Method and system for predicting energy consumption of operating vehicle based on multi-source data fusion

Abstract

The application discloses a method and system for predicting energy consumption of a commercial vehicle based on multi-source data fusion, relates to the technical field of Internet of Vehicles data processing, and comprises the following steps: first, dividing effective operation data of the commercial vehicle according to changes in operation tasks, and second, dividing the data according to changes in road energy consumption states to form a time-space segment index table of the commercial vehicle; and then, performing slope driving coupling analysis on each road energy consumption segment according to the time-space segment index table of the commercial vehicle to obtain load inversion rules, and correcting an explicit load state to generate a road section level energy consumption influence state vector. The application performs slope driving coupling analysis on road energy consumption segments through the time-space segment index table of the commercial vehicle, establishes load inversion rules and corrects the explicit load state, and thus accurately maps the load influence to the road section level energy consumption influence state vector.

Description

Technical Field

[0001] This invention relates to the field of vehicle network data processing technology, and in particular to a method and system for predicting the energy consumption of operating vehicles based on multi-source data fusion. Background Technology

[0002] Energy consumption prediction for commercial vehicles falls under the technical field of vehicle operation data processing and intelligent transportation energy consumption management. It typically describes the vehicle operation process using onboard control data, positioning data, load record data, road condition data, and traffic and meteorological data. The data is then processed, segmented, and feature extracted at a unified time scale. Combined with historical energy consumption records, an energy consumption prediction relationship is established to obtain the energy consumption trend of the vehicle in the current driving segment or the remaining operating task.

[0003] In conventional methods, the energy consumption prediction of operating vehicles often uses fixed time windows or fixed mileage windows as prediction units, which makes it difficult to accurately correspond to changes in operating tasks caused by loading, unloading, parking, and route switching. At the same time, conventional methods usually directly use explicit load records or superficially similar historical segments for prediction, which makes it difficult to handle prediction biases caused by load record lag and inconsistent residual causes of similar segments. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a method for predicting the energy consumption of operating vehicles based on multi-source data fusion to solve the problem of insufficient prediction accuracy caused by changes in operating tasks and load state deviations in the energy consumption prediction of operating vehicles.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, the present invention provides a method for predicting the energy consumption of operating vehicles based on multi-source data fusion, which includes: collecting multi-source operating data of operating vehicles, performing time synchronization, reliable fusion and anomaly correction on the multi-source operating data, and generating effective operating data of operating vehicles. The effective operating data of operating vehicles is first segmented according to changes in operating tasks, and then further segmented according to changes in road energy consumption status to form a spatiotemporal segment index table of operating vehicles. Based on the spatiotemporal segment index table of operating vehicles, slope-driven coupling analysis is performed on each road energy consumption segment to obtain load inversion rules, and explicit load states are inverted and corrected to generate road segment-level energy consumption influence state vectors. Residual cause analysis was performed on the energy consumption impact state vector at the road segment level, an energy consumption residual cause-effect graph was constructed, and only similar operating segments with consistent residual causes were migrated to obtain the initial dual-scale energy consumption prediction of operating vehicles. Based on the vehicle's historical residual drift, driver energy consumption offset, and load correction deviation, individualized residual compensation is performed on the initial dual-scale energy consumption prediction of operating vehicles to generate the corrected dual-scale energy consumption prediction of operating vehicles.

[0007] As a preferred embodiment of the multi-source data fusion-based energy consumption prediction method for operating vehicles described in this invention, the multi-source operating data includes vehicle control data, positioning data, load record data, scheduling task data, road condition data, and traffic meteorological data.

[0008] As a preferred embodiment of the multi-source data fusion-based energy consumption prediction method for operating vehicles according to the present invention, the steps of performing time synchronization, reliable fusion, and anomaly correction on multi-source operating data to generate effective operating data for operating vehicles are as follows: A unified timeline mapping is performed on various acquisition times in the multi-source operational data, and time compensation is performed based on the offset relationship between the data generation time and the data reception time to form time-synchronized multi-source operational data; Data credibility is determined based on the data integrity and temporal consistency of time-synchronized multi-source operational data, and stable correlated data is screened through credibility gating to form reliable operational data; The system performs state trajectory consistency verification on the reliable operation data, eliminates abnormal disturbances that deviate from the vehicle's dynamic evolution and operational task progress, and generates valid operation data for the operating vehicles.

[0009] As a preferred embodiment of the multi-source data fusion-based method for predicting the energy consumption of operating vehicles according to the present invention, the steps of first segmenting the effective operating data of operating vehicles according to changes in operating tasks, and second segmenting it according to changes in road energy consumption status to form a spatiotemporal segment index table for operating vehicles are as follows: Based on the changes in operational tasks in the effective operational data of operating vehicles, the location of task changes is identified, and the effective operational data of operating vehicles is segmented for the first time to form operational task segments. Based on the changes in road energy consumption status in the operation task segments, identify the energy consumption boundary positions corresponding to the energy consumption abrupt change characteristics, and then perform secondary segmentation of the operation task segments according to the energy consumption boundary positions to form road energy consumption segments. Based on the vehicle operating status and external driving status within the road energy consumption segment, the road energy consumption segment is marked with its operating status to form a road energy consumption segment, and then indexed and organized in chronological order to form a spatiotemporal segment index table for operating vehicles.

[0010] As a preferred embodiment of the multi-source data fusion-based method for predicting the energy consumption of operating vehicles according to the present invention, the steps for obtaining load inversion rules by performing slope-driven coupling analysis on each road energy consumption segment are as follows: Based on the spatiotemporal segment index table of operating vehicles, the time range and operation status marker of each road energy consumption segment are located, and the slope change, drive response and energy consumption per unit distance are extracted from the effective operation data of operating vehicles to form segment energy consumption coupling relationship data. Based on the analysis of the energy consumption coupling relationship data of the segments, the combined impact of slope change on driving response and energy consumption per unit distance is analyzed, and the combined impact is mapped to the explicit load state to establish load inversion rules.

[0011] As a preferred embodiment of the multi-source data fusion-based energy consumption prediction method for operating vehicles described in this invention, the explicit load state is inverted and corrected to generate a road segment-level energy consumption influence state vector, the steps of which are as follows: The explicit load state is inverted and corrected using load inversion rules. Load offsets that do not match the segment energy consumption coupling data in the explicit load state are identified, and load correction state data is generated. The load correction status data is embedded into the operating status marker of the road energy consumption segment, and the load impact and road segment energy consumption sensitivity are correlated and encoded to form a road segment-level energy consumption impact status vector.

