Data processing method for vehicle-mounted AI knowledge base

Abstract

The application provides a kind of data processing method for vehicle-mounted AI knowledge base, belongs to data processing technical field, it is first to realize the accurate digital quantification of the single knowledge obsolescence degree and update lag degree of vehicle-mounted AI knowledge base, abandon the fuzzy judgment mode of traditional artificial experience, output accurate value through corresponding mathematical expression, can accurately distinguish mild, moderate and severe obsolescence and lag grade, and the quantization accuracy is reproducible and has no subjective bias, solve the problem that the existing technology cannot quantify the data failure degree and update lag degree, provide accurate data support for the update and cleaning of knowledge base;Through three-dimensional obsolescence quantification and hierarchical update mechanism, it can accurately capture various implicit and explicit obsolete knowledge, intervene in mild obsolete knowledge in advance, and completely clean and replace severe obsolete knowledge, after using the application scheme, the proportion of obsolete data in vehicle-mounted knowledge base is reduced, and the long-term resident expired invalid data is eliminated.

Description

Technical Field

[0001] This invention belongs to the field of data processing technology, specifically relating to a data processing method for in-vehicle AI knowledge bases. Background Technology

[0002] With the rapid iteration of intelligent connected vehicle technology, in-vehicle AI systems have become an important carrier for vehicle intelligence, connectivity, and human-centered interaction. They are widely used in intelligent driving assistance, in-vehicle intelligent interaction, intelligent road condition analysis, in-vehicle service recommendation, and vehicle fault self-diagnosis scenarios. As the main data support for in-vehicle AI systems, the in-vehicle AI knowledge base stores multi-dimensional structured and semi-structured knowledge data, including road condition rules, traffic regulations, vehicle operation and maintenance parameters, in-vehicle interaction scripts, scenario service logic, and fault judgment standards. The real-time performance, accuracy, and even effectiveness of this data directly determine the decision-making accuracy, interactive experience, and operational safety of the in-vehicle AI system.

[0003] Current in-vehicle AI knowledge base data processing systems mostly adopt a fixed operation and maintenance model of offline batch entry and periodic manual updates. The overall update mechanism is lagging behind and lacks data timeliness control, resulting in significant technical defects such as outdated data and delayed updates. This makes it difficult to match complex and dynamic in-vehicle operation scenarios. The existing knowledge base update cycle is generally monthly or quarterly, and iteration can only be completed by manually importing new data in batches and deleting old data. This cannot match high-frequency dynamic data change scenarios such as traffic regulations updates, dynamic changes in road conditions, and urban road network reconstruction. At the same time, the existing technology lacks a quantitative judgment mechanism for the timeliness status of individual knowledge in the knowledge base. It cannot accurately distinguish between slightly outdated, severely outdated, and completely invalid knowledge data. It can only rely on general screening based on human experience, which easily leads to problems such as the accidental deletion of valid knowledge, long-term retention of outdated knowledge, and incomplete matching of new knowledge.

[0004] Specifically, the technical defects of existing technologies include: prominent data staleness issues. Knowledge such as traffic rules, road speed limits, and traffic restriction policy parameters in vehicle scenarios are highly time-sensitive. Fixed-period update mode results in a large amount of outdated and biased data residing in the knowledge base for a long time, causing problems such as AI interaction response errors, driving assistance judgment errors, and false alarms in fault detection; serious update lag. Existing technologies lack real-time data synchronization and incremental update triggering mechanisms, relying on manual operation and maintenance to trigger updates, which cannot respond to immediate data changes. Moreover, batch update mode consumes a lot of computing power and has low update efficiency. The computing power resources of vehicle edge terminals are limited, and frequent batch updates can easily lead to system lag and response delays.

[0005] In summary, the existing data processing technology of in-vehicle AI knowledge bases cannot meet the high real-time, high precision and high reliability requirements of AI operation in intelligent connected vehicles. There is an urgent need for a new data processing method that can accurately quantify the degree of data obsolescence, perform real-time incremental iteration and match the computing power of in-vehicle edges. Summary of the Invention

[0006] The purpose of this section is to outline some aspects of the embodiments of the present invention and to briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section, as well as in the abstract and title of the present application, to avoid obscuring the purpose of this section, the abstract and title of the invention. Such simplifications or omissions shall not be used to limit the scope of the present invention.

[0007] In view of the aforementioned existing problems, the present invention is proposed.

[0008] To solve the above-mentioned technical problems, the present invention provides the following technical solution: This invention provides a data processing method for in-vehicle AI knowledge bases, including: Standardized collection of knowledge metadata for in-vehicle AI knowledge base; The knowledge metadata is calculated using a knowledge obsolescence quantification calculation model; The knowledge update lag metric calculation model is applied to calculate the knowledge update lag. The criteria for determining the level of knowledge obsolescence and lag are based on the obsolescence and lag of knowledge updates.

[0009] Furthermore, a standardized method for collecting knowledge metadata for in-vehicle AI knowledge bases is implemented, specifically including:

[0010] For each piece of knowledge in the vehicle AI knowledge base, five types of objective metadata are collected. These five types of objective metadata are the knowledge metadata, which include the effective duration of the knowledge benchmark, the actual residence time of the knowledge, the frequency of knowledge scenario matching, the frequency of knowledge source iteration, and the deviation of effective knowledge matching.

[0011] Furthermore, the knowledge within the in-vehicle AI knowledge base includes traffic regulations, road network and traffic conditions, vehicle operation and maintenance, in-vehicle interaction, and scenario service knowledge.

[0012] Furthermore, the effective duration of the knowledge benchmark is the set effective duration of a single piece of knowledge in the in-vehicle AI knowledge base. This effective duration is the average update interval of similar knowledge in history, which is the average time span between two adjacent updates of this knowledge. The actual knowledge residence time is the duration of a single piece of knowledge in the in-vehicle AI knowledge base from the end of the last update to the current moment. The knowledge scenario matching frequency is the cumulative number of times a single piece of knowledge in the in-vehicle AI knowledge base is called, matched, and used for decision-making by the in-vehicle AI system. The knowledge source iteration frequency is the number of times a single piece of knowledge in the in-vehicle AI knowledge base has been updated by knowledge in the cloud server up to the current moment. The effective knowledge matching deviation is the deviation value between a single piece of knowledge in the in-vehicle AI knowledge base and the latest knowledge in the cloud server used to update that knowledge.

