A vehicle-mounted data analysis method and system

Abstract

The application discloses a kind of vehicle-mounted data analysis method and system, it is related to vehicle-mounted data analysis field, for solving the nonlinear misplacement of multiple-source data timestamp, it include: identifying multiple-source event;Multiple-source event includes internal event and external V2X event, and records the original timestamp and data source of multiple-source event;According to the preset event causal relationship rule set, the causal relationship between multiple-source event is inferred, and event causal chain is obtained;The confidence of each original timestamp of multiple-source event is evaluated, and the confidence of each original timestamp is obtained;Confidence evaluation is based on data source, signal quality index, the matching degree of environmental context and event causal chain where vehicle is located;Based on event causal chain and confidence, the timestamp of multiple-source event is dynamically corrected.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of vehicle data analysis, and in particular to a vehicle data analysis method and system. BACKGROUND

[0002] In the daily operation scenario of new energy vehicle fleet, the vehicle data analysis system needs to integrate vehicle internal sensor data and external V2X communication data to realize intelligent management of the vehicle fleet. In order to ensure the time consistency of multi-source data, the system presets a "time stamp alignment rule". This rule usually performs time stamp calibration and synchronization on the received data packets based on communication link delay statistics and data generation timing experience model, for example, linearly compensates the time stamp of V2X message to make it consistent with the time reference of vehicle internal data. In the conventional communication environment, this set of rules can effectively ensure the accuracy of data fusion and provide reliable data basis for energy consumption prediction and vehicle fleet scheduling.

[0003] However, in specific environments such as industrial parks where electromagnetic interference is frequent, the vehicle V2X communication antenna is exposed to a complex electromagnetic environment for a long time and continuously suffers from occasional and instantaneous electromagnetic disturbances. For example, when the vehicle passes by large industrial equipment, the electromagnetic radiation generated instantaneously may cause temporary decline in V2X signal quality, random fluctuation in data packet transmission delay, and the fluctuation presents nonlinear characteristics, which is difficult to predict. Influenced by this, part of the external traffic flow information received by the vehicle computing unit has a nonlinear deviation from the time stamp of the high-precision data in the vehicle, which is difficult to completely correct by the existing "time stamp alignment rule". This deviation causes the data from different sources to be slightly "misaligned" in the time dimension, for example, an emergency braking event detected by an internal sensor may not be accurately associated with the external event that caused the braking due to the time stamp lag of the external traffic flow data. The existing "time stamp alignment rule" mainly performs linear compensation based on average delay or simple sliding window matching, and for the delay fluctuation caused by intermittent electromagnetic interference with randomness and nonlinear characteristics, the correction ability is insufficient, and it cannot accurately capture and correct this dynamic and irregular time deviation, thereby causing the accuracy of data fusion to decrease.

[0004] Moreover, under operating pressure, the driver may choose a route with unstable communication signals based on personal experience, or choose to skip retry or delay upload when V2X data upload fails, resulting in partial key V2X data not being uploaded to the cloud platform in a timely and complete manner, further exacerbating the incompleteness of the data, making the system's perception of the environment where the vehicle is located "blind". Due to the above nonlinear deviation of the timestamp and the delay or absence of V2X data upload, the vehicle-mounted data analysis system faces serious challenges when constructing the data set required for energy consumption prediction. There is a significant "information gap" in time and "local incompleteness" in content between the external traffic flow information received by the system and the actual driving behavior and battery state data of the vehicle. This decline in data quality reduces the prediction ability of the energy consumption prediction logic for real-world dynamic traffic environments, and the adaptability and general ability of complex traffic scenarios are greatly compromised, and the real law of energy consumption under different traffic conditions cannot be accurately reflected.

[0005] Finally, the vehicle-mounted data analysis system is seriously challenged when it faces the "multi-source data timestamp nonlinear misalignment and local absence" problem caused by the occasional fluctuations in V2X communication within the industrial park, the specific operating habits of the driver under operating pressure, and the limitations of the current "timestamp alignment rule". The "multi-source data timestamp alignment rule" and "feature weight set in energy consumption prediction logic" within the system cannot work effectively, resulting in a significant reduction in data fusion efficiency, delays in real-time decision-making, and insufficient accuracy. More seriously, the "priority judgment logic of vehicle fleet dispatching instructions" cannot accurately obtain and evaluate the real situation of the single vehicle state and real-time road conditions when formulating a global joint scheduling scheme, thus exacerbating the overall decline in the operating efficiency of the entire new energy vehicle fleet, forming a complex technical dilemma that cannot be solved by a single technical improvement. SUMMARY

[0006] The present application provides a vehicle-mounted data analysis method for solving the problem of nonlinear misalignment of multi-source data timestamps.

[0007] In a first aspect, to solve the above technical problems, the present application provides a vehicle-mounted data analysis method, comprising: identifying multi-source events; the multi-source events include internal events and external V2X events, and recording the original timestamps and data sources of the multi-source events;

[0008] According to the preset event causal relationship rule set, the causal relationship between the multi-source events is inferred to obtain an event causal chain;

[0009] The confidence of each original timestamp of the multi-source events is evaluated to obtain the confidence of each original timestamp; the confidence evaluation is based on the data source, the signal quality indicator, the environmental context in which the vehicle is located, and the matching degree of the event causal chain;

[0010] Based on the event causal chain and the confidence, the timestamps of the multi-source events are dynamically corrected.

[0011] Through the technical solution, the vehicle internal sensor data and external V2X communication data can be effectively integrated, the multi-source events are identified, the causal relationship is inferred, the timestamp confidence is evaluated and dynamically corrected, thereby solving the problems of nonlinear misplacement and local missing of the timestamps of the multi-source data, improving the accuracy of data fusion, and providing a reliable data basis for energy consumption prediction and vehicle fleet scheduling.

[0012] Further, the multi-source events are identified, including:

[0013] Raw data from vehicle internal sensors and external V2X communication are acquired; the raw data includes at least one of the following: vehicle speed, brake pedal depth, steering angle, battery charge and discharge state, front traffic signal light state, road intersection congestion degree, and front obstacle warning;

[0014] According to the raw data, the multi-source events are identified.