[0012] As a preferred embodiment of the multi-source data fusion-based energy consumption prediction method for operating vehicles described in this invention, the steps of performing residual cause analysis on the road segment-level energy consumption influence state vector and constructing an energy consumption residual cause-effect graph are as follows: Based on the deviation comparison between the road segment-level energy consumption influence state vector and the historical energy consumption benchmark of similar road energy consumption segments, the energy consumption residual data of the current road energy consumption segment is determined. Counterfactual residual attribution and time-series lag verification are used to conduct residual cause analysis on energy consumption residual data, establish causal relationships between energy consumption residual data and energy consumption disturbance factors, and construct an energy consumption residual causal graph.

[0013] As a preferred embodiment of the multi-source data fusion-based energy consumption prediction method for operating vehicles according to the present invention, the steps for obtaining the initial dual-scale energy consumption prediction of operating vehicles are as follows: Based on the energy consumption residual cause-effect graph, the residual causes of historical operation segments are matched, and similar operation segments with the same residual causes as the current road energy consumption segment are selected to form a set of similar operation segments. Based on the energy consumption residual cause-effect diagram, the weight of the residual cause consistency is determined, and the segment energy consumption increment is embedded into the road segment-level energy consumption influence state vector according to the load sensitivity correction coefficient to calculate the initial short-term energy consumption prediction value. The remaining road energy consumption segments are read from the spatiotemporal segment index table of operating vehicles, and the energy consumption increments of similar segments of the remaining road energy consumption segments are recursively accumulated according to the weight of the consistency of residual causes to calculate the initial predicted value of the remaining operating task energy consumption.

[0014] As a preferred embodiment of the multi-source data fusion-based energy consumption prediction method for operating vehicles described in this invention, the steps for performing individualized residual compensation on the initial dual-scale energy consumption prediction of operating vehicles to generate a corrected dual-scale energy consumption prediction of operating vehicles are as follows: Based on the change in the initial dual-scale energy consumption prediction of operating vehicles and the historical prediction deviation of current vehicles in the same type of road energy consumption segment, determine the vehicle residual drift compensation amount corresponding to the historical residual drift of the vehicle. The vehicle residual drift compensation amount is superimposed and verified with the driver energy consumption offset formed by the current driver under the same operating conditions to obtain the driver-related residual compensation amount. By matching the driver-related residual compensation amount with the load correction deviation, the effect of vehicle load changes on the increase or decrease of the driver-related residual compensation amount is determined, and an individualized residual compensation amount is obtained. The individualized residual compensation amount is used to synchronously compensate the initial two-scale energy consumption prediction of operating vehicles, thereby generating the corrected two-scale energy consumption prediction of operating vehicles.

[0015] Secondly, the present invention provides an energy consumption prediction system for operating vehicles based on multi-source data fusion, comprising: a data acquisition module, which collects multi-source operating data of operating vehicles, performs time synchronization, reliable fusion and anomaly correction on the multi-source operating data, and generates effective operating data of operating vehicles; The operation time-space segmentation module divides the effective operation data of operating vehicles into the first segment according to the changes in operation tasks, and the second segment according to the changes in road energy consumption status, forming an index table of operating vehicle time-space segments. The load inversion correction module, based on the spatiotemporal segment index table of operating vehicles, performs slope-driven coupling analysis on each road energy consumption segment to obtain load inversion rules, and performs inversion correction on explicit load states to generate a road segment-level energy consumption influence state vector. The residual causal migration prediction module performs residual cause analysis on the energy consumption influence state vector at the road segment level, constructs an energy consumption residual causal graph, and migrates only similar running segments with consistent residual causes to obtain the initial dual-scale energy consumption prediction of operating vehicles. The individualized residual compensation module performs individualized residual compensation on the initial dual-scale energy consumption prediction of operating vehicles based on vehicle historical residual drift, driver energy consumption offset, and load correction deviation, generating the corrected dual-scale energy consumption prediction of operating vehicles.

[0016] The beneficial effects of this invention are as follows: by performing slope-driven coupling analysis on road energy consumption segments through the spatiotemporal segment index table of operating vehicles, load inversion rules are established and explicit load states are corrected, achieving accurate mapping of load influence to the segment-level energy consumption influence state vector; by screening similar operating segments with consistent residual causes through the energy consumption residual causal graph, the causal constraint transfer of the initial operating vehicle dual-scale energy consumption prediction is realized, and combined with individualized residual compensation, the corrected operating vehicle dual-scale energy consumption prediction is generated, which can be used for short-term energy consumption and remaining operating task energy consumption prediction of operating vehicles, improving the prediction pertinence and operational adaptability. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart of a method for predicting the energy consumption of operating vehicles based on multi-source data fusion.

[0019] Figure 2 This is a schematic diagram of an energy consumption prediction system for operating vehicles based on multi-source data fusion.

[0020] Figure 3 This is a flowchart for individualized residual compensation.

[0021] Figure 4 This is a schematic diagram illustrating the principle of load inversion correction and residual causal transfer. Detailed Implementation

[0022] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0023] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0024] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0025] Reference Figures 1-4This is one embodiment of the present invention, which provides a method for predicting the energy consumption of operating vehicles based on multi-source data fusion, including the following steps: S1. Collect multi-source operation data of operating vehicles, perform time synchronization, reliable fusion and anomaly correction on the multi-source operation data, and generate valid operation data of operating vehicles.

[0026] Multi-source operational data includes vehicle control data, positioning data, payload recording data, scheduling task data, road condition data, and traffic and meteorological data; Furthermore, vehicle control data and positioning data are collected in real time through the vehicle control unit and positioning terminal of the operating vehicle; load recording data and scheduling task data are obtained through the vehicle weighing device and vehicle scheduling management terminal; and road condition data and traffic meteorological data are obtained through electronic maps, road perception interfaces and traffic meteorological data interfaces.

[0027] A unified timeline mapping is performed on various acquisition times in the multi-source operational data, and time compensation is performed based on the offset relationship between the data generation time and the data reception time to form time-synchronized multi-source operational data; Furthermore, a unified timeline is established based on the operating time of vehicles. The acquisition times of multi-source operating data are matched with the unified timeline to obtain a unified time stamp for each data item. The time offset is calculated based on the data generation time and data reception time of each data item, and the unified time stamp is shifted forward or backward to compensate for the time offset, resulting in a compensated time stamp. The multi-source operating data is rearranged and aligned according to the compensated time stamp, so that the vehicle control data, positioning data, load record data, scheduling task data, road status data, and traffic meteorological data at the same operating time form a corresponding relationship, forming time-synchronized multi-source operating data.