[0013] Furthermore, methods for calculating knowledge metadata using a knowledge obsolescence metric computational model specifically include: The obsolescence of knowledge is obtained by calculating the knowledge metadata using the mathematical expression of the knowledge obsolescence quantification calculation model.

[0014] Furthermore, the mathematical expression for the quantitative computational model of knowledge obsolescence is: ; In this mathematical expression, As to the obsolescence of knowledge; For the actual duration of knowledge retention; The effective duration of the knowledge benchmark; For the frequency of knowledge source iteration; Match frequency to knowledge scenarios; This is to address the bias in the effective matching of knowledge.

[0015] Furthermore, methods for calculating the knowledge update lag using a quantitative computational model include: The mathematical expression for constructing a quantitative calculation model of knowledge update lag is used to quantify the lag degree of a single knowledge update, that is, to calculate the knowledge update lag degree.

[0016] Furthermore, the mathematical expression for the quantitative computational model of knowledge update lag is: ; In this mathematical expression, The update lag of a single piece of knowledge. For the current moment, This refers to the moment when this piece of knowledge was most recently updated. For the duration of the knowledge benchmark in effect, For the frequency of knowledge source iteration, This represents the maximum frequency of knowledge source iterations for all knowledge in the same category as this knowledge.

[0017] Furthermore, the specific criteria for determining the level of knowledge obsolescence and lag include: In terms of the obsolescence of knowledge At that time, the obsolescence level of the knowledge was determined to be slightly obsolete; In terms of the obsolescence of knowledge At that time, the obsolescence level of the knowledge was determined to be moderately obsolete; In terms of the obsolescence of knowledge When this happens, the level of obsolescence of the knowledge is determined to be severely obsolete; Lag in knowledge updates In this case, the lag level of the knowledge update is determined to be slightly lagging; Lag in knowledge updates When this happens, the knowledge update lag level is determined to be moderate lag; Lag in knowledge updates When this happens, the knowledge update lag level is determined to be severely lagging.

[0018] Furthermore, the specific criteria for classifying knowledge as outdated and lagging include: When the obsolescence level of knowledge is determined to be slightly obsolete or the update lag level of knowledge is determined to be slightly lagging, periodic inspection and updates are performed on the knowledge. When the knowledge is determined to be moderately outdated or moderately lagging in terms of update lag, an incremental partial update is performed on the knowledge. When the obsolescence level of knowledge is determined to be severely obsolete or the update lag level is determined to be seriously lagging, the knowledge will be either fully replaced and updated or directly cleaned up and deleted.

[0019] The beneficial effects of the present invention are as follows, compared with the prior art: This invention achieves for the first time accurate digital quantification of the obsolescence and update lag of individual knowledge entries in an in-vehicle AI knowledge base. It abandons the fuzzy judgment mode of traditional human experience and outputs accurate values ​​through corresponding mathematical expressions. It can accurately distinguish between mild, moderate, and severe obsolescence and lag levels, with reproducible quantification accuracy and no subjective bias. This solves the problem that existing technologies cannot quantify the degree of data failure and update lag, providing accurate data support for knowledge base updates and cleaning. Through a three-dimensional obsolescence quantification and hierarchical update mechanism, it can accurately capture various implicit and explicit obsolete knowledge, intervene early in mildly obsolete knowledge, and thoroughly clean and replace severely obsolete knowledge. After adopting this invention, the proportion of obsolete data in the in-vehicle knowledge base is reduced, long-term expired and invalid data is eliminated, and the knowledge matching accuracy and decision-making accuracy of the in-vehicle AI system are improved. This invention improves efficiency and avoids problems such as interaction errors, driver assistance judgment errors, and false alarms caused by outdated data. It abandons the traditional fixed-cycle batch update mode and establishes a real-time monitoring and hierarchical triggering incremental update mechanism. Emergency delayed knowledge is triggered for updates at millisecond levels, while routine delayed knowledge is updated silently at regular intervals. Compared to the existing monthly or quarterly update modes, the knowledge update response latency is shortened, and the update lag problem is solved. At the same time, the incremental update mode avoids full data iteration, reducing the computing power consumption of the vehicle terminal for updates and effectively solving the system lag and response latency problems caused by batch updates, matching the computing power constraints of vehicle edge hardware. This invention achieves automated governance from metadata collection, timeliness quantification, level determination, and incremental updates to dynamic cleaning, without the need for manual operation and maintenance intervention, reducing the operation and maintenance costs of the vehicle knowledge base. Simultaneously, a local cloud-based two-way synchronous closed loop continuously optimizes the quantitative benchmark parameters, achieving dynamic iterative improvement in data processing accuracy. The timeliness, accuracy, and stability of the knowledge base data are continuously optimized, ensuring no data accumulation, no residual failures, and no update omissions during long-term operation. This invention avoids problems such as computational errors, judgment distortions, and logical conflicts, reducing the failure rate of the data processing module in the vehicle AI knowledge base. This enhances the operational safety and reliability of the intelligent connected vehicle's in-vehicle AI system, providing stable, accurate, and real-time data support for intelligent driving and intelligent interaction functions. Attached Figure Description

[0020] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. 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. Wherein: Figure 1 This is an overall flowchart of the data processing method for vehicle-mounted AI knowledge base described in this invention. Detailed Implementation

[0021] 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. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0022] Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without inventive effort should fall within the scope of protection of this invention.

[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] According to an embodiment of the present invention, in combination Figure 1 The flowchart shown illustrates a data processing method for an in-vehicle AI knowledge base, as described in this invention, comprising: S1. Standardize the collection of knowledge metadata for the in-vehicle AI knowledge base. Note the following in this step: In a preferred embodiment, the method for standardized collection of knowledge metadata for an in-vehicle AI knowledge base specifically includes: For each piece of knowledge in the vehicle AI knowledge base, five types of objective metadata are collected. These five types of objective metadata are the knowledge metadata, which include the effective duration of the knowledge benchmark, the actual residence time of the knowledge, the frequency of knowledge scenario matching, the frequency of knowledge source iteration, and the deviation of effective knowledge matching.