[0015] Through the technical solution, rich vehicle internal and external raw data can be acquired to provide comprehensive information input for subsequent event identification and causal inference, thereby improving the accuracy and comprehensiveness of event identification.

[0016] More specifically, in some embodiments, according to the raw data, the multi-source events are identified, including:

[0017] If the vehicle speed decreases by more than 5 m / s within 0.5 s, an emergency braking event is identified;

[0018] If the vehicle speed is 0, a stop event is identified.

[0019] Through the technical solution, based on specific vehicle behavior parameters, key driving events can be accurately identified to provide clear event definitions for subsequent causal relationship analysis, thereby improving the accuracy and automation level of event identification.

[0020] Preferably, according to a preset event causal relationship rule set, the causal relationship between the multi-source events is inferred to obtain an event causal chain, including:

[0021] If the front obstacle warning is earlier in time than the emergency braking event, and the time interval between them is between 0 ms and 500 ms, then there is a causal relationship between them;

[0022] If the red light of the road intersection signal is earlier in time than the stop event, and the time interval between them is between 0 ms and 2000 ms, then there is a causal relationship between them;

[0023] If the received signal strength indication (RSSI) reported by the V2X module is lower than -90 dBm or the bit error rate (BER) is higher than 10 -3 , and the V2X data packet transmission delay or loss is close in time, and the current location of the vehicle is in a predefined high-interference area, then there is a causal relationship between the two.

[0024] Through the technical solution, the preset causal relationship rule set can be used to effectively infer the potential causal relationship between multiple source events in combination with the time sequence and specific conditions, so that an event causal chain with logic is constructed, thereby providing solid logical support for subsequent timestamp correction.

[0025] On the basis described above, the application further proposes that the confidence of each original timestamp of the multiple source events is evaluated, including:

[0026] For each event in the multiple source events, the original confidence of each event original timestamp is obtained;

[0027] When the RSSI is lower than the RSSI threshold, the original confidence is reduced by a first step; when the BER is higher than the BER threshold, the original confidence is reduced by a second step;

[0028] In the case where the event and the high-confidence event form an event causal chain, the original confidence is increased by a third step; the high-confidence event is an event with an original confidence greater than a confidence threshold.

[0029] Through the technical solution, the signal quality, environmental context, and matching degree of the event causal chain can be comprehensively considered for fine confidence evaluation of each original timestamp, so that the reliability of the timestamp can be more accurately reflected, and a more reliable basis can be provided for subsequent dynamic correction.

[0030] As a technical improvement, the timestamps of the multiple source events are dynamically corrected based on the event causal chain and the confidence, including:

[0031] The spatial matching degree score and the time logic score of the event causal chain hypothesis are evaluated, and weighted summation is performed to obtain a support score;

[0032] According to the support score, the optimal event causal chain is determined; the optimal event causal chain is the event causal chain with the highest support score;

[0033] The time information of the external V2X event is calibrated according to the optimal event causal chain.

[0034] By the technical scheme, the spatial matching degree and the time logic of the event causal chain can be evaluated, and the optimal causal chain can be determined, so that dynamic correction of the external V2X event timestamp is realized, the nonlinear time deviation problem is effectively solved, and the accuracy of data fusion is improved.

[0035] To perfect the scheme, the time information of the external V2X event is calibrated according to the optimal event causal chain, including:

[0036] For the internal event of the optimal event causal chain and the external V2X event, the timestamp of the internal event is used as a time reference.

[0037] According to the difference between the timestamp of the internal event and the average reaction time of the driver, the timestamp of the external V2X event is determined.

[0038] By the technical scheme, the timestamp of the external V2X event can be accurately calibrated by using the high-confidence internal event timestamp as a reference and combining experience parameters such as driver reaction time, so that the accuracy and practicability of timestamp correction are further improved.

[0039] In an embodiment, the spatial matching degree score satisfies the following relationship:

[0040] K = max (0, 1-L-Z) ;

[0041] L is the ratio of the distance between the external V2X event position and the vehicle position to the preset maximum relevant distance, and Z is the ratio of the direction angle between the direction in which the vehicle points to the external V2X event position and the driving direction of the vehicle to 180 degrees.

[0042] By the technical scheme, the spatial matching degree between the external V2X event and the vehicle can be evaluated in a quantitative manner, providing an objective basis for the selection of the causal chain, so that the accuracy of causal inference is improved.

[0043] In another embodiment, the time logic score satisfies the following relationship:

[0044] If the external V2X event occurrence time is before the anchor event occurrence time, and the time interval between the two falls within the driver reaction time range, the time logic score is 1, otherwise it is 0.

[0045] By the technical scheme, the time logic relationship between the external V2X event and the anchor event can be evaluated in a quantitative manner, further enhancing the reliability of the causal chain, so that the accuracy of timestamp correction is improved.

[0046] In a second aspect, the application also discloses a vehicle-mounted data analysis system, which comprises:

[0047] An identification module is configured to identify multi-source events, which include internal events and external V2X events, and record original timestamps and data sources of the multi-source events.

[0048] A causal inference module is configured to infer causal relationships between the multi-source events according to a preset event causal relationship rule set, and obtain an event causal chain.

[0049] A confidence evaluation module is configured to evaluate the confidence of each original timestamp of the multi-source events, and obtain the confidence of each original timestamp. The confidence evaluation is based on the data sources, signal quality indicators, environmental context in which the vehicle is located, and matching degree of the event causal chain.

[0050] A correction module is configured to dynamically correct the timestamps of the multi-source events based on the event causal chain and the confidence.

[0051] By the technical solution, a system architecture for realizing a vehicle-mounted data analysis method is provided. The modular design clearly divides the functional units, facilitating the implementation, deployment and maintenance of the system, and effectively solving the problems of non-linear misplacement and local missing of multi-source data timestamps.

[0052] Advantages

[0053] The vehicle-mounted data analysis method disclosed in the present application identifies multi-source events (including internal events and external V2X events), and records original timestamps and data sources thereof, thereby laying a foundation for subsequent analysis. On this basis, the causal relationships between the multi-source events are inferred according to a preset event causal relationship rule set, and an event causal chain is constructed, thereby revealing the internal logical relationship between events. For each original timestamp, the confidence is evaluated based on the data sources, signal quality indicators, environmental context in which the vehicle is located, and matching degree of the event causal chain, thereby obtaining the confidence of each timestamp, and effectively quantifying the reliability of the timestamp. Finally, the timestamps of the multi-source events are dynamically corrected based on the constructed event causal chain and the evaluated confidence.

