charging requirement. The energy consumption saving amount and the total energy consumption data are used to calculate the energy consumption ratio. The energy consumption ratio is the ratio of energy consumption saving amount to total energy consumption data. This ratio is used to quantify the proportion of energy consumption saved by the enterprise's vehicles due to the optimization of charging routes to the total energy consumption. Extract the current battery state of charge data and the battery state of full charge data of the enterprise vehicle from the real-time vehicle status data of the multi-source operation data set, and combine them with the standard range data of the enterprise vehicle to calculate the estimated driving range data in the current charging cycle. The estimated driving range data is the product of the standard range data and the proportion of the current battery state of charge data to the battery state of full charge data. The mileage utilization rate is calculated based on the estimated mileage data and the standard range data of the enterprise vehicle. The mileage utilization rate is the ratio of the estimated mileage data to the standard range data of the enterprise vehicle. This ratio is used to evaluate the utilization of the vehicle's range capacity during the current charging cycle. The mileage utilization rate data of the enterprise's vehicles in the past multiple charging cycles are extracted from the historical and cost data of the multi-source operation data set. The average value of the mileage utilization rate data in the past multiple charging cycles is calculated. The difference between the mileage utilization rate data of each past charging cycle and the average value is calculated. The average value of all the differences is then calculated to obtain the monthly stability variance. The monthly stability variance is used to reflect the fluctuation of mileage utilization rate. The battery cycle count data and battery health baseline cycle count data of the enterprise vehicle are extracted from the real-time vehicle status data of the multi-source operation data set. Combined with the energy consumption ratio, the battery health correlation parameters are calculated by a preset correlation formula. The battery health correlation parameters are used to correlate and reflect the impact of energy consumption ratio and battery cycle status on battery health. The energy consumption ratio, mileage utilization rate, monthly stability variance and battery health-related parameters are integrated to generate a set of key business indicators for enterprise vehicles. The set of key business indicators is used to comprehensively evaluate the energy consumption and energy-saving effect and operating status of enterprise vehicles from multiple dimensions. Furthermore, the method also includes calculating driver stability based on the set of key business indicators and the multi-source operational data set, a step that includes: The monthly stability variance is extracted from the set of key business indicators. The historical and cost data of the drivers corresponding to the company's vehicles are extracted from the set of multi-source operation data. The average value of the historical monthly stability variance data is calculated and the average value is determined as the stability benchmark value. The monthly stability variance is calculated by comparing it with the stability benchmark value to obtain the stability deviation value, which is used to reflect the degree of deviation between the current monthly stability and the historical average stability. Extract vehicle speed fluctuation data, acceleration and deceleration frequency data, and route deviation data corresponding to the driver's driving process from the real-time vehicle status data of the multi-source operation data set. The vehicle speed fluctuation data is the magnitude of speed change during driving, the acceleration and deceleration frequency data is the number of acceleration and deceleration operations per unit time, and the route deviation data is the number of times the driver deviates from the planned route during driving. For vehicle speed fluctuation data, acceleration and deceleration frequency data, and route deviation number data, corresponding fluctuation weight coefficients are set respectively. Based on the fluctuation weight coefficients, the vehicle speed fluctuation data, acceleration and deceleration frequency data, and route deviation number data are weighted and summed to obtain the driving behavior fluctuation coefficient. This driving behavior fluctuation coefficient is used to comprehensively reflect the fluctuation of the driver's driving behavior. A preset baseline score for driver stability is set. Based on the stability deviation value and the driving behavior fluctuation coefficient, the corresponding deduction score is calculated. The preset baseline score is subtracted from the deduction score to obtain the initial value of driver stability. The deduction score is positively correlated with the stability deviation value and the driving behavior fluctuation coefficient. Extract the number of times a driver has followed the charging route optimization strategy and the total number of charging times from the historical and cost data of the multi-source operation data set, and calculate the strategy compliance rate, which is the ratio of the number of times the charging route optimization strategy has been followed to the total number of charging times. The corresponding bonus score is determined based on the policy compliance rate. The bonus score is positively correlated with the policy compliance rate. The initial driver stability value is added to the bonus score to obtain the final driver stability score. The driver stability score is used to evaluate the performance of the company's vehicle drivers in terms of driving behavior stability and charging route strategy execution stability.