[0025] It should be noted that the in-vehicle AI system is the same as the in-vehicle intelligent system; the hardware devices of the in-vehicle AI system include the in-vehicle edge terminal, which is a controller with a built-in AI chip installed in the vehicle; the data processing method for the in-vehicle AI knowledge base runs on the in-vehicle edge terminal and is executed by the data processing module of the in-vehicle AI system, and the in-vehicle AI knowledge base is stored on the in-vehicle edge terminal; each piece of knowledge in the in-vehicle AI knowledge base can be assigned a unique knowledge ID; the in-vehicle edge terminal communicates with the cloud server, and the cloud server stores the latest knowledge of the same type as the various types of knowledge in the in-vehicle AI knowledge base, which is used to update the relevant types of knowledge.

[0026] In a preferred embodiment, the knowledge in the vehicle AI knowledge base includes traffic regulations, road network and traffic conditions, vehicle operation and maintenance, vehicle interaction, and scenario service knowledge.

[0027] It should be noted that traffic regulations can include six types: weekday license plate number-based traffic restrictions, holiday toll-free highway regulations, non-motorized vehicle restrictions on expressways, emergency lane traffic management regulations, urban truck traffic time restrictions, and traffic light priority rules. Weekday license plate number-based traffic restrictions refer to the traffic management rules implemented by local traffic management departments to alleviate urban congestion during designated hours on weekdays and within the main urban area, based on the last digit of the vehicle's license plate. These rules apply to adjusted workdays but not to statutory holidays and weekends. Holiday toll-free highway regulations refer to the regulations regarding which vehicles are exempt from tolls on national toll highways during the statutory holidays of Spring Festival, Qingming Festival, Labor Day, and National Day. The toll is determined based on the time the vehicle leaves the highway exit; the rules for prohibiting non-motorized vehicles on expressways refer to which vehicles are allowed to travel on urban expressways, elevated roads, or tunnel main lines which are dedicated to motorized vehicles for fast passage; the regulations for emergency lane traffic management refer to which lanes on highways and urban expressways are dedicated emergency rescue lanes, and that emergency rescue lanes are prohibited from being used for normal driving, waiting in congestion, or temporary rest; the rules for restricting the passage of trucks in urban areas are time-sharing control rules for large and medium-sized trucks, dump trucks, and freight vehicles, which are prohibited during the day or restricted at night to ensure the order of traffic on urban roads; the rules for traffic light priority refer to the general priority of going straight, turning left, and turning right when the traffic lights at intersections are in a green light state.

[0028] Furthermore, road network traffic condition knowledge can include four types: standard speed limit parameters for urban arterial roads, speed limit thresholds for highway tunnels, traffic parameters for bridge sections, and traffic parameters for steep slope sections. Standard speed limit parameters for urban arterial roads refer to the legally mandated maximum speed standards for urban arterial roads as specified in national road regulations. These parameters are updated with road reconstruction and traffic management optimization. Speed ​​limit thresholds for highway tunnels are the maximum speed limits for vehicles inside highway tunnels as stipulated by national traffic management regulations. The speed limits for some short tunnels, long tunnels, and special weather scenarios may be adjusted differently, and adjustments to tunnel management rules will lead to iterative updates of these parameters. Traffic parameters for bridge sections include set bridge speed limits, crosswind warning thresholds, or load limits. Traffic parameters for steep slope sections include uphill speed limits, downhill speed limits, or safe following distance standards on steep slopes. New regulations will cause delays in updating and staleness of traffic parameters for bridge sections and steep slope sections.

[0029] Furthermore, vehicle operation and maintenance knowledge can include three types: vehicle routine maintenance cycle, vehicle motor set operating temperature threshold, and vehicle fault code corresponding judgment rules. The vehicle fault code corresponding judgment rules are preset standard knowledge rules that define the fault type, triggering conditions, fault level, and troubleshooting basis for various types of vehicle fault codes.

[0030] Furthermore, in-vehicle interaction knowledge can include three types: navigation activation guidance text, vehicle fault prompt explanation text, and abnormal speed safety reminder text. Navigation activation guidance text refers to the standard guidance text associated with the navigation function of the in-vehicle AI system. After the user initiates the navigation operation, the in-vehicle AI system uses fixed interactive text content for route initialization, route broadcasting, function guidance, and status prompts. Vehicle fault prompt explanation text refers to fault popularization and explanation text knowledge. When the vehicle detects a fault, it uses standardized prompt text to explain the meaning, impact, and precautions of the fault to the user in layman's terms. Abnormal speed safety reminder text refers to standard prompt text knowledge. When conditions such as speeding, low-speed lane occupation, or abnormal speed fluctuations occur, the in-vehicle AI system actively broadcasts or displays safety warning prompts in pop-up windows.

[0031] Furthermore, scenario-based service knowledge can include four types: commuting route congestion warning service logic, highway driving safety reminder triggering rules, destination parking intelligent recommendation logic, and vehicle range-limited travel warning rules. The commuting route congestion warning service logic refers to the intelligent service logic that automatically identifies commuting times, matches real-time traffic congestion data, or predicts congestion trends for users' daily fixed commuting routes, triggering congestion broadcasts or route avoidance recommendations in advance. The highway driving safety reminder triggering rules refer to the standardized triggering rules where, after a vehicle enters a highway scenario, the onboard AI system determines driving risks based on highway speed thresholds, following distance, road segment attributes, or driving status, automatically triggering speeding reminders, fatigue reminders, special road segment warnings, or insufficient distance reminders. The destination parking intelligent recommendation logic refers to the service recommendation rules where, when a vehicle arrives near its destination, the onboard AI system intelligently filters, sorts, and pushes the best parking lot. The vehicle range-limited travel warning rules refer to the warning judgment rules where, based on the vehicle's determination that the range cannot cover the entire journey, the onboard AI system proactively triggers range-limited reminders or pushes charging point locations.