[0054] By the technical solution, the application effectively solves the problems of non-linear misplacement and local absence of multi-source data timestamps caused by electromagnetic interference, driver operation habits and limitations of existing "timestamp alignment rules". Specifically, by introducing event causal chain inference and timestamp confidence evaluation mechanism, the application can more accurately capture and correct the delay fluctuations caused by intermittent electromagnetic interference, which have random and nonlinear characteristics, overcoming the shortcomings of existing linear compensation or simple sliding window matching. At the same time, by dynamically correcting the timestamps, the application can eliminate the "misplacement" of data from different sources in the time dimension, so that the emergency braking events detected by the internal sensors can be accurately associated with the external events that caused the braking, thereby making up for the problems of "information gap" and "local incompleteness" in the prior art. As can be seen, the application significantly improves the data fusion accuracy of the vehicle-mounted data analysis system, enhances the prediction ability of the energy consumption prediction logic for the real-world dynamic traffic environment, and provides more accurate single-vehicle state and real-time road condition information for priority judgment of vehicle fleet dispatching instructions, ultimately improving the overall new energy vehicle fleet operation efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0055] Figure 1 is a vehicle-mounted data analysis method flowchart provided by an embodiment of the application;

[0056] Figure 2 is another vehicle-mounted data analysis method flowchart provided by an embodiment of the application;

[0057] Figure 3 is a vehicle-mounted data analysis system structure diagram provided by an embodiment of the application. DETAILED DESCRIPTION

[0058] The technical solutions in the application will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the application, rather than all the embodiments. The components of the application described and shown in the accompanying drawings can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but only represents selected embodiments of the application. Based on the embodiments of the application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of the application.

[0059] It should be noted that: similar reference numbers and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings. Meanwhile, in the description of the application, the terms "first", "second", etc. are only used to distinguish the description, and cannot be understood as indicating or implying relative importance.

[0060] Traditional existing vehicle-mounted data analysis systems mainly rely on preset timestamp alignment rules to calibrate timestamps through linear compensation or simple sliding window matching when processing multi-source data. However, in a complex electromagnetic interference environment, the timestamps of V2X communication data may exhibit non-linear, random fluctuations, leading to subtle deviations that are difficult to correct between high-precision data inside the vehicle. Such deviations cause different sources of data to be "misaligned" in the time dimension, affecting the accuracy of data fusion and reducing the reliability of energy consumption prediction and fleet scheduling.

[0061] To this end, the present application proposes a vehicle-mounted data analysis method, comprising:

[0062] Identifying multi-source events; multi-source events include internal events and external V2X events, and recording the original timestamps and data sources of multi-source events;

[0063] According to the preset event causal relationship rule set, the causal relationship between multi-source events is inferred to obtain an event causal chain;

[0064] Conducting confidence assessment on each original timestamp of the multi-source events to obtain the confidence of each original timestamp; the confidence assessment is based on the data source, signal quality indicator, environmental context of the vehicle, and matching degree of the event causal chain;

[0065] Based on the event causal chain and the confidence, the timestamps of the multi-source events are dynamically corrected.

[0066] By introducing the event causal relationship inference and timestamp confidence assessment mechanism, the present application can effectively identify and correct the non-linear time deviation between multi-source data, thereby significantly improving the accuracy and reliability of vehicle-mounted data analysis and providing more accurate data support for intelligent fleet management.

[0067] In order to more easily and clearly understand the vehicle-mounted data analysis method proposed by the present application, the key terms and implementation environment involved therein will be described in detail below. This method is mainly applied to a vehicle-mounted data analysis system, which is usually deployed on new energy vehicles and is responsible for integrating and processing various data generated during vehicle operation.

[0068] where "multi-source events" refer to events detected or reported by different data sources during the vehicle's journey. These events can be divided into two categories: one is "internal events", mainly from the vehicle's internal sensors, such as sudden braking, stopping, steering, and other driving behavior events, as well as battery charging and discharging status and other vehicle state events. The other is "external V2X events", mainly from the information obtained by the vehicle and the external environment through V2X (Vehicle-to-Everything) communication, such as the status of the front traffic signal, the degree of road congestion, and the front obstacle warning, etc.

[0069] "Original timestamp" refers to the initial time information recorded by the data source system when each event is detected or received. These timestamps are the basis for the order and duration of events, but may be biased due to transmission delays, system clock synchronization, and other factors.

[0070] "Data source" refers to the entity that generates or reports events, such as the vehicle's internal CAN bus, radar sensors, cameras, or external V2X communication modules, road side units (RSU), etc. Different data sources may have different time accuracy and reliability.

[0071] "Event causal relationship rule set" is a set of predefined logical rules that describe the possible causal relationships between different events. For example, "front obstacle warning" may lead to "sudden braking event", or "red light at the intersection" may lead to "stopping event". These rules are the basis for inferring the causal chain of events.

[0072] "Event causal chain" refers to a sequence formed by connecting a series of multi-source events with causal relationships by applying the event causal relationship rule set. It reveals the logical order and mutual influence of event occurrence.

[0073] "Confidence" is a quantitative evaluation of the reliability of each original timestamp. It takes into account the reliability of the data source, signal quality, the environmental context of the vehicle, and the matching degree of the event causal chain, etc., reflecting the accuracy of the timestamp.

[0074] "Dynamic correction" refers to the real-time adjustment and optimization of the original timestamp based on the event causal chain and confidence, to eliminate or reduce time bias, making the time information of multi-source events more consistent and accurate.

[0075] This method aims to solve the problem of nonlinear misplacement and local loss of multi-source data timestamps in complex electromagnetic interference environments, by identifying multi-source events, inferring causal relationships, evaluating timestamp confidence, and dynamically correcting, thereby improving the accuracy and reliability of vehicle-mounted data analysis.

[0076] Reference Figure 1The application provides a vehicle-mounted data analysis method, comprising the following steps:

[0077] S1, identifying multi-source events and recording original timestamps and data sources of the multi-source events.