[0028] It should be noted that the offset between the data generation time and the data reception time refers to the time difference between the actual collection and generation of the same data and its reception and recording. For example, the location data is collected after the vehicle location is acquired, but due to transmission delay, it enters the same operating time as the vehicle control data.

[0029] Data credibility is determined based on the data integrity and temporal consistency of time-synchronized multi-source operational data, and stable correlated data is screened through credibility gating to form reliable operational data.

[0030] Furthermore, data integrity refers to whether time-synchronized multi-source operational data has sufficient valid fields and continuous records at the same operational moment. This is mainly determined by the proportion of missing fields, the duration of continuous missing data collection, and the valid range of values. For example, if vehicle control data has a speed field but lacks a drive response field, data integrity will be reduced. Timing consistency refers to whether time-synchronized multi-source operational data conforms to the continuous driving process and operational task execution sequence on a unified time axis. This is mainly determined by the magnitude of data changes between adjacent operational moments, the time offset between different data sources, and the order of vehicle state changes. For example, if positioning data shows a location jump but the vehicle control data does not show a corresponding speed change, timing consistency will be reduced.

[0031] Data credibility is generated for each data item by ensuring data integrity and temporal consistency, and the data credibility is used as the degree to open the credibility gating. Data with high credibility and a stable correspondence with other data at the same operating time is retained, while data with low credibility or that disrupts the continuity of vehicle operation is suppressed or replaced to form reliable operating data.

[0032] The system performs state trajectory consistency verification on the reliable operation data, eliminates abnormal disturbances that deviate from the vehicle's dynamic evolution and operational task progress, and generates valid operation data for the operating vehicles.

[0033] Furthermore, based on the time interval between adjacent operating moments in the reliable operating data, the driving speed, vehicle speed change range, and drive response change range corresponding to the positioning distance are calculated. The driving speed corresponding to the positioning distance is compared with the vehicle speed for consistency, and the vehicle speed change range and drive response change range are matched according to the historical normal operating range of the same vehicle model. When the driving speed corresponding to the positioning distance deviates significantly from the vehicle speed, or the vehicle speed changes abruptly but the drive response does not change accordingly, it is determined that the driving relationship does not conform to continuous driving. At the same time, the task node sequence is determined according to the scheduling task data, the vehicle position change is matched with the task node position, and it is determined whether the load status change occurs at the loading or unloading position. If the vehicle position has crossed the task node but the task status has not changed accordingly, or the load status changes abruptly at a non-loading / unloading position, it is determined that the operation task execution sequence does not conform to continuous driving.

[0034] It should be noted that abnormal disturbances in vehicle dynamics evolution refer to data deviations in reliable operational data that do not conform to the continuous driving pattern of the vehicle, such as inconsistencies between the positioning speed and the onboard speed, or sudden changes in onboard speed without a corresponding change in the drive response. Abnormal disturbances in the progress of operational tasks refer to data deviations in reliable operational data that do not conform to the execution sequence of operational tasks, such as the vehicle's position crossing task nodes without a change in the task status, or sudden changes in the load status at non-loading / unloading positions.

[0035] S2. The effective operating data of the operating vehicles is first segmented according to the changes in operating tasks, and then further segmented according to the changes in road energy consumption status to form a spatiotemporal segment index table of operating vehicles.

[0036] Based on the changes in operational tasks in the effective operational data of operating vehicles, the location of task changes is identified, and the effective operational data of operating vehicles is segmented for the first time to form operational task segments.

[0037] Furthermore, when the scheduling task data changes from not executed to executed and the location data is within the departure point range, the corresponding time marker is identified as the task start position, corresponding to the task initiation stage; when the load recording data continuously rises within the loading point range and reaches a stable state, the corresponding time marker is identified as the loading change position, corresponding to the heavy-load transportation stage after loading is completed; when the location data enters the station or unloading point range and the vehicle speed is continuously lower than the stopping threshold, the corresponding time marker is identified as the station stopping position, corresponding to the stopping and waiting stage; when the load recording data continuously decreases within the unloading point range and reaches a stable state, the corresponding time marker is identified as the unloading change position, corresponding to the empty return stage after unloading; when the location data continuously deviates from the original operating route and matches a new route identifier, the corresponding time marker is identified as the route switching position, corresponding to the route switching stage; the effective operation data of the operating vehicle is first segmented according to the time markers corresponding to the above task change positions to form operating task segments.

[0038] It should be noted that changes in operational tasks refer to changes in the start and end of tasks, loading, unloading, station stops, route switching, or task status transitions of operating vehicles during the current operational task. The stopping threshold is set based on the type of operating vehicle, positioning speed measurement error, and historical station stopping speed distribution, through historical stopping sample quantile statistics and manual calibration verification. An exemplary value range is 0 to 5 km / h.

[0039] Based on the changes in road energy consumption status in the operational task segments, the energy consumption boundary positions corresponding to the energy consumption abrupt change characteristics are identified, and the operational task segments are further segmented according to the energy consumption boundary positions to form road energy consumption segments.

[0040] Furthermore, a road energy consumption state sequence is constructed based on continuous operation records in the operational task segments. The unit distance energy consumption and drive response changes at each operation moment are normalized on the same scale, and coupled and mapped according to the degree of deviation between the two from the sliding reference interval, converting the unit distance energy consumption and drive response changes at each operation moment into the corresponding road energy consumption state value. The sliding reference interval is formed by adjacent stable operation moments, and the normal fluctuation range of the road energy consumption state value within the sliding reference interval is calculated (such as the fluctuation range formed by the upper and lower standard deviations of the sliding reference mean). When the road energy consumption state value at a certain operation moment continuously deviates from the normal fluctuation range, and the deviation trend is maintained in subsequent operation moments, the time marker and road position corresponding to the beginning of the deviation are identified as the energy consumption boundary position.

[0041] Based on the time stamp and road location of the energy consumption boundary position in the operation task segment, the continuous operation records between two adjacent energy consumption boundary positions are divided into a sub-segment. The continuous operation records from the start of the operation task segment to the first energy consumption boundary position are divided into the first sub-segment, and the continuous operation records from the last energy consumption boundary position to the end of the operation task segment are divided into the last sub-segment. Sub-segments with road energy consumption status values ​​less than the merging threshold in adjacent sub-segments are merged to form a road energy consumption segment.

[0042] It should be noted that changes in road energy consumption status refer to the changes in the relationship between energy consumption per unit distance and drive response caused by changes in road conditions or driving status within an operational task segment. Abrupt energy consumption changes refer to a sudden increase or decrease in energy consumption per unit distance, accompanied by a simultaneous abnormal increase or decrease in drive response, even when vehicle speed and load conditions have not changed significantly. For example, the drive response may remain relatively constant while driving continuously on the same road segment, but energy consumption per unit distance may continuously increase, or the energy consumption per unit distance may remain relatively constant while the drive response suddenly increases.