[0032] In a preferred embodiment, the knowledge baseline effective duration is the baseline effective duration of a single piece of knowledge in the vehicle-mounted AI knowledge base. This baseline effective duration is the average update interval of similar knowledge in history, which is the average time span between two adjacent updates of the knowledge. The actual knowledge residence time is the duration of a single piece of knowledge in the vehicle-mounted AI knowledge base from the end of the last update to the current moment. The knowledge scenario matching frequency is the cumulative number of times a single piece of knowledge in the vehicle-mounted AI knowledge base is called, matched, and used for decision-making by the vehicle-mounted AI system. The knowledge source iteration frequency is the number of times a single piece of knowledge in the vehicle-mounted AI knowledge base has been updated by knowledge in the cloud server up to the current moment. The effective knowledge matching deviation is the deviation value between a single piece of knowledge in the vehicle-mounted AI knowledge base and the latest knowledge in the cloud server used to update that knowledge.

[0033] It should be noted that the cumulative number of times a single piece of knowledge is invoked, matched, and used in decision-making by the in-vehicle AI system specifically includes the following five aspects of the effective count: When the knowledge is traffic regulations, the corresponding cumulative valid counts include the number of times the in-vehicle AI system uses this knowledge to verify traffic rules, execute traffic restriction judgments, speeding compliance checks, traffic management rule reminders, and traffic priority judgments; When the knowledge is road network and traffic condition knowledge, the corresponding cumulative valid counts include the number of times the onboard AI system applies the knowledge to match road speed limits and bridge or steep slope sections; When the knowledge is vehicle operation and maintenance knowledge, the corresponding cumulative valid counts include the number of times the in-vehicle AI system uses the knowledge to perform vehicle fault code matching and judgment, vehicle operating condition self-check, maintenance cycle reminder, and vehicle motor temperature parameter compliance verification. When the knowledge is in-vehicle interaction knowledge, the corresponding cumulative valid counts include the number of times the in-vehicle AI system uses the knowledge to perform user voice question and answer, navigation function operation explanation and broadcast, fault text prompts, driving safety reminder pop-ups and voice outputs; When this knowledge serves as scenario-based knowledge, the corresponding cumulative valid counts include the number of times the in-vehicle AI system applies this knowledge for commuter route matching, highway scenario safety matching, parking recommendation, low battery warning, and travel risk prediction. The cumulative valid number of times a single piece of knowledge is called, matched, and used for decision-making by the vehicle AI system does not include the vehicle AI system's boot preloading, background global scanning, traversal and retrieval of the vehicle AI knowledge base, cache refresh, silent pre-matching, and background behaviors that are read-only data and do not output any AI decisions or services. All of these are not included in the cumulative valid number of times.