[0078] The multi-source events include internal events and external V2X events.

[0079] As a possible implementation manner, the system can preset a series of event trigger conditions. For example, when the vehicle internal sensor detects that the vehicle speed sharply decreases in a short time, it can be identified as an "emergency braking event"; when the vehicle remains stationary for a long time, it can be identified as a "parking event".

[0080] Specifically, if the vehicle speed decreases by more than 5 m / s in 0.5 seconds, it is identified as an emergency braking event.

[0081] If the vehicle speed is 0, it is identified as a parking event.

[0082] For external V2X events, the system can listen to the messages received by the V2X communication module, for example, when a message that the front traffic signal light state changes to red is received, it can be identified as a "signal light red event"; when a road intersection congestion warning message is received, it can be identified as a "road intersection congestion event".

[0083] When these events are identified, the system records the original timestamps and data sources of each event, for example, the timestamp of the internal event can come from the time synchronization module of the vehicle CAN bus, and the timestamp of the external V2X event comes from the timestamp of the data packet received by the V2X communication module.

[0084] S2, according to the preset event causal relationship rule set, deduce the causal relationship between the multi-source events, and obtain an event causal chain.

[0085] As a possible implementation manner, the system can maintain a database containing a variety of causal relationship rules. For example, a rule can be defined as: if the "front obstacle warning event" is earlier in time than the "emergency braking event", and the time interval between the two is within a certain range, then the front obstacle warning is considered to be the cause of the emergency braking. Another rule can be: if the "road intersection signal light red event" is earlier in time than the "parking event", and the time interval between the two is within another preset range, then the signal light red is considered to be the cause of the parking. After identifying the multi-source events, the system will traverse these rules to try to connect events with potential causal relationship, thereby constructing one or more event causal chains. For example, when "front obstacle warning" is detected, followed by "emergency braking", and then "vehicle deceleration", the system can deduce an event causal chain composed of the three events.

[0086] Specifically, if the front obstacle warning is earlier in time than the hard braking event, and the time interval between them is between 0 ms and 500 ms, then there is a causal relationship between them;

[0087] If the red light at the intersection is earlier in time than the stop event, and the time interval between them is between 0 ms and 2000 ms, then there is a causal relationship between them;

[0088] If the received signal strength indication (RSSI) reported by the V2X module is lower than -90 dBm or the bit error rate (BER) is higher than 10 -3 and is close in time to the V2X data packet transmission delay or loss, and the current location of the vehicle is in a predefined high-interference area, then there is a causal relationship between them.

[0089] Specifically, the above rule set aims to accurately identify the causal relationship between different multi-source events by defining a series of specific timing and scenario conditions. Among them, the first rule "if the front obstacle warning is earlier in time than the hard braking event, and the time interval between them is between 0 ms and 500 ms, then there is a causal relationship between them" is used to capture the typical reaction behavior of the driver to the front obstacle. The front obstacle warning is usually issued by the vehicle's perception system (such as radar, camera), while the hard braking event reflects the actual braking behavior of the vehicle. The time interval between the two is limited to 0 ms to 500 ms, which is based on the average reaction time of human drivers to ensure that the inferred causal relationship has high reasonableness and accuracy.

[0090] Further, the second rule "if the red light at the intersection is earlier in time than the stop event, and the time interval between them is between 0 ms and 2000 ms, then there may be a causal relationship between them" aims to identify the potential association between traffic signal changes and vehicle stopping behavior. The red light at the intersection is a kind of external V2X event, while the stop event is an internal event. Considering that the reaction time of the driver to the traffic signal may vary due to various factors (such as attention, speed, road conditions), the time interval is set to a wider range of 0 ms to 2000 ms, and the expression "may exist causal relationship" is used to reflect the non-deterministic or probabilistic nature of this association.

[0091] In addition, the third rule "if the received signal strength indication (RSSI) reported by the V2X module is lower than -90 dBm or the bit error rate (BER) is higher than 10 -3, and the vehicle's current location is in a predefined high-interference area, then there may be a cause-effect relationship between them. The received signal strength indication (RSSI) and the bit error rate (BER) are key indicators of the quality of V2X signals. When these indicators fall below preset thresholds (e.g., RSSI below -90 dBm or BER above 10 -3 ), it indicates that the V2X communication environment is poor, which is prone to cause transmission delays or losses of data packets. By correlating these signal quality problems with abnormal behaviors of V2X data packets and high-interference areas where the vehicle is located, the potential causes of V2X data event abnormalities can be inferred, providing important contextual information for subsequent timestamp corrections.

[0092] In some preferred embodiments, the application is implemented as follows:

[0093] Suppose a vehicle detects a sudden obstacle in front of it at time T1 through its internal sensors and issues a front obstacle warning. Then, at time T1 + 300 milliseconds, the vehicle's brake pedal depth sensor detects that the driver has performed an emergency braking operation, identifying it as an emergency braking event. According to the cause-effect relationship rules of the application, since the front obstacle warning is earlier than the emergency braking event and the time interval between them is 300 milliseconds, which falls within the range of 0 milliseconds to 500 milliseconds, the system will infer that there is a clear cause-effect relationship between the front obstacle warning and the emergency braking event.

[0094] For another example, when the vehicle approaches an intersection, the V2X module receives information that the intersection signal light turns red at time T2. Then, at time T2 + 1500 milliseconds, the vehicle's speed sensor shows that the vehicle's speed has dropped to 0, identifying it as a stop event. According to the rules of the application, since the intersection signal light turning red is earlier than the stop event and the time interval between them is 1500 milliseconds, which falls within the range of 0 milliseconds to 2000 milliseconds, the system will infer that there may be a cause-effect relationship between the intersection signal light turning red and the stop event.

[0095] For another example, in a certain area, the vehicle's V2X module reports that the received signal strength indication (RSSI) is continuously below -90 dBm, while the bit error rate (BER) is above 10 -3 . During this period, the system observes that V2X data packets have transmission delays and losses, and the vehicle's current location is identified as a predefined high-interference area. According to the rules of the application, the system will infer that there may be a cause-effect relationship between the signal quality problems of the V2X module and the abnormal transmission of V2X data packets, indicating that the abnormal behavior of external V2X events may be caused by the deterioration of the communication environment, rather than the absence or irrelevance of the event itself. These specific rule applications make the construction of event cause-effect chains more accurate and intelligent.