[0043] Based on the vehicle operating status and external driving status within the road energy consumption segment, the road energy consumption segment is marked with its operating status to form a road energy consumption segment with operating status marking. The segment is then indexed and organized in chronological order to form a spatiotemporal segment index table for operating vehicles.

[0044] Furthermore, when marking the operating status of road energy consumption segments, a stable driving state is marked when the vehicle speed is stable and the drive response is smooth; an acceleration driving state is marked when the vehicle speed increases and the drive response strengthens; a deceleration braking state is marked when the vehicle speed decreases or braking occurs; and a low-speed stopping state is marked when the vehicle speed is below the stopping threshold and the position change is small. When the road energy consumption status value is within the fluctuation range of the sliding reference, it is marked as a normal energy consumption state; when it is continuously higher than the fluctuation range of the sliding reference, it is marked as a high-resistance energy consumption state; and when it is continuously lower than the fluctuation range of the sliding reference, it is marked as a low-resistance energy consumption state. This forms road energy consumption segments with operating status markings, which are then organized in chronological order to form a spatiotemporal segment index table for operating vehicles.

[0045] It should be noted that the merging threshold is a difference limit used to determine whether adjacent sub-segments belong to the same road energy consumption state. It is set based on the fluctuation distribution of historical road energy consumption state values ​​for the same operating route, and an exemplary value can be 10% to 30% of the width of the sliding reference fluctuation range. Vehicle operating state refers to the driving state within a road energy consumption segment, characterized by vehicle speed, drive response, and braking state, such as stable driving state, acceleration driving state, deceleration braking state, and low-speed stopping state. External driving state refers to the external constraint state within a road energy consumption segment, characterized by road energy consumption state values ​​and changes in the driving environment, such as normal energy consumption state, high-resistance energy consumption state, low-resistance energy consumption state, and disturbance energy consumption state.

[0046] S3. Based on the spatiotemporal segment index table of operating vehicles, perform slope-driven coupling analysis on each road energy consumption segment to obtain load inversion rules, and perform inversion correction on explicit load states to generate road segment-level energy consumption influence state vectors.

[0047] Based on the spatiotemporal segment index table of operating vehicles, the time range and operating status marker of each road energy consumption segment are located, and the slope change, drive response and energy consumption per unit distance are extracted from the effective operating data of operating vehicles to form segment energy consumption coupling relationship data.

[0048] Furthermore, based on the spatiotemporal segment index table of operating vehicles, the segment number, start time marker, end time marker, and operating status marker of each road energy consumption segment are read. The corresponding continuous records in the valid operating data of the operating vehicles are extracted using the start time marker and end time marker to obtain the segment operation record corresponding to the road energy consumption segment. Within the segment operation record, the slope change is calculated based on the elevation difference or road slope marker corresponding to the adjacent road positions. The drive response is determined based on the response amplitude of the drive torque or drive power relative to the vehicle speed change. The unit distance energy consumption relationship is determined based on the ratio of segment energy consumption to segment travel distance. The slope change, drive response, and unit distance energy consumption relationship are then organized in the same time sequence to form segment energy consumption coupling relationship data.

[0049] It should be noted that gradient change refers to the change in elevation difference or road gradient markings between adjacent road locations within a road energy consumption segment as the travel distance increases. For example, the road gradient increases as a vehicle enters an uphill section from a flat road. Drive response refers to the change in drive torque or drive power generated by the vehicle within a road energy consumption segment to adapt to changes in speed and road resistance. For example, the drive power increases as the gradient increases after the vehicle enters an uphill section. Energy consumption per unit distance refers to the correspondence between energy consumption and travel distance within a road energy consumption segment. For example, the amount of electricity or fuel consumed by a vehicle within a certain kilometer of road energy consumption segment.

[0050] Based on the analysis of the energy consumption coupling relationship data of the segments, the combined impact of slope change on driving response and energy consumption per unit distance is analyzed, and the combined impact is mapped to the explicit load state to establish load inversion rules.

[0051] Furthermore, based on the segment energy consumption coupling relationship data, a comparison window is set before and after the slope change at the starting point within the same road energy consumption segment. The average slope difference, average drive response difference, and energy consumption difference per unit distance are calculated for the before and after comparison windows, and the average drive response difference and energy consumption difference per unit distance are compared with the normal fluctuation range of similar road energy consumption segments. When the average slope difference is positive and both the average drive response difference and energy consumption difference per unit distance exceed the corresponding normal fluctuation range, it is determined that the increase in slope has an enhanced effect on the drive response and energy consumption per unit distance. When the average slope difference is negative and both the average drive response difference and energy consumption difference per unit distance are lower than the corresponding normal fluctuation range, it is determined that the decrease in slope has a weakening effect on the drive response and energy consumption per unit distance. The enhanced or weakened effect is then matched with the expected drive response range and energy consumption range per unit distance under the explicit load state. If the enhanced effect is higher than the corresponding range of the explicit load state, the explicit load state is determined to be too low. If the weakened effect is lower than the corresponding range of the explicit load state, the explicit load state is determined to be too high. The determination relationship of the explicit load state being too low or too high is organized into a load inversion rule. The load inversion rule can reverse the influence of the real load by using the coupling relationship between slope change, drive response and energy consumption per unit distance when the explicit load state is inaccurate or lagging, so that the road segment-level energy consumption influence state vector is closer to the actual energy consumption formation process of operating vehicles.

[0052] It should be noted that explicit load status refers to the load status of an operating vehicle directly determined by the on-board weighing device, load record data, or loading and unloading task record, such as empty, half-loaded, fully loaded, or the corresponding load weight.

[0053] The explicit load state is inverted and corrected using load inversion rules. Load offsets that do not match the segment energy consumption coupling data in the explicit load state are identified, and load correction state data is generated.

[0054] Furthermore, the reference drive response range and reference unit distance energy consumption range corresponding to the explicit load state are read through load inversion rules. The actual drive response amplitude and actual unit distance energy consumption amplitude in the segment energy consumption coupling relationship data are compared with the reference drive response range and reference unit distance energy consumption range, respectively. When the actual drive response amplitude and actual unit distance energy consumption amplitude are both higher than the corresponding range of the explicit load state, it is determined that the explicit load state has a low load offset. When the actual drive response amplitude and actual unit distance energy consumption amplitude are both lower than the corresponding range of the explicit load state, it is determined that the explicit load state has a high load offset. The explicit load state is then adjusted or compensated according to the low load offset or high load offset to generate load correction state data.