[0034] Furthermore, the knowledge effective matching bias specifically includes: When a single piece of knowledge in the vehicle-mounted AI knowledge base is the city's weekday license plate number restriction rule, the deviation value between the single piece of knowledge in the vehicle-mounted AI knowledge base and the latest knowledge in the cloud server used to update that knowledge is the deviation value between the city's weekday license plate number restriction rule in the vehicle-mounted AI knowledge base and the latest city's weekday license plate number restriction rule in the cloud server used to update that rule. In other words, if the city's weekday license plate number restriction rule in the vehicle-mounted AI knowledge base and the latest city's weekday license plate number restriction rule in the cloud server are completely consistent, the deviation value is 0; otherwise, the deviation value is 1. When a single piece of knowledge in the vehicle-mounted AI knowledge base is the regulation of free highway passage during holidays, the deviation value between the single piece of knowledge in the vehicle-mounted AI knowledge base and the latest knowledge in the cloud server used to update the regulation of free highway passage during holidays is the deviation value between the regulation of free highway passage during holidays in the vehicle-mounted AI knowledge base and the latest regulation of free highway passage during holidays in the cloud server used to update the regulation of free highway passage during holidays. In other words, if the regulation of free highway passage during holidays in the vehicle-mounted AI knowledge base and the latest regulation of free highway passage during holidays in the cloud server are completely consistent, the deviation value is 0; otherwise, the deviation value is 1. When a single piece of knowledge in the vehicle-mounted AI knowledge base is a rule prohibiting non-motorized vehicles from using expressways, the deviation between the single piece of knowledge in the vehicle-mounted AI knowledge base and the latest knowledge in the cloud server used to update that knowledge is the deviation between the single piece of knowledge in the vehicle-mounted AI knowledge base and the latest rule in the cloud server used to update that rule. In other words, if the rule in the vehicle-mounted AI knowledge base and the latest rule in the cloud server used to update that rule are completely identical, the deviation is 0; otherwise, the deviation is 1. When a single piece of knowledge in the vehicle-mounted AI knowledge base is the emergency lane traffic management regulation, the deviation between the single piece of knowledge in the vehicle-mounted AI knowledge base and the latest knowledge in the cloud server used to update that knowledge is the deviation between the emergency lane traffic management regulation in the vehicle-mounted AI knowledge base and the latest emergency lane traffic management regulation in the cloud server used to update that emergency lane traffic management regulation. In other words, if the emergency lane traffic management regulation in the vehicle-mounted AI knowledge base and the latest emergency lane traffic management regulation in the cloud server are completely consistent, the deviation value is 0; otherwise, the deviation value is 1. When a single piece of knowledge in the vehicle-mounted AI knowledge base is a rule restricting the passage of trucks in urban areas during certain hours, the deviation between this single piece of knowledge in the vehicle-mounted AI knowledge base and the latest knowledge in the cloud server used to update this knowledge is the deviation between the urban truck passage time restriction rule in the vehicle-mounted AI knowledge base and the latest urban truck passage time restriction rule in the cloud server used to update this rule. In other words, if the urban truck passage time restriction rule in the vehicle-mounted AI knowledge base and the latest urban truck passage time restriction rule in the cloud server are completely consistent, the deviation value is 0; otherwise, the deviation value is 1. When a single piece of knowledge in the vehicle-mounted AI knowledge base is a traffic light priority rule, the deviation between this single piece of knowledge in the vehicle-mounted AI knowledge base and the latest knowledge used to update this knowledge in the cloud server is the deviation between the traffic light priority rule in the vehicle-mounted AI knowledge base and the latest traffic light priority rule used to update this traffic light priority rule in the cloud server. In other words, if the traffic light priority rule in the vehicle-mounted AI knowledge base and the latest traffic light priority rule used to update this traffic light priority rule in the cloud server are completely consistent, the deviation value is 0; otherwise, the deviation value is 1. When a single piece of knowledge in the vehicle-mounted AI knowledge base is the standard speed limit parameter of an urban arterial road, the deviation between the single piece of knowledge in the vehicle-mounted AI knowledge base and the latest knowledge in the cloud server used to update that knowledge is the deviation between the standard speed limit parameter of the urban arterial road in the vehicle-mounted AI knowledge base and the latest standard speed limit parameter of the urban arterial road in the cloud server used to update that knowledge. That is, first calculate the absolute value of the difference between the standard speed limit parameter of the urban arterial road in the vehicle-mounted AI knowledge base and the latest standard speed limit parameter of the urban arterial road in the cloud server used to update that knowledge. The normalized value of this absolute value is the deviation value. The normalized value of this absolute value can be obtained using existing normalization algorithms. When a single piece of knowledge in the vehicle-mounted AI knowledge base is the speed limit threshold for a high-speed tunnel, the deviation between the single piece of knowledge in the vehicle-mounted AI knowledge base and the latest knowledge in the cloud server used to update that knowledge is the deviation between the speed limit threshold for that high-speed tunnel in the vehicle-mounted AI knowledge base and the latest speed limit threshold for that high-speed tunnel in the cloud server used to update that high-speed tunnel. That is, the absolute value of the difference obtained by subtracting the latest speed limit threshold for that high-speed tunnel in the cloud server from the speed limit threshold for that high-speed tunnel in the vehicle-mounted AI knowledge base is the deviation value. The normalized value of the absolute value can be obtained using existing normalization algorithms. When a single piece of knowledge in the vehicle-mounted AI knowledge base is the traffic parameters of a bridge section, the deviation between the single piece of knowledge in the vehicle-mounted AI knowledge base and the latest knowledge in the cloud server used to update that knowledge is the deviation between the traffic parameters of that bridge section in the vehicle-mounted AI knowledge base and the latest traffic parameters of that bridge section in the cloud server used to update that traffic parameters. That is, first calculate the absolute value of the difference between the traffic parameters of the bridge section in the vehicle-mounted AI knowledge base and the latest traffic parameters of that bridge section in the cloud server used to update that traffic parameters. The normalized value of this absolute value is the deviation value. The normalized value of this absolute value can be obtained using existing normalization algorithms. When a single piece of knowledge in the vehicle-mounted AI knowledge base is the passage parameters for a steep road section, the deviation between the single piece of knowledge in the vehicle-mounted AI knowledge base and the latest knowledge in the cloud server used to update that knowledge is the deviation between the passage parameters for that steep road section in the vehicle-mounted AI knowledge base and the latest passage parameters for that steep road section in the cloud server used to update that passage parameters. That is, the absolute value of the difference obtained by subtracting the latest passage parameters for that steep road section in the cloud server from the passage parameters for that steep road section in the vehicle-mounted AI knowledge base is the deviation value. The normalized value of the absolute value can be obtained using existing normalization algorithms. When a single piece of knowledge in the vehicle AI knowledge base is the vehicle's routine maintenance cycle, the deviation between the single piece of knowledge in the vehicle AI knowledge base and the latest knowledge in the cloud server used to update that knowledge is the deviation between the vehicle's routine maintenance cycle in the vehicle AI knowledge base and the latest vehicle routine maintenance cycle in the cloud server used to update that knowledge. That is, first calculate the absolute value of the difference between the vehicle's routine maintenance cycle in the vehicle AI knowledge base and the latest vehicle routine maintenance cycle in the cloud server used to update that knowledge. The normalized value of this absolute value is the deviation value. The normalized value of this absolute value can be obtained using existing normalization algorithms. When a single piece of knowledge in the vehicle's AI knowledge base sets the operating temperature threshold for the vehicle's motor, the deviation between this single piece of knowledge in the vehicle's AI knowledge base and the latest knowledge in the cloud server used to update this knowledge is the deviation between the operating temperature threshold set for the vehicle's motor in the vehicle's AI knowledge base and the latest operating temperature threshold set for the vehicle's motor in the cloud server. In other words, the absolute value of the difference between the operating temperature threshold set for the vehicle's motor in the vehicle's AI knowledge base and the latest operating temperature threshold set for the vehicle's motor in the cloud server is calculated. The normalized value of this absolute value is the deviation value, and the normalized value can be obtained using existing normalization algorithms. When a single piece of knowledge in the vehicle AI knowledge base is a corresponding judgment rule for a vehicle's fault code, the deviation value between the single piece of knowledge in the vehicle AI knowledge base and the latest knowledge used to update that knowledge in the cloud server is the deviation value between the corresponding judgment rule for the vehicle's fault code in the vehicle AI knowledge base and the latest corresponding judgment rule for the vehicle's fault code in the cloud server. In other words, if the corresponding judgment rule for the vehicle's fault code in the vehicle AI knowledge base and the latest corresponding judgment rule for the vehicle's fault code in the cloud server are completely consistent, the deviation value is 0; otherwise, the deviation value is 1. When a single piece of knowledge in the in-vehicle AI knowledge base is navigation activation guidance interactive text, the deviation value between the single piece of knowledge in the in-vehicle AI knowledge base and the latest knowledge in the cloud server used to update that knowledge is the deviation value between the navigation activation guidance interactive text in the in-vehicle AI knowledge base and the latest navigation activation guidance interactive text in the cloud server used to update that navigation activation guidance interactive text. That is, first, the navigation activation guidance interactive text in the in-vehicle AI knowledge base and the latest navigation activation guidance interactive text in the cloud server used to update that navigation activation guidance interactive text are obtained by using sentence embedding methods in Natural Language Processing (NLP) to obtain two vectors, and then the Euclidean distance between the two vectors is calculated. This Euclidean distance is the deviation value. When a single piece of knowledge in the in-vehicle AI knowledge base is a vehicle fault prompt explanation, the deviation value between the single piece of knowledge in the in-vehicle AI knowledge base and the latest knowledge in the cloud server used to update that knowledge is the deviation value between the vehicle fault prompt explanation in the in-vehicle AI knowledge base and the latest vehicle fault prompt explanation in the cloud server used to update that knowledge. That is, first, the vehicle fault prompt explanation in the in-vehicle AI knowledge base and the latest vehicle fault prompt explanation in the cloud server used to update that knowledge are obtained by using sentence embedding methods in Natural Language Processing (NLP) to obtain two vectors, and then the Euclidean distance between the two vectors is calculated. This Euclidean distance is the deviation value. When a single piece of knowledge in the vehicle AI knowledge base is a vehicle speed abnormality safety warning text, the deviation value between the single piece of knowledge in the vehicle AI knowledge base and the latest knowledge in the cloud server used to update that knowledge is the deviation value between the vehicle speed abnormality safety warning text in the vehicle AI knowledge base and the latest vehicle speed abnormality safety warning text in the cloud server used to update that knowledge. That is, first, the vehicle speed abnormality safety warning text in the vehicle AI knowledge base and the latest vehicle speed abnormality safety warning text in the cloud server are used to update that knowledge by sentence embedding methods in natural language processing (NLP) to obtain two vectors, and then the Euclidean distance between the two vectors is calculated. This Euclidean distance is the deviation value. When a single piece of knowledge in the vehicle-mounted AI knowledge base is the logic for a commuter route congestion warning service, the deviation between the single piece of knowledge in the vehicle-mounted AI knowledge base and the latest knowledge in the cloud server used to update that knowledge is the deviation between the commuter route congestion warning service logic in the vehicle-mounted AI knowledge base and the latest commuter route congestion warning service logic in the cloud server used to update that commuter route congestion warning service logic. In other words, if the commuter route congestion warning service logic in the vehicle-mounted AI knowledge base and the latest commuter route congestion warning service logic in the cloud server are completely consistent, the deviation value is 0; otherwise, the deviation value is 1. When a single piece of knowledge in the vehicle AI knowledge base is a high-speed driving safety reminder trigger rule, the deviation value between the single piece of knowledge in the vehicle AI knowledge base and the latest knowledge in the cloud server used to update that knowledge is the deviation value between the high-speed driving safety reminder trigger rule in the vehicle AI knowledge base and the latest high-speed driving safety reminder trigger rule in the cloud server used to update that high-speed driving safety reminder trigger rule. That is, if the high-speed driving safety reminder trigger rule in the vehicle AI knowledge base and the latest high-speed driving safety reminder trigger rule in the cloud server are completely consistent, the deviation value is 0; otherwise, the deviation value is 1. When a single piece of knowledge in the vehicle AI knowledge base is the intelligent recommendation logic for the destination parking lot, the deviation value between the single piece of knowledge in the vehicle AI knowledge base and the latest knowledge in the cloud server used to update that knowledge is the deviation value between the intelligent recommendation logic for the destination parking lot in the vehicle AI knowledge base and the latest intelligent recommendation logic for the destination parking lot in the cloud server used to update that logic. In other words, if the intelligent recommendation logic for the destination parking lot in the vehicle AI knowledge base and the latest intelligent recommendation logic for the destination parking lot in the cloud server are completely consistent, the deviation value is 0; otherwise, the deviation value is 1. When a single piece of knowledge in the in-vehicle AI knowledge base is a rule for warning travel due to insufficient vehicle range, the deviation between this single piece of knowledge in the in-vehicle AI knowledge base and the latest knowledge in the cloud server used to update this rule is the deviation between the in-vehicle AI knowledge base rule for warning travel due to insufficient vehicle range and the latest knowledge in the cloud server used to update this rule. In other words, if the in-vehicle AI knowledge base rule for warning travel due to insufficient vehicle range and the latest knowledge in the cloud server used to update this rule are completely identical, the deviation is 0; otherwise, the deviation is 1.