[0096] S3, performing confidence evaluation on each original timestamp of the multi-source event to obtain a confidence of each original timestamp.

[0097] The confidence evaluation is based on data sources, signal quality indicators, environmental context in which the vehicle is located, and matching degree of the event causal chain.

[0098] As a possible implementation, for each event in the multi-source event, an original confidence of the original timestamp of each event is obtained;

[0099] When the RSSI is lower than the RSSI threshold, the original confidence is reduced by a first step size;

[0100] When the BER is higher than the BER threshold, the original confidence is reduced by a second step size;

[0101] In the case where the event forms an event causal chain with a high-confidence event, the original confidence is increased by a third step size; the high-confidence event is an event with an original confidence greater than a confidence threshold.

[0102] S4, dynamically correcting the timestamps of the multi-source event based on the event causal chain and the confidence.

[0103] As a possible implementation, the system can first evaluate the overall reliability of each event causal chain, for example, by calculating the average confidence of all event timestamps in the causal chain. Then, the system can select the event causal chain with the highest confidence as the reference for correction according to the evaluation result. For each event in the causal chain, especially those external V2X events with low confidence, the system can use the timestamps of other high-confidence events in the causal chain as a reference, combined with a preset time delay model or driver reaction time model, to adjust the original timestamp. For example, if a causal chain shows that "front obstacle warning" leads to "emergency braking", and the timestamp confidence of "emergency braking" is high, while the timestamp confidence of "front obstacle warning" is low, the system can adjust the original timestamp of "front obstacle warning" forward or backward according to the average reaction time of the driver, so that it is more logically consistent with the timestamp of "emergency braking". This dynamic correction process is continuous and can adapt to changes in the vehicle operating environment, optimizing the time consistency of multi-source data in real time.

[0104] The vehicle-mounted data analysis method proposed in this application integrates multi-source event identification, event causal relationship inference, timestamp confidence evaluation, and dynamic correction, forming a closed-loop, self-adaptive data processing process, aiming to solve the problem of nonlinear misplacement and local missing of multi-source data timestamps in the prior art.

[0105] Specifically, when the vehicle is running in a complex environment, the system first "identifies multi-source events", including internal events from vehicle internal sensors (such as vehicle speed, brake pedal depth, steering angle, etc.) and external V2X events from V2X communication (such as front traffic signal status, road intersection congestion degree, front obstacle warning, etc.). In the identification process, the "original timestamp and data source" of each event is accurately recorded. For example, when the vehicle internal sensor detects a sudden brake, the occurrence time of the event and its information from the internal sensor are recorded; at the same time, when the V2X module receives a front obstacle warning, its receiving time and its information from V2X communication are also recorded.

[0106] Subsequently, "according to the preset event causal relationship rule set, the causal relationship between multi-source events is inferred, and the event causal chain is obtained". For example, if the system detects that the "front obstacle warning" event is earlier than the "sudden brake" event in time, and the time interval between the two is within the preset range, it is inferred that there is a causal relationship between the two, and they are connected into an event causal chain. The construction of such causal chain provides an important logical basis for subsequent timestamp correction.

[0107] Then, "the confidence of each original timestamp of multi-source events is evaluated, and the confidence of each original timestamp is obtained". This evaluation process is dynamic and multi-dimensional, which is "based on data source, signal quality index, vehicle environment context and matching degree of event causal chain". For example, the timestamp from a high-precision internal sensor usually has a high initial confidence; while the timestamp of an external V2X event, its confidence will be affected by V2X signal strength (RSSI), bit error rate (BER) and other signal quality indicators, in the "environment context" such as serious electromagnetic interference industrial park, its confidence may be further reduced. At the same time, if the timestamp of an event is highly consistent with other high-confidence events in the "event causal chain" in which it is located in time logic, the confidence of the event timestamp will also be improved. For example, if the timestamp confidence of the "sudden brake" event is high, and it forms a strong causal chain with the "front obstacle warning" event, the timestamp confidence of the "front obstacle warning" event will also be improved accordingly.

[0108] Finally, "Dynamic Correction of Multi-source Event Timestamps Based on Event Causal Chain and Confidence". The system will comprehensively consider the logicality of the event causal chain and the confidence of each timestamp, and adjust those timestamps with low confidence or deviation from the logicality of the causal chain. For example, for a causal chain consisting of "obstacle ahead warning" and "emergency brake", if the timestamp confidence of "emergency brake" is high, and the timestamp confidence of "obstacle ahead warning" is low, the system can use the timestamp of "emergency brake" as a reference, combined with the average reaction time of the driver, to correct the original timestamp of "obstacle ahead warning", making it more consistent with the causal logic in time. This dynamic correction mechanism can effectively compensate for the timestamp deviation caused by the nonlinear delay and data loss of V2X communication, ensuring the consistency of multi-source data in the time dimension.

[0109] Through the above-mentioned cooperation, the method of the present application can overcome the limitations of the existing "timestamp alignment rule" in dealing with nonlinear and random time deviation. It not only can identify multi-source events, but more importantly, it can understand the causal relationship between events and quantitatively evaluate the reliability of each timestamp, and finally realize dynamic and accurate correction of the timestamp. This enables the vehicle-mounted data analysis system to obtain higher quality and more accurate data sets, thereby significantly improving the accuracy of energy consumption prediction, the efficiency of vehicle fleet scheduling, and the adaptability to complex traffic scenarios, effectively solving the "multi-source data timestamp nonlinear misalignment and local loss" problem described in the foregoing technical problems, and providing a solid technical foundation for the intelligent management of new energy vehicle fleets.

[0110] In one possible design, as shown in Figure 2 To determine image region blur, the present application can further include the following steps:

[0111] S101, acquiring original data from vehicle internal sensors and external V2X communication.

[0112] The original data includes at least one of the following: vehicle speed, brake pedal depth, steering angle, battery charge and discharge state, front traffic signal light state, road intersection congestion degree, and front obstacle warning.