[0055] The load correction status data is embedded into the operating status marker of the road energy consumption segment, and the load impact and road segment energy consumption sensitivity are correlated and encoded to form a road segment-level energy consumption impact status vector.

[0056] Furthermore, the load correction status data is bound to the operating status markers of road energy consumption segments according to the same segment number. Based on the load level difference or load value difference between the load correction status data and the explicit load status, the load offset degree of the load correction status data relative to the explicit load status is analyzed. The change in energy consumption per unit distance of the road energy consumption segment under the corresponding operating status marker is statistically analyzed. The load offset degree and the change in energy consumption per unit distance are encoded accordingly, so that each road energy consumption segment forms a segment-level energy consumption influence state vector that includes the load correction effect and the segment energy consumption sensitivity.

[0057] It should be noted that load impact refers to the effect of the deviation of load correction state data relative to explicit load state on the drive response and energy consumption per unit distance in road energy consumption segments. Segment energy consumption sensitivity refers to the degree to which the energy consumption per unit distance of a road energy consumption segment fluctuates with changes in load correction state data under the corresponding operating state marker.

[0058] S4. Perform residual cause analysis on the energy consumption impact state vector at the road segment level, construct an energy consumption residual cause-effect graph, and only migrate similar operating segments with consistent residual causes to obtain the initial dual-scale energy consumption prediction of operating vehicles. By comparing the deviation between the road segment-level energy consumption influence state vector and the historical energy consumption benchmark of similar road segments, the energy consumption residual data of the current road segment is determined.

[0059] Furthermore, based on the operating state markers and load correction states in the road segment-level energy consumption influence state vector, historical samples belonging to the same operating state as the current road energy consumption segment are selected from historical road energy consumption segments, and the average or quantile stable value of energy consumption per unit distance of the same historical samples is used as the historical energy consumption benchmark. Then, the energy consumption per unit distance corresponding to the current road energy consumption segment is converted into the energy consumption characterization value of the current segment. The difference between the energy consumption characterization value of the current segment and the historical energy consumption benchmark is compared, and the degree of energy consumption that the current road energy consumption segment is higher or lower than that of the same historical operating state is determined according to the direction and magnitude of the difference, thus forming the energy consumption residual data of the current road energy consumption segment.

[0060] It should be noted that similar road energy consumption segments refer to historical road energy consumption segments that have the same or similar operating status markings, load correction statuses, and road energy consumption statuses as the current road energy consumption segment. Historical energy consumption benchmarks refer to the stable reference values ​​of energy consumption per unit distance formed by similar road energy consumption segments in historical operating records, which can be determined by the historical average, median, or stable quantile values ​​of energy consumption per unit distance for similar road energy consumption segments.

[0061] Counterfactual residual attribution and time-series lag verification are used to conduct residual cause analysis on energy consumption residual data, establish causal relationships between energy consumption residual data and energy consumption disturbance factors, and construct an energy consumption residual causal graph.

[0062] Furthermore, counterfactual residual attribution refers to, after the energy consumption residual data has been determined, taking the actual operating state of the current road energy consumption segment as the factual state, replacing a certain energy consumption disturbance factor with the historical stable state of the same type of road energy consumption segment, while keeping other energy consumption disturbance factors unchanged, and determining the contribution intensity of the corresponding energy consumption disturbance factor by comparing the change range of the energy consumption residual data before and after the replacement; time lag verification refers to verifying the time of change of the energy consumption disturbance factor and the time of appearance of the energy consumption residual data. Only when the energy consumption disturbance factor changes first, and the energy consumption residual data changes accordingly within a reasonable lag time, is the causal relationship between the energy consumption disturbance factor and the energy consumption residual data retained.

[0063] When performing residual cause analysis on energy consumption residual data, the contribution intensity of each energy consumption disturbance factor to the energy consumption residual data is first calculated by counterfactual residual attribution. Then, interference factors with high contribution intensity but no temporal relationship are eliminated by time series lag verification. Energy consumption disturbance factors with contribution intensity that meets the requirements and time series relationship are determined as effective residual causes. When constructing the energy consumption residual cause-effect graph, the effective residual causes and energy consumption residual data are used as causal nodes. The effective residual causes are directed to the energy consumption residual data to form causal edges. The weights of the causal edges are determined according to the contribution intensity and lag stability to form the energy consumption residual cause-effect graph.

[0064] It should be noted that energy consumption disturbance factors refer to operational factors that can cause changes in energy consumption residual data, including load correction deviations, changes in road energy consumption status, driver energy consumption deviations, abnormal vehicle power response, and changes in the external driving environment. The energy consumption residual cause-effect graph can identify the true causes from the energy consumption residual data and only allows similar operational segments with consistent residual causes to participate in migration prediction, thereby avoiding erroneous migration of seemingly similar data with different sources of energy consumption deviation, and improving the accuracy of dual-scale energy consumption prediction for operating vehicles.

[0065] Based on the residual cause-effect graph of energy consumption, the residual causes of historical operation segments are matched, and similar operation segments with the same residual causes as the current road energy consumption segment are selected to form a set of similar operation segments.

[0066] Furthermore, the effective residual causes and causal edge weights corresponding to the current road energy consumption segment are used as matching benchmarks. The historical effective residual causes and historical causal edge weights corresponding to each historical operation segment are read from the historical operation segments, and the effective residual causes are compared one by one to see if they belong to the same type and whether the difference in causal edge weights is within a preset similarity range. When the historical operation segment and the current road energy consumption segment have the same dominant effective residual causes and the difference in causal edge weights is less than the preset similarity threshold, the historical operation segment is determined to be a similar operation segment with the same residual causes as the current road energy consumption segment, and the similar operation segments that meet the conditions are summarized to form a set of similar operation segments.

[0067] It should be noted that the similarity range can be set based on the fluctuation distribution of historical causal edge weights in similar road energy consumption segments, for example, taking the range formed by one standard deviation above and below the mean of historical causal edge weights. The similarity threshold can be set based on the residual cause matching error between the current road energy consumption segment and historical operation segments, for example, taking the 75th quantile of the historical matching error of similar road energy consumption segments as the threshold.

[0068] Based on the energy consumption residual cause-effect diagram, the weight of the residual cause consistency is determined, and the segment energy consumption increment is embedded into the road segment-level energy consumption influence state vector according to the load sensitivity correction coefficient to calculate the initial short-term energy consumption prediction value.