[0035] S2. Apply the knowledge obsolescence metric calculation model to calculate the knowledge metadata. Note that the following points should be noted in this step: In a preferred embodiment, the method for calculating knowledge metadata using a knowledge obsolescence metric calculation model specifically includes: The obsolescence of knowledge is obtained by calculating the knowledge metadata using the mathematical expression of the knowledge obsolescence quantification calculation model.

[0036] It should be noted that the obsolescence of knowledge is used to accurately quantify the obsolescence of a single piece of knowledge in the vehicle AI knowledge base. In other words, the obsolescence of knowledge is used to characterize the degree of deviation and obsolescence of knowledge content from the latest knowledge. The higher the obsolescence value, the more serious the obsolescence problem of knowledge.

[0037] In a preferred embodiment, the mathematical expression of the knowledge obsolescence quantification calculation model is: ; In this mathematical expression, Assess the obsolescence of knowledge within the in-vehicle AI knowledge base; For the actual duration of knowledge retention; The effective duration of the knowledge benchmark; For the frequency of knowledge source iteration; Match frequency to knowledge scenarios; This is to address the bias in the effective matching of knowledge.

[0038] It should be noted that the mathematical expression of the knowledge obsolescence metric calculation model is the formula for calculating knowledge obsolescence; the time aging term in the mathematical expression of the knowledge obsolescence metric calculation model... This time decay normalization operator is constructed based on the actual knowledge residency time and the effective duration of the knowledge benchmark. The effective duration of the knowledge benchmark is generated statistically from similar knowledge histories, accurately conforming to the objective update patterns of different types of knowledge without subjective bias. The shorter the actual knowledge residency time, the lower the obsolescence, accurately reflecting the natural aging pattern of knowledge over time. Compared to existing technologies that calculate fixed time decay coefficients, this operator relies entirely on objective time parameters, eliminating subjective bias and matching all duration types of in-vehicle knowledge. It solves the problem of subjective coefficient interference in the accuracy of existing time aging calculations. Furthermore, this operator exhibits a non-linear decay law, conforming to the true aging characteristics of in-vehicle knowledge—stable in the early stages and rapidly failing in the later stages—offering higher accuracy than linear calculation methods. The frequency iteration term in the mathematical expression of the knowledge obsolescence metric calculation model... This is a knowledge iteration activity normalization operator, constructed based on the objective ratio of external iteration frequency to internal usage frequency. A higher knowledge source iteration frequency indicates frequent knowledge updates and rapid content changes; a larger value for this term indicates faster knowledge obsolescence. Conversely, a higher knowledge scenario matching frequency indicates frequent use and real-time verification of the knowledge in automotive scenarios, resulting in higher accuracy; a smaller value for this iteration frequency term suppresses the increase in obsolescence. This operator effectively aligns with the iterative characteristics of knowledge. The matching deviation term in the mathematical expression of the knowledge obsolescence metric calculation model... This is a knowledge content accuracy correction term, directly representing the actual deviation between currently stored knowledge and the latest knowledge. It is a direct quantitative result of knowledge obsolescence. The first two operators predict the trend of knowledge obsolescence from the dimensions of time and iteration. The accuracy of content is used to verify the actual degree of obsolescence. The three factors form a calculation logic for trend prediction and actual verification. This matching deviation item supplements the problem of instantaneous data deviation that cannot be covered by time and frequency, and improves the comprehensiveness of obsolescence measurement.