[0113] The raw data from the vehicle internal sensors and external V2X communication is obtained by a data acquisition module integrated in the vehicle system, which collects various types of information generated during the vehicle operation in real time. Specifically, the vehicle internal sensors can include but are not limited to speed sensors, brake pedal position sensors, steering angle sensors, battery management systems (BMS), etc., for obtaining internal operating parameters such as vehicle speed, brake pedal depth, steering angle, battery charge and discharge status, etc. At the same time, the external V2X communication module is responsible for receiving communication information from other vehicles (V2V), roadside units (V2I) or pedestrian devices (V2P), such as the status of the front traffic signal, the degree of road congestion and the warning of the front obstacle, etc. These raw data provide a comprehensive information base for subsequent event recognition.

[0114] S102, identifying a multi-source event according to the raw data.

[0115] Specifically, if the vehicle speed decreases by more than 5 meters per second within 0.5 seconds, it is identified as a sudden braking event;

[0116] If the vehicle speed is 0, it is identified as a parking event.

[0117] The scheme of the present application provides specific quantitative standards for the identification of multi-source events, making the event recognition process more accurate and automated. By setting "vehicle speed decreases by more than 5 meters per second within 0.5 seconds" as the identification condition of sudden braking event, and "vehicle speed is 0" as the identification condition of parking event, the key events with clear physical meaning and driving behavior characteristics can be effectively extracted from the continuous raw data stream. This identification method based on specific threshold avoids ambiguous judgment, ensures the accuracy and consistency of the identified events, and lays a solid foundation for subsequent event causal chain construction, timestamp confidence assessment and dynamic correction.

[0118] Through the above technical scheme, the present application can significantly improve the accuracy and reliability of multi-source event recognition. By introducing specific quantitative identification rules, such as the clear definition of sudden braking event and parking event, the misidentification and missed identification can be effectively reduced. This precise event recognition capability makes the subsequent event causal relationship inference more accurate, the confidence assessment of timestamp more reliable, and ultimately improves the overall precision and practicality of vehicle data analysis, providing more reliable data support for driving assistance systems and automatic driving decisions.

[0119] In some embodiments of the present application, the causal relationship between the multi-source events is inferred according to the preset event causal relationship rule set. However, in actual application, if the event causal relationship rule set is not specific enough or fails to fully consider the timing logic and environmental factors of event occurrence, the accuracy of the causal relationship inference may be insufficient, thereby affecting the reliability of subsequent timestamp correction.

[0120] In a possible design, in order to obtain the confidence of each original timestamp based on confidence evaluation of each original timestamp of the multi-source events, the method further includes:

[0121] S201. For each event in the multi-source events, an original confidence of an original timestamp of each event is obtained.

[0122] As a possible implementation manner, the system pre-stores a confidence table, and the confidence table includes the relationship between different events and different event timestamp confidences. The original confidence of the original timestamp of each event can be obtained according to the confidence table.

[0123] S202. When the RSSI is lower than the RSSI threshold, the original confidence is reduced by a first step.

[0124] S203. When the BER is higher than the BER threshold, the original confidence is reduced by a second step.

[0125] S204. In the case that the event forms an event causal chain with a high-confidence event, the original confidence is increased by a third step.

[0126] The high-confidence event is an event with an original confidence greater than a confidence threshold.

[0127] When the RSSI is lower than the preset RSSI threshold, it indicates that the signal reception quality is poor, and the data transmission may be interfered. At this time, the reliability of the timestamp is correspondingly reduced, and therefore the original confidence is reduced by a first step. The first step can be a preset fixed value, for example, 0.1, or dynamically adjusted according to the degree of signal strength decrease. Similarly, when the BER is higher than the preset BER threshold, it indicates that there are more data transmission errors, and the accuracy of the timestamp may be affected. At this time, the original confidence is reduced by a second step, for example, the second step is 0.15. For example, the third step can be 0.2.

[0128] The scheme of the present application effectively solves the possible deficiencies of the time stamp confidence evaluation of the traditional method in the face of dynamically changing signal quality and complex causal relationship by introducing a dynamic confidence adjustment mechanism for the original time stamp. Specifically, when the V2X communication signal quality decreases due to environmental interference or distance attenuation (manifested as RSSI below a threshold or BER higher than a threshold), the system can timely identify and accordingly reduce the confidence of the relevant event time stamp, thereby avoiding the deviation of subsequent analysis caused by low-quality data. At the same time, by utilizing the internal logic of the event causal chain, when an event forms a reasonable causal relationship with a confirmed high-confidence event, the time stamp confidence of the event will be improved, which is equivalent to utilizing the internal redundancy information and logical consistency of the system to cross-verify the reliability of the time stamp. This two-way adjustment mechanism makes the confidence evaluation of the time stamp more refined and adaptive.

[0129] Through the above technical scheme, the present application can significantly improve the accuracy and robustness of the multi-source event time stamp confidence evaluation in vehicle-mounted data analysis. Especially in complex traffic environment and variable network conditions, the method can dynamically reflect the real reliability of the time stamp, effectively reducing the error caused by poor signal quality or data uncertainty. In addition, by utilizing the matching degree of the event causal chain to enhance the confidence, the system can better utilize the logical association between events, thereby improving the reliability of event causal chain inference and time stamp dynamic correction as a whole, providing a solid data foundation for more accurate driving assistance and automatic driving decision-making.

[0130] In some preferred embodiments, the following is illustrated by a specific example. Assume that a vehicle receives an external V2X event about the status of a traffic signal ahead during driving, and records its original timestamp. The initial evaluation may assign a medium confidence to this timestamp, for example, 0.7. However, if the RSSI reported by the V2X module at this time is lower than the preset RSSI threshold, for example, -90 dBm, the system will reduce the confidence of the original timestamp of the external V2X event by the first step (for example, 0.1), making it 0.6. At the same time, if the vehicle's internal sensors identify a "stop event" with a high confidence (for example, 0.9), and according to the preset event causal relationship rule set, the "traffic signal ahead turns red" event and the "stop event" form a reasonable event causal chain in time and logic, and the "stop event" is determined to be a high-confidence event (original confidence 0.9 is greater than confidence threshold 0.8). In this case, even if the V2X signal quality is poor, since the external V2X event forms a causal chain with the high-confidence internal event, the system will increase the confidence of the original timestamp of the external V2X event by the third step (for example, 0.05), making it change from 0.6 to 0.65. Through this dynamic adjustment, even in the case of signal quality fluctuations, the system can take into account multiple factors to obtain a more accurate and reliable timestamp confidence.