[0069] Furthermore, the baseline energy consumption of the current road energy consumption segment is first determined based on the segment-level energy consumption influence state vector of the current road energy consumption segment. Then, the residual cause consistency weight between each similar operating segment and the current road energy consumption segment is determined based on the energy consumption residual cause graph. The segment energy consumption increments of similar operating segments are weighted and summarized according to the residual cause consistency weight. At the same time, the load influence amplification or weakening processing is applied to the weighted and summarized segment energy consumption increment based on the load sensitivity correction coefficient and load correction deviation of the current road energy consumption segment. Finally, the processed segment energy consumption increment is superimposed on the baseline energy consumption of the current road energy consumption segment to obtain the initial short-term energy consumption prediction value.

[0070] It should be noted that the load sensitivity correction coefficient can be determined based on the corresponding slope between the change in load correction deviation and the change in energy consumption per unit distance in similar road energy consumption segments. This coefficient characterizes the impact of load correction deviation on the energy consumption increment of the current road energy consumption segment. The segment energy consumption increment refers to the increase or decrease in actual energy consumption of a similar operating segment relative to the historical energy consumption benchmark of similar road energy consumption segments. It represents the transferable energy consumption deviation predicted by similar operating segments for the current road energy consumption segment.

[0071] The initial short-time energy consumption prediction value is calculated using the following expression: in, This represents the initial short-term energy consumption forecast. Indicates the first The baseline energy consumption of a road energy consumption segment. Indicates the first The segment-level energy consumption influence state vector of a road energy consumption segment Indicates the first Load-sensitive correction factor for each road energy consumption segment Indicates the first Load correction deviation of each road energy consumption segment Indicates the first A set of similar running segments with consistent residual causes corresponding to each road energy consumption segment. Indicates the first The first similar running segment and the first Consistent weighting of residual causes among different road energy consumption segments Indicates the first The energy consumption increment of a segment with similar operation segments. This indicates the sequence number of the current road energy consumption segment in the spatiotemporal segment index table of operating vehicles.

[0072] The remaining road energy consumption segments are read from the spatiotemporal segment index table of operating vehicles, and the energy consumption increments of similar segments of the remaining road energy consumption segments are recursively accumulated according to the weight of the consistency of residual causes to calculate the initial predicted value of the remaining operating task energy consumption.

[0073] Furthermore, starting from the current road energy consumption segment, the remaining road energy consumption segments are read sequentially from the spatiotemporal segment index table of operating vehicles. The corresponding baseline energy consumption is determined based on the road segment-level energy consumption influence state vector of each remaining road energy consumption segment. The segment energy consumption increment is determined based on the set of similar operating segments corresponding to each remaining road energy consumption segment. The segment energy consumption increment is then filtered and weighted using the residual cause consistency weight. At the same time, the weighted segment energy consumption increment is adjusted by combining the corresponding load sensitivity correction coefficient and load correction deviation to obtain the segment predicted energy consumption of each remaining road energy consumption segment. Finally, the remaining road energy consumption segments are recursively accumulated according to their order in the spatiotemporal segment index table of operating vehicles to calculate the initial predicted value of remaining operating task energy consumption.

[0074] It should be noted that the energy consumption increment of similar segments refers to the energy consumption offset formed by the actual energy consumption of similar operating segments relative to the historical energy consumption benchmark of similar road energy consumption segments. It is used to migrate and characterize the possible increase or decrease in energy consumption of the remaining road energy consumption segments.

[0075] The initial predicted energy consumption for remaining operational tasks is calculated using the following expression: ; in, This represents the initial predicted energy consumption for remaining operational tasks. This indicates the sequence number of the remaining road energy consumption segment in the spatiotemporal segment index table for operating vehicles. This indicates the sequence number of the last road energy consumption segment for the remaining operational tasks. Indicates the first The segment-level energy consumption influence state vector of each remaining road energy consumption segment Indicates the first The baseline energy consumption of each remaining road energy consumption segment. Indicates the first Load-sensitive correction factor for each remaining road energy consumption segment Indicates the first Load correction deviation of each remaining road energy consumption segment Indicates the first A set of similar running segments corresponding to each remaining road energy consumption segment. Indicates the first The first similar running segment and the first The residual causes of each remaining road energy consumption segment are weighted consistently. Indicates the first The energy consumption increment of a segment with similar operation.

[0076] The initial dual-scale energy consumption forecast for operating vehicles consists of the initial short-term energy consumption forecast and the initial remaining operating task energy consumption forecast.

[0077] S5. Based on the vehicle's historical residual drift, driver energy consumption offset, and load correction deviation, perform individualized residual compensation on the initial dual-scale energy consumption prediction of operating vehicles to generate the corrected dual-scale energy consumption prediction of operating vehicles.

[0078] Based on the changes in the initial dual-scale energy consumption prediction of operating vehicles and the historical prediction deviation of current vehicles in the same type of road energy consumption segment, the vehicle residual drift compensation amount corresponding to the historical residual drift of the vehicle is determined.

[0079] Furthermore, the current road energy consumption segment corresponding to the initial dual-scale energy consumption prediction of the operating vehicle is matched with the historical prediction record of the current vehicle in the same type of road energy consumption segment. The historical prediction deviation between the historical predicted energy consumption and the historical actual energy consumption is read, and it is judged in chronological order whether the historical prediction deviation has a continuous upward or downward drift trend. When the historical prediction deviation remains in the same direction in multiple consecutive road energy consumption segments of the same type, the vehicle's historical residual drift is determined according to the average offset of the historical prediction deviation, and the vehicle's historical residual drift is converted into a vehicle residual drift compensation amount to compensate for the prediction deviation caused by the current vehicle's long-term high or low energy consumption.

[0080] It should be noted that historical prediction deviation refers to the continuous upward, continuous downward, or fluctuating convergence trend of the difference between the historical predicted energy consumption and the historical actual energy consumption of the current vehicle in the same type of road energy consumption segment over time. Vehicle historical residual drift refers to the individual vehicle energy consumption deviation trend formed by the long-term consistent change in the deviation between the historical predicted energy consumption and the historical actual energy consumption of the current vehicle in the same type of road energy consumption segment.

[0081] The vehicle residual drift compensation amount is superimposed and verified with the driver energy consumption offset formed by the current driver under the same operating conditions to obtain the driver-related residual compensation amount.