[0039] Furthermore, the mathematical expression of the knowledge obsolescence quantification calculation model is a combined quantification model, which avoids subjective intervention errors and comprehensively quantifies the obsolescence of knowledge from three dimensions: time aging, iteration characteristics, and content accuracy. This solves the technical problems of single-dimensional judgment and subjective coefficient interference in existing technologies. At the same time, the formula calculation logic is simple and can be matched with the low computing power operation scenario of vehicle edge terminals.

[0040] S3. Calculate the knowledge update lag using a quantitative calculation model. Note that the following points should be noted in this step: In a preferred embodiment, the method for calculating the knowledge update lag using a knowledge update lag quantification model specifically includes: To address the shortcomings of existing technologies in quantifying the degree of knowledge update lag, this invention constructs a mathematical expression for a quantitative calculation model of knowledge update lag. This expression is used to accurately quantify the degree of lag in the update of a single piece of knowledge, that is, to calculate the knowledge update lag degree. This accurately reflects the delay defect in the update of the in-vehicle AI knowledge base. The larger the update lag degree value, the more serious the knowledge update lag problem is, and the higher the risk of the in-vehicle AI system using outdated data.

[0041] In a preferred embodiment, the mathematical expression of the knowledge update lag quantification calculation model is the formula for calculating the knowledge update lag degree. The mathematical expression of the knowledge update lag quantification calculation model is as follows: ; In this mathematical expression, The update lag of a single piece of knowledge within the vehicle's AI knowledge base. For the current moment, This refers to the moment when this piece of knowledge was most recently updated. This refers to the effective duration of the knowledge benchmark, which is the average interval between updates of similar knowledge over history. The average update interval is the mean of the time span between two consecutive updates of this knowledge. This represents the frequency of knowledge source iterations for this piece of knowledge. This represents the maximum frequency of knowledge source iterations for all knowledge in the same category as this knowledge.

[0042] It should be noted that the key term for time lag in the mathematical expression of the knowledge update lag quantification calculation model is... To update the lag time normalization operator, The actual lag time after knowledge update. The effective duration of the knowledge benchmark is used to standardize and quantify the degree of lag by comparing the actual lag duration with the effective duration of the knowledge benchmark. When the data source is updated, the longer the synchronization interval of the in-vehicle AI knowledge base, the larger the numerator value and the higher the lag. For time-sensitive knowledge with a shorter effective duration of the knowledge benchmark, the lag is higher for the same actual lag duration, accurately reflecting the extremely low fault tolerance of short-term knowledge updates. The iterative difference correction term in the mathematical expression of the knowledge update lag quantification calculation model... To correct the knowledge iteration characteristics, an operator is constructed based on the difference between the maximum knowledge source iteration frequency of all knowledge in the same category and the knowledge source iteration frequency of this knowledge. A square root operation is introduced to optimize the numerical distribution. Without manual parameter intervention, this operator is used to determine the maximum knowledge source iteration frequency of all knowledge in the same category whose knowledge source iteration frequency is close to the industry maximum. Approaching zero, the value of the knowledge iteration characteristic correction operator approaches zero, weakening the lag and preventing the overjudgment of small lags in high-frequency iterated knowledge. For low-frequency updated knowledge with iteration frequencies far below the industry maximum, the value of the knowledge iteration characteristic correction operator is increased to amplify the lag, accurately identifying inefficient knowledge that has not been updated for a long time and has severe iteration lag. The +1 operation in the denominator of the iteration difference correction term avoids division by zero errors, ensuring that the calculation logic is absolutely smooth and there are no operational anomalies. The square root operation can optimize numerical convergence, making the lag value distribution more uniform and the hierarchical judgment more accurate. Compared with the linear correction method, it has stronger fault tolerance and wider adaptability.

[0043] Furthermore, the mathematical expression of the knowledge update lag quantification calculation model combines the two dimensions of absolute lag duration and relative iteration characteristics, solving the problems of existing technologies that rely solely on manual judgment of lag and lack quantitative standards. It exhibits extremely high stability, while the formula distinguishes the lag differences of knowledge with different timeliness and iteration characteristics, achieving differentiated and accurate quantification and avoiding the judgment distortion caused by uniform standard judgment. It is well matched to the needs of evaluating the update lag of multi-type and multi-timeliness knowledge in vehicle AI knowledge bases, and has extremely low computing power consumption, matching the real-time computing scenarios of vehicle edge terminals.

[0044] S4. Implement a knowledge obsolescence and update lag classification standard based on knowledge obsolescence and lag. Note that the following points should be noted in this step: In a preferred embodiment, the criteria for determining the level of knowledge obsolescence and lag specifically include: In terms of the obsolescence of knowledge At that time, the obsolescence level of the knowledge was determined to be slightly obsolete, with a small knowledge deviation; In terms of the obsolescence of knowledge At that time, the obsolescence level of the knowledge was determined to be moderate, indicating that the knowledge had significant biases. In terms of the obsolescence of knowledge When this happens, the level of obsolescence of the knowledge is determined to be severely obsolete, meaning the knowledge is essentially invalid. Lag in knowledge updates If the knowledge update lag level is determined to be slightly lagging, the knowledge update delay is within the fault tolerance range and does not affect the operation of the vehicle AI system; Lag in knowledge updates At that time, the knowledge update lag level was determined to be moderate lag, indicating that the knowledge posed a certain operational risk. Lag in knowledge updates If the knowledge is found to be severely outdated, it is considered to pose a security risk.