[0131] In some embodiments of the application described above, a scheme for dynamically correcting the timestamps of multi-source events based on event causal chains and confidence is proposed. However, in actual applications, especially in complex traffic environments, there may be multiple potential event causal chains, and the reliability of these causal chains may differ. If the most reliable event causal chain cannot be effectively evaluated and selected, direct timestamp correction may result in inaccurate correction results, especially for external V2X events that are easily affected by external factors, the accuracy of their time information will be difficult to fully guarantee.

[0132] To this end, the application further proposes that the above-mentioned dynamic correction of the timestamps of multi-source events based on event causal chains and confidence includes:

[0133] S301, evaluate the spatial matching score and the time logicality score of the event causal chain hypothesis, and perform weighted summation to obtain a support score.

[0134] Specifically, the spatial matching score satisfies the following relationship:

[0135] K = max(0, 1-L-Z);

[0136] L is the ratio of the distance between the external V2X event position and the vehicle position and the preset maximum relevant distance; Z is the ratio of the direction angle between the direction of the vehicle pointing to the external V2X event position and the driving direction of the vehicle and 180 degrees.

[0137] The preset maximum relevant distance can be set according to the actual application scene and the effective range of V2X communication, for example, can be set to 50 meters, 100 meters or 200 meters, etc.

[0138] The time logic score satisfies the following relationship:

[0139] If the external V2X event occurrence time is before the anchor point event occurrence time, and the time interval between the two falls within the driver reaction time range, the time logic score is 1, otherwise 0.

[0140] S302, determining the optimal event causal chain according to the support score.

[0141] Among them, the optimal event causal chain is the event causal chain with the highest support score.

[0142] S303, calibrating the time information of the external V2X event according to the optimal event causal chain.

[0143] Specifically, for the internal event and the external V2X event of the optimal event causal chain, the timestamp of the internal event is used as the time reference;

[0144] According to the difference between the timestamp of the internal event and the average driver reaction time, the timestamp of the external V2X event is determined.

[0145] Through the above technical solution, the present application can overcome the problem of inaccurate correction caused by the lack of event causal chain reliability evaluation mechanism in the traditional method when processing multi-source event timestamp correction. Specifically, by introducing the spatial matching degree score and the time logic score, and combining the weighted sum to obtain the support score, the present application can finely evaluate multiple potential event causal chains, thereby accurately identifying the optimal event causal chain that best meets the actual situation. This mechanism ensures that the timestamp correction is based on a verified and highly reliable causal relationship, significantly improving the calibration accuracy of the external V2X event time information and the overall robustness of the system. Thus, a more accurate and reliable time reference can be provided for vehicle data analysis, thereby supporting more accurate driving behavior analysis, environmental perception and decision making.

[0146] In some preferred embodiments, a specific example is given below. Suppose that while a vehicle is in motion, it simultaneously receives a "forward emergency braking warning" event from V2X communication and a "driver's emergency braking" event from an internal vehicle sensor. The system first infers multiple possible causal chains of events based on preset rules, such as:

[0147] 1. Causal chain A: Emergency braking warning ahead -> Driver brakes suddenly.

[0148] 2. Causal chain B: Driver brakes suddenly (no external cause).

[0149] To determine which causal chain is superior, the system evaluates the spatial matching score and temporal logicality score for each causal chain hypothesis. For causal chain A, assuming the reported emergency braking location of the external V2X event is very close to the vehicle's current location and the warning direction is consistent with the vehicle's travel direction, its spatial matching score is high. Simultaneously, if the V2X warning time is earlier than the driver's emergency braking time, and the time interval falls within the driver's average reaction time (e.g., 200 milliseconds to 1000 milliseconds), its temporal logicality score is also high. For causal chain B, since there is no external V2X warning as a trigger, its spatial matching score and temporal logicality score related to the external V2X event may be low or inapplicable.

[0150] Assume the spatial fit score has a weight of 0.6 and the temporal logicality score has a weight of 0.4. If causal chain A has a spatial fit score of 0.9 and a temporal logicality score of 0.8, then its support score... If the support score of causal chain B is assumed to be 0.3 (since it lacks support from external V2X events), then by comparison, the support score of causal chain A (0.86) is higher than that of causal chain B (0.3), therefore causal chain A is determined to be the optimal event causal chain.

[0151] Once causal chain A is determined to be the optimal event causal chain, the system uses the timestamp of the internal event of the driver's emergency braking as a time reference. The original timestamps of external V2X events are calibrated based on the driver's average reaction time (e.g., assumed to be 500 milliseconds). For example, if the driver's emergency braking occurs at time T, the calibrated timestamp for the V2X emergency braking warning will be determined as T - 500 milliseconds, thus more accurately reflecting the actual occurrence time of the external V2X event and eliminating potential delays from V2X communication. In this way, even with transmission delays in V2X data, its timestamps can be accurately calibrated, ensuring the accuracy of data analysis.

[0152] like Figure 3 As shown in the figure, this embodiment of the invention also provides an in-vehicle data analysis system. The system includes:

[0153] a recognition module configured to recognize multi-source events, the multi-source events including internal events and external V2X events, and record original timestamps and data sources of the multi-source events;

[0154] a causal inference module configured to infer causal relationships between the multi-source events according to a preset event causal relationship rule set, and obtain an event causal chain;

[0155] a confidence evaluation module configured to evaluate confidence of each original timestamp of the multi-source events, and obtain confidence of each original timestamp, the confidence evaluation being based on the data sources, a signal quality indicator, an environmental context in which the vehicle is located, and a matching degree of the event causal chain;

[0156] a correction module configured to dynamically correct the timestamps of the multi-source events based on the event causal chain and the confidence.