7. The charging path planning method for energy conservation of enterprise vehicles according to claim 6, characterized in that, The method further includes performing economic value estimation processing based on the set of key business indicators and the multi-source operational data set. This step includes: Extract the energy consumption ratio and mileage utilization rate from the set of key business indicators; extract the annual average mileage data of enterprise vehicles, the historical charging unit price data for each charge, and the corresponding charging volume data for each charge from the historical and cost data of the multi-source operation data set. The weighted average electricity price is calculated based on the historical charging unit price data and the corresponding charging volume data. The weighted average electricity price is the sum of the products of each charging volume data and the corresponding historical charging unit price data divided by the sum of each charging volume data. This weighted average electricity price is used to reflect the actual electricity price level of the company's vehicle charging. Based on the aforementioned energy consumption ratio, annual average mileage data, predicted unit energy consumption, and weighted average electricity price, the annual energy consumption savings of the company's vehicles are calculated. The annual energy consumption savings are calculated by multiplying the energy consumption ratio, annual average mileage data, predicted unit energy consumption, and weighted average electricity price by the annual average number of charging cycles for the company's vehicles. The annual average number of charging cycles is extracted from historical and cost data from a multi-source operational data set. Battery health-related parameters are extracted from the set of key business indicators. The current battery capacity data and initial battery capacity data of the enterprise vehicle are extracted from the real-time vehicle status data of the multi-source operation data set. The battery capacity decay rate is calculated. The battery capacity decay rate is the difference between the initial battery capacity data and the current battery capacity data divided by the initial battery capacity data. This battery capacity decay rate is used to reflect the degree of battery capacity loss. Based on the battery health-related parameters and the battery capacity decay rate, the estimated battery life extension time is calculated using a preset life estimation formula. The life estimation formula is constructed based on the mitigating effect of the battery health-related parameters on battery decay and the correlation between the battery capacity decay rate and battery life. Extract enterprise vehicle battery replacement cost data and battery maintenance cost data from the historical and cost data of the multi-source operation data set, and calculate the indirect cost savings data corresponding to the battery life extension based on the estimated battery life extension duration. The indirect cost savings data is the amount of savings allocated to the battery replacement cost data and battery maintenance cost data according to the life extension ratio. The annual energy consumption savings and the indirect cost savings corresponding to the extended battery life are integrated to generate an economic value estimation result for the company's vehicles. The economic value estimation result is used to quantify the economic benefits brought about by the charging path optimization strategy.

8. The charging path planning method for energy conservation of enterprise vehicles according to claim 6, characterized in that, The method further includes calculating a battery health index based on the set of key business indicators and the multi-source operational data set. This step includes: Battery health-related parameters and energy consumption ratio are extracted from the set of key business indicators. The current battery temperature, current battery voltage, and current battery current data of the enterprise vehicle are extracted from the real-time vehicle status data of the multi-source operation data set. The current battery temperature, current battery voltage, and current battery current data are real-time collected battery operating status data. The current battery temperature data, current battery voltage data, current battery current data, battery health-related parameters, and energy consumption ratio are input into a pre-trained battery health assessment model. The battery health assessment model is a machine learning model trained based on a large amount of historical battery health data, and includes a feature preprocessing layer, a feature weighting layer, and an output layer. The battery health assessment model uses a feature preprocessing layer to process the input battery current temperature data, battery current voltage data, battery current current data, battery health related parameters, and energy consumption ratio, converting different types of data into features of a unified dimension to generate a battery state feature vector. The feature preprocessing includes data normalization and feature mapping. Based on the battery state feature vector, each feature in the vector is assigned a corresponding health impact weight by the feature weighting layer of the battery health assessment model. The health impact weight is determined by training based on the actual degree of influence of each feature on the battery health state, and features with greater influence have larger weight values. The output layer of the battery health assessment model performs a comprehensive calculation on the weighted battery state feature vector, maps the weighted feature vector to the corresponding health score, and obtains the initial value of the battery health index. The output layer uses a preset activation function to realize the conversion of feature vector to health score. Extract health index data of the enterprise's vehicle battery from historical and cost data of the multi-source operation data set, analyze the changing trend of the past health index data, and check whether the current battery health index initial value conforms to the changing trend. If the initial value of the current battery health index conforms to the past trend, it is determined as the final battery health index; if it does not conform, the initial value of the battery health index is corrected based on the past trend to obtain the final battery health index. The battery health index is used to intuitively and quantitatively assess the current health status of the company's vehicle batteries.