[0082] Furthermore, the vehicle residual drift compensation amount is correlated with the current driver's historical driving records under similar operating conditions. The difference between the current driver's historical actual energy consumption and the benchmark energy consumption of similar vehicles is calculated, and the difference is used as the driver's energy consumption offset. When the driver's energy consumption offset and the vehicle residual drift compensation amount are in the same direction, it indicates that the driver's operation difference and the vehicle's individual difference jointly cause the energy consumption to be too high or too low. The two are superimposed to form the driver-related residual compensation amount. When the driver's energy consumption offset and the vehicle residual drift compensation amount are in opposite directions, it indicates that the driver's operation difference and the vehicle's individual difference have a canceling effect. The difference between the two is used to form the driver-related residual compensation amount.

[0083] It should be noted that the driver-related residual compensation amount can perform directional verification and unified compensation for the energy consumption deviation of individual vehicles and the energy consumption deviation caused by the driver's operating habits, so that the initial dual-scale energy consumption prediction amount of operating vehicles is more in line with the actual energy consumption performance of the current vehicle and current driver combination.

[0084] By matching the driver-related residual compensation amount with the load correction deviation, the effect of vehicle load changes on the increase or decrease of the driver-related residual compensation amount is determined, and an individualized residual compensation amount is obtained.

[0085] Furthermore, when matching the driver-related residual compensation amount with the load correction deviation, it is first determined whether the load correction deviation corresponds to an explicit load state that is too low or too high, and then it is determined whether the driver-related residual compensation amount corresponds to a compensation for high energy consumption or a compensation for low energy consumption. When the load correction deviation indicates an explicit load state that is too low and the driver-related residual compensation amount is a compensation for high energy consumption, it indicates that the load factor will amplify the driver-related residual compensation amount. When the load correction deviation indicates an explicit load state that is too high and the driver-related residual compensation amount is a compensation for low energy consumption, it indicates that the load factor will enhance the low energy consumption compensation trend. When the two are opposite, it indicates that the load factor has a weakening effect on the driver-related residual compensation amount. The driver-related residual compensation amount is amplified or weakened according to the load sensitivity correction degree corresponding to the load correction deviation to obtain an individualized residual compensation amount.

[0086] It should be noted that load correction deviation refers to the difference between the explicit load state and the load state obtained after correction by the load inversion rule, used to characterize the degree of deviation between the load record and the actual load effect. Vehicle load change refers to the change in load state caused by the increase or decrease in the weight carried by the operating vehicle during loading, unloading, or task execution. Individualized residual compensation can unify the vehicle historical residual drift, driver energy consumption offset, and load correction deviation into targeted compensation for the initial two-scale energy consumption prediction of the operating vehicle, making the prediction more consistent with the actual operating energy consumption under the current vehicle, driver, and load state.

[0087] The individualized residual compensation amount is used to synchronously compensate the initial two-scale energy consumption prediction of operating vehicles, thereby generating the corrected two-scale energy consumption prediction of operating vehicles.

[0088] Furthermore, the individualized residual compensation is allocated according to the current road energy consumption segment corresponding to the short-term prediction and the remaining road energy consumption segment corresponding to the remaining task prediction. For the initial short-term energy consumption prediction value in the initial dual-scale energy consumption prediction of operating vehicles, the individualized residual compensation corresponding to the current road energy consumption segment is directly superimposed. For the initial remaining operating task energy consumption prediction value in the initial dual-scale energy consumption prediction of operating vehicles, the individualized residual compensation is recursively allocated according to the segment order and segment energy consumption ratio of the remaining road energy consumption segments, and the allocated compensation is superimposed on the corresponding segment prediction energy consumption, and the corrected dual-scale energy consumption prediction of operating vehicles is generated.

[0089] This embodiment also provides an energy consumption prediction system for operating vehicles based on multi-source data fusion, including: a data acquisition module, which collects multi-source operating data of operating vehicles, performs time synchronization, reliable fusion and anomaly correction on the multi-source operating data, and generates effective operating data of operating vehicles; The operation time-space segmentation module divides the effective operation data of operating vehicles into the first segment according to the changes in operation tasks, and the second segment according to the changes in road energy consumption status, forming an index table of operating vehicle time-space segments. The load inversion correction module, based on the spatiotemporal segment index table of operating vehicles, performs slope-driven coupling analysis on each road energy consumption segment to obtain load inversion rules, and performs inversion correction on explicit load states to generate a road segment-level energy consumption influence state vector. The residual causal migration prediction module performs residual cause analysis on the energy consumption influence state vector at the road segment level, constructs an energy consumption residual causal graph, and migrates only similar running segments with consistent residual causes to obtain the initial dual-scale energy consumption prediction of operating vehicles. The individualized residual compensation module performs individualized residual compensation on the initial dual-scale energy consumption prediction of operating vehicles based on vehicle historical residual drift, driver energy consumption offset, and load correction deviation, generating the corrected dual-scale energy consumption prediction of operating vehicles.

[0090] In summary, this invention achieves accurate mapping of load influence to road segment-level energy consumption influence state vectors by: performing slope-driven coupling analysis on road energy consumption segments using an index table of spatiotemporal segments of operating vehicles; establishing load inversion rules and correcting explicit load states; and filtering similar operating segments with consistent residual causes through an energy consumption residual causal graph to achieve causal constraint transfer of the initial dual-scale energy consumption prediction of operating vehicles. Furthermore, it generates corrected dual-scale energy consumption predictions of operating vehicles by combining individualized residual compensation. These predictions can be used for short-term energy consumption and remaining operating task energy consumption prediction of operating vehicles, improving prediction relevance and operational adaptability.