[0157] The embodiments of the present application further provide a computer readable storage medium. All or part of the processes of the method embodiments above can be instructed by a computer program to relevant hardware to complete, the program can be stored in the computer readable storage medium, and the program can include the processes of the method embodiments above when executed. The computer readable storage medium can be an internal storage unit of the task execution device (including the data sending end and / or the data receiving end) of any of the foregoing embodiments, such as a hard disk or a memory of the task execution device. The computer readable storage medium can also be an external storage device of the terminal device, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. Further, the computer readable storage medium can include both the internal storage unit of the task execution device and the external storage device. The computer readable storage medium is used to store the computer program and other programs and data required by the task execution device. The computer readable storage medium can also be used to temporarily store data that has been output or will be output.

[0158] In addition, each functional unit in each embodiment of the present application can be integrated in one processing unit, or each unit can exist physically, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of a software functional unit.

[0159] The integrated unit, if implemented in the form of a software function unit and sold or used as an independent product, can be stored in a readable storage medium. Based on such understanding, the technical solutions of the embodiments of the present application essentially or say the part that contributes to the prior art or the whole or part of the technical solutions can be embodied in the form of a software product. The software product is stored in a storage medium, includes a plurality of instructions to make a device (which can be a single-chip microcomputer, a chip, etc.) or a processor execute all or part of the steps of the embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a ROM, a RAM, a magnetic disk or an optical disk, and various storage program codes.

[0160] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any change or replacement within the technical scope disclosed in the present application should be covered within the protection scope of the present application.

Claims

1. A method of in-vehicle data analysis, characterized by, The method comprises: identifying multi-source events and recording original timestamps and data sources of the multi-source events; the multi-source events include internal events and external V2X events; inferring causal relationships between the multi-source events according to a preset event causal relationship rule set to obtain an event causal chain; performing confidence evaluation on each original timestamp of the multi-source events to obtain a confidence of each original timestamp; the confidence evaluation is based on the data sources, signal quality indicators, environmental context in which the vehicle is located, and matching degree of the event causal chain; performing dynamic correction on the timestamps of the multi-source events based on the event causal chain and the confidence; the dynamic correction on the timestamps of the multi-source events based on the event causal chain and the confidence comprises: evaluating a spatial matching score and a time logic score of the event causal chain hypothesis, and performing weighted summation to obtain a support score; determining an optimal event causal chain according to the support score; the optimal event causal chain is an event causal chain with the highest support score; calibrating time information of the external V2X events according to the optimal event causal chain; the calibration of the time information of the external V2X events according to the optimal event causal chain comprises: using a timestamp of an internal event of the optimal event causal chain as a time reference; determining a timestamp of an external V2X event according to a difference between the timestamp of the internal event and an average reaction time of a driver.

2. The method of claim 1, wherein, The method comprises: obtaining original data from internal sensors of a vehicle and external V2X communication; the original data includes at least one of the following: vehicle speed, brake pedal depth, steering angle, battery charging and discharging state, front traffic signal light state, road intersection congestion degree, and front obstacle warning; identifying the multi-source events according to the original data.

3. The in-vehicle data analysis method of claim 2, wherein, The identification of the multi-source events according to the original data comprises: if the vehicle speed decreases by more than 5 m / s within 0.5 s, an emergency braking event is identified; if the vehicle speed is 0, a stop event is identified.

4. The in-vehicle data analysis method of claim 3, wherein, The inference of the causal relationships between the multi-source events according to the preset event causal relationship rule set to obtain the event causal chain comprises: if the front obstacle warning is earlier than the emergency braking event in time, and the time interval between them is between 0 ms and 500 ms, a causal relationship exists between them; if the red light of the road intersection signal is earlier than the stop event in time, and the time interval between them is between 0 ms and 2000 ms, a causal relationship exists between them; if a received signal strength indicator (RSSI) reported by a V2X module is lower than -90 dBm or a bit error rate (BER) is higher than 10-3, and is close to a V2X data packet transmission delay or loss in time, and the current position of the vehicle is in a predefined high interference area, a causal relationship exists between them.

5. The in-vehicle data analysis method of claim 1, wherein, The confidence evaluation on each original timestamp of the multi-source events comprises: for each event in the multi-source events, obtaining an original confidence of each event original timestamp; decrease the original confidence level by a first step size when the RSSI is lower than an RSSI threshold value; decrease the original confidence level by a second step size when the BER is higher than a BER threshold value; increase the original confidence level by a third step size in a case that the event forms the event causal chain with a high confidence event; the high confidence event is an event with an original confidence level greater than a confidence threshold value.

6. The method of claim 1, wherein, the spatial matching score satisfies the following relationship: K = max(0, 1-L-Z); L is a ratio of a distance between the external V2X event position and the vehicle position to a preset maximum relevant distance; Z is a ratio of a direction angle between a direction in which the vehicle points to the external V2X event position and a driving direction of the vehicle to 180 degrees.

7. The in-vehicle data analysis method of claim 1, wherein, the temporal logic score satisfies the following relationship: if the external V2X event occurrence time is before the anchor event occurrence time, and a time interval between the two falls within a driver reaction time range, the temporal logic score is 1, otherwise, the temporal logic score is 0.

8. An in-vehicle data analysis system characterized by comprising: The system comprises: an identification module configured to identify multiple-source events; the multiple-source events comprise internal events and external V2X events, and record original time stamps and data sources of the multiple-source events; a causal inference module configured to infer causal relationships between the multiple-source events according to a preset event causal relationship rule set, to obtain an event causal chain; a confidence level evaluation module configured to evaluate a confidence level of each original time stamp of the multiple-source events; the confidence level evaluation is based on the data sources, signal quality indicators, an environmental context in which the vehicle is located, and a matching degree of the event causal chain; a correction module configured to dynamically correct time stamps of the multiple-source events based on the event causal chain and the confidence level; the correction module is further configured to: evaluate a spatial matching score and a temporal logic score of the event causal chain hypothesis, and perform weighted summation to obtain a support score; determine an optimal event causal chain according to the support score; the optimal event causal chain is an event causal chain with the highest support score; calibrate time information of the external V2X event according to the optimal event causal chain; the calibration of the time information of the external V2X event according to the optimal event causal chain comprises: using a time stamp of an internal event of the optimal event causal chain as a time reference; determining a time stamp of the external V2X event according to a difference between the time stamp of the internal event and an average driver reaction time.

Images