9. The charging path planning method for energy conservation of enterprise vehicles according to claim 7, characterized in that, The method further includes calculating the route optimization potential based on the economic value estimation results and the multi-source operational data set; this step includes: Extract the annual electricity consumption savings and the indirect cost savings corresponding to the battery life extension from the economic value estimation results, and add the annual electricity consumption savings and indirect cost savings to obtain the company's total annual economic benefits for vehicles. Extract the annual average number of times the enterprise's vehicle charging route is optimized from the historical and cost data of the multi-source operation data set, and divide the annual total economic benefit data by the annual average number of times the charging route is optimized to obtain the average economic benefit of a single charging route optimization for the enterprise's vehicle. Extract the total number of vehicles in the enterprise fleet and the average number of times the charging path is optimized annually for each vehicle from the historical and cost data of the multi-source operation data set. Calculate the total number of times the enterprise fleet is optimized annually based on the total number of vehicles and the average number of times the charging path is optimized annually for each vehicle. The total number of times the enterprise fleet is optimized annually is the sum of the average number of times the charging path is optimized annually for each vehicle. Multiply the total number of times the company's fleet has optimized charging routes in a year by the average economic benefit of each charging route optimization to obtain the estimated total economic benefit of the company's fleet in a year. This estimated total economic benefit is used to estimate the annual economic benefit that the entire fleet can obtain through charging route optimization. From the environmental and map data of the multi-source operation data set, candidate charging station distribution density data and fixed station charging station distribution density data within the enterprise fleet's operating area are extracted, and the charging station distribution difference coefficient is calculated. The charging station distribution difference coefficient is the ratio of candidate charging station distribution density data to fixed station charging station distribution density data. This charging station distribution difference coefficient is used to reflect the degree of difference between the two types of charging station distribution. Based on the estimated annual total economic benefits and the charging station distribution difference coefficient, the basic value of route optimization potential is calculated using a preset potential calculation formula. The potential calculation formula is constructed based on the relationship between the estimated annual total economic benefits and the charging station distribution difference on the route optimization potential. Extract the past route optimization implementation effect data of the enterprise fleet from the historical and cost data of the multi-source operation data set. The past route optimization implementation effect data includes the ratio of the actual economic benefits obtained from past route optimization to the estimated benefits. Based on the past route optimization implementation effect data, the baseline value of route optimization potential is adjusted. The baseline value of route optimization potential is multiplied by the benefit ratio in the past route optimization implementation effect data to obtain the final route optimization potential. The route optimization potential is used to quantify the potential economic benefits that the company's fleet can achieve through charging route optimization.

10. A charging path planning system for energy conservation in enterprise vehicles, characterized in that, include: processor; A machine-readable storage medium for storing machine-executable instructions of the processor; The processor is configured to execute the charging path planning method for energy conservation of enterprise vehicles as described in any one of claims 1 to 9 by executing the machine-executable instructions.

Images