[0091] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for predicting the energy consumption of operating vehicles based on multi-source data fusion, characterized in that, include: Collect multi-source operation data of operating vehicles, perform time synchronization, reliable fusion and anomaly correction on the multi-source operation data, and generate valid operation data of operating vehicles; The effective operating data of operating vehicles is first segmented according to changes in operating tasks, and then further segmented according to changes in road energy consumption status to form a spatiotemporal segment index table of operating vehicles. Based on the spatiotemporal segment index table of operating vehicles, slope-driven coupling analysis is performed on each road energy consumption segment to obtain load inversion rules, and explicit load states are inverted and corrected to generate road segment-level energy consumption influence state vectors. The steps for obtaining load inversion rules by performing slope-driven coupling analysis on each road energy consumption segment are as follows: Based on the spatiotemporal segment index table of operating vehicles, the time range and operation status marker of each road energy consumption segment are located, and the slope change, drive response and energy consumption per unit distance are extracted from the effective operation data of operating vehicles to form segment energy consumption coupling relationship data. Based on the analysis of the energy consumption coupling relationship data of the segments, the combined impact of slope change on driving response and energy consumption per unit distance is analyzed, and the combined impact is mapped to the explicit load state to establish load inversion rules; The residual cause analysis is performed on the energy consumption impact state vector at the road segment level, an energy consumption residual cause-effect graph is constructed, and only similar operating segments with consistent residual causes are migrated to obtain the initial dual-scale energy consumption prediction of operating vehicles. The steps are as follows: Based on the deviation comparison between the road segment-level energy consumption influence state vector and the historical energy consumption benchmark of similar road energy consumption segments, the energy consumption residual data of the current road energy consumption segment is determined. Counterfactual residual attribution and time-series lag verification are used to conduct residual cause analysis on energy consumption residual data, establish causal relationships between energy consumption residual data and energy consumption disturbance factors, and construct an energy consumption residual causal graph. Based on the energy consumption residual cause-effect graph, the residual causes of historical operation segments are matched, and similar operation segments with the same residual causes as the current road energy consumption segment are selected to form a set of similar operation segments. Based on the energy consumption residual cause-effect diagram, the weight of the residual cause consistency is determined, and the segment energy consumption increment is embedded into the road segment-level energy consumption influence state vector according to the load sensitivity correction coefficient to calculate the initial short-term energy consumption prediction value. The remaining road energy consumption segments are read from the spatiotemporal segment index table of operating vehicles, and the energy consumption increments of similar segments of the remaining road energy consumption segments are recursively accumulated according to the weight of consistent residual causes to calculate the initial energy consumption prediction value of the remaining operating tasks. Based on the vehicle's historical residual drift, driver energy consumption offset and load correction deviation, individualized residual compensation is performed on the initial dual-scale energy consumption prediction of operating vehicles to generate the corrected dual-scale energy consumption prediction of operating vehicles.

2. The method for predicting the energy consumption of operating vehicles based on multi-source data fusion as described in claim 1, characterized in that: The multi-source operational data includes vehicle control data, positioning data, load recording data, scheduling task data, road condition data, and traffic and meteorological data.

3. The method for predicting the energy consumption of operating vehicles based on multi-source data fusion as described in claim 2, characterized in that, The steps for performing time synchronization, reliable fusion, and anomaly correction on multi-source operational data to generate valid operational data for commercial vehicles are as follows: A unified timeline mapping is performed on various acquisition times in the multi-source operational data, and time compensation is performed based on the offset relationship between the data generation time and the data reception time to form time-synchronized multi-source operational data; Data credibility is determined based on the data integrity and temporal consistency of time-synchronized multi-source operational data, and stable correlated data is screened through credibility gating to form reliable operational data; The system performs state trajectory consistency verification on the reliable operation data, eliminates abnormal disturbances that deviate from the vehicle's dynamic evolution and operational task progress, and generates valid operation data for the operating vehicles.

4. The method for predicting the energy consumption of operating vehicles based on multi-source data fusion as described in claim 1, characterized in that, The steps for first segmenting the effective operating data of commercial vehicles according to changes in operating tasks, and second segmenting it according to changes in road energy consumption status to form a spatiotemporal segment index table for commercial vehicles are as follows: Based on the changes in operational tasks in the effective operational data of operating vehicles, the location of task changes is identified, and the effective operational data of operating vehicles is segmented for the first time to form operational task segments. Based on the changes in road energy consumption status in the operation task segments, identify the energy consumption boundary positions corresponding to the energy consumption abrupt change characteristics, and then perform secondary segmentation of the operation task segments according to the energy consumption boundary positions to form road energy consumption segments. Based on the vehicle operating status and external driving status within the road energy consumption segment, the road energy consumption segment is marked with its operating status to form a road energy consumption segment, and then indexed and organized in chronological order to form a spatiotemporal segment index table for operating vehicles.

5. The method for predicting the energy consumption of operating vehicles based on multi-source data fusion as described in claim 1, characterized in that, The explicit load state is inverted and corrected to generate a road segment-level energy consumption influence state vector. The steps are as follows: The explicit load state is inverted and corrected using load inversion rules. Load offsets that do not match the segment energy consumption coupling data in the explicit load state are identified, and load correction state data is generated. The load correction status data is embedded into the operating status marker of the road energy consumption segment, and the load impact and road segment energy consumption sensitivity are correlated and encoded to form a road segment-level energy consumption impact status vector.

6. The method for predicting the energy consumption of operating vehicles based on multi-source data fusion as described in claim 1, characterized in that, The steps for performing individualized residual compensation on the initial dual-scale energy consumption prediction of operating vehicles to generate the corrected dual-scale energy consumption prediction of operating vehicles are as follows: Based on the changes in the initial dual-scale energy consumption prediction of operating vehicles and the historical prediction deviation of current vehicles in the same type of road energy consumption segment, determine the vehicle residual drift compensation amount corresponding to the historical residual drift of the vehicle. The vehicle residual drift compensation amount is superimposed and verified with the driver energy consumption offset formed by the current driver under the same operating conditions to obtain the driver-related residual compensation amount. By matching the driver-related residual compensation amount with the load correction deviation, the effect of vehicle load changes on the increase or decrease of the driver-related residual compensation amount is determined, and individualized residual compensation amount is obtained. The individualized residual compensation amount is used to synchronously compensate the initial two-scale energy consumption prediction of operating vehicles, thereby generating the corrected two-scale energy consumption prediction of operating vehicles.

7. A system for predicting the energy consumption of operating vehicles based on multi-source data fusion, based on the method for predicting the energy consumption of operating vehicles based on multi-source data fusion as described in any one of claims 1 to 6, characterized in that, include: The data acquisition module collects multi-source operating data from operating vehicles, performs time synchronization, reliable fusion, and anomaly correction on the multi-source operating data, and generates valid operating data for operating vehicles. The operation time-space segmentation module divides the effective operation data of operating vehicles into the first segment according to the changes in operation tasks, and the second segment according to the changes in road energy consumption status, forming an index table of operating vehicle time-space segments. The load inversion correction module, based on the spatiotemporal segment index table of operating vehicles, performs slope-driven coupling analysis on each road energy consumption segment to obtain load inversion rules, and performs inversion correction on explicit load states to generate a road segment-level energy consumption influence state vector. The residual causal migration prediction module performs residual cause analysis on the road segment-level energy consumption influence state vector, constructs an energy consumption residual causal graph, and migrates only similar running segments with consistent residual causes to obtain the initial dual-scale energy consumption prediction of operating vehicles. The individualized residual compensation module performs individualized residual compensation on the initial dual-scale energy consumption prediction of operating vehicles based on vehicle historical residual drift, driver energy consumption offset, and load correction deviation, generating the corrected dual-scale energy consumption prediction of operating vehicles.

Images