[0045] In a preferred embodiment, the criteria for determining the level of knowledge obsolescence and lag further include: When the obsolescence level of knowledge is determined to be slightly obsolete or the update lag level of knowledge is determined to be slightly lagging, periodic inspection and updates are performed on the knowledge. When the knowledge is determined to be moderately outdated or moderately lagging in terms of update lag, an incremental partial update is performed on the knowledge. When the obsolescence level of knowledge is determined to be severely obsolete or the update lag level is determined to be seriously lagging, the knowledge will be either fully replaced and updated or directly cleaned up and deleted.

[0046] It should be noted that the methods for periodically checking and updating this knowledge may specifically include: With a fixed inspection cycle of one month, at the end of each fixed inspection cycle, the data processing module of the vehicle AI system compares the latest knowledge of the same type stored in the cloud server with the current knowledge. If the comparison finds that the latest knowledge is different from the current knowledge, only the content with the difference is updated in the current knowledge to make it consistent with the latest knowledge; if the comparison finds that the latest authoritative knowledge is no different from the current knowledge, the current knowledge is retained without any operation.

[0047] Furthermore, methods for incrementally updating this knowledge locally may specifically include: When the obsolescence level of knowledge is determined to be moderately obsolete or the update lag level is determined to be moderately lagging, the data processing module of the vehicle AI system compares the latest knowledge of the same type used to update this type of knowledge with the knowledge stored in the cloud server in real time. If the comparison finds that the latest knowledge is different from the knowledge, only the content with the difference is updated in the knowledge to make it consistent with the latest knowledge; if the comparison finds that the latest knowledge is the same as the knowledge, the knowledge is retained without any operation.

[0048] Furthermore, methods for completely replacing or directly deleting this knowledge may include: When determining that the obsolescence level of knowledge is severely obsolete or the update lag level of knowledge is severely lagging, if the knowledge scenario matching frequency of knowledge is within the effective duration of the two consecutive knowledge benchmarks prior to this moment... =0, the data processing module of the vehicle AI system will directly clean and delete the knowledge in real time; If, within the effective duration of two consecutive knowledge benchmarks prior to this moment, the knowledge scenario matching frequency of this knowledge... 0. The data processing module of the vehicle AI system will replace and update the knowledge in real time with the latest knowledge of the same type stored in the cloud server.

[0049] Furthermore, the obsolescence thresholds of 0.3 and 0.8 are set based on the low fault tolerance and safety critical inflection point, matching the knowledge aging accumulation characteristics and vehicle computing power constraints; the lag thresholds of 0.2 and 0.6 are divided based on the normal synchronization delay and the risk mutation critical point, taking into account the system fault tolerance capability and vehicle operation safety. The hierarchical gradient is scientific, has strong engineering matching, and is not subjectively arbitrary.

[0050] This invention addresses the technical shortcomings of existing in-vehicle AI knowledge bases, such as outdated data, delayed updates, lack of quantitative evaluation standards, low update efficiency, and strong subjectivity of human intervention. It proposes a data processing method for in-vehicle AI knowledge bases. This invention abandons the traditional manual experience-based judgment and fixed-cycle batch update mode, constructs a quantitative calculation formula, and achieves accurate quantification of the outdatedness and update lag of individual knowledge items in the in-vehicle knowledge base, accurately distinguishing the timeliness and expiration levels of different knowledge. It establishes a real-time incremental update mechanism with multiple trigger conditions to replace the traditional batch update mode, matching the computing power characteristics of in-vehicle edge terminals. It enables dynamic cleaning, iteration, and matching of knowledge base data, solving the problem of outdated and delayed data, and improving the decision-making accuracy, interaction stability, and operational security of in-vehicle AI systems.

[0051] It should be recognized that embodiments of the present invention may be implemented or carried out by computer hardware, a combination of hardware and software, or by computer instructions stored in a non-transitory computer-readable storage medium.

[0052] The method can be implemented using standard programming techniques, including a non-transitory computer-readable storage medium configured with a computer program in the computer program, wherein the storage medium is configured such that the computer operates in a specific and predefined manner.

[0053] Each program can be implemented in a high-level procedural or object-oriented programming language to communicate with the computer system; however, if required, the program can be implemented in assembly or machine language.

[0054] In any case, the language can be either compiled or interpreted.

[0055] Furthermore, for this purpose, the program can run on programmed application-specific integrated circuits.

[0056] The processes described herein (or variations and / or combinations thereof) can be executed under the control of one or more computer systems configured with executable instructions, and can be implemented by hardware or a combination thereof as code (e.g., executable instructions, one or more computer programs, or one or more applications) that commonly executes on one or more processors. The computer program includes a plurality of instructions executable by one or more processors.

[0057] Furthermore, the method can be implemented in any suitable computing platform, including but not limited to personal computers, minicomputers, mainframes, workstations, networked or distributed computing environments, standalone or integrated computer platforms, or in communication with charged particle tools or other imaging devices.

[0058] Various aspects of the present invention can be implemented in machine-readable code stored on a non-transitory storage medium or device, whether portable or integrated into a computing platform, such as a hard disk, optical read and / or write storage medium, RAM, ROM, etc., such that it can be read by a programmable computer, and when the storage medium or device is read by the computer, it can be used to configure and operate the computer to perform the processes described herein.

[0059] Furthermore, machine-readable code, or parts thereof, can be transmitted via wired or wireless networks.

[0060] When such media includes instructions or programs that combine with a microprocessor or other data processor to implement the steps described above, the invention described herein includes these and other different types of non-transitory computer-readable storage media.

[0061] 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 data processing method for in-vehicle AI knowledge bases, characterized in that, include: Standardized collection of knowledge metadata for in-vehicle AI knowledge base; The knowledge metadata is calculated using a knowledge obsolescence quantification calculation model; The knowledge update lag metric calculation model is applied to calculate the knowledge update lag. The criteria for determining the obsolescence and lag of knowledge are based on the obsolescence and lag of its updates. The mathematical expression for the quantitative calculation model of knowledge obsolescence is: ； In this mathematical expression, As to the obsolescence of knowledge; For the actual duration of knowledge retention; The effective duration of the knowledge benchmark; For the frequency of knowledge source iteration; Match frequency to knowledge scenarios; To address the bias in effective knowledge matching; The mathematical expression for the quantitative calculation model of knowledge update lag is: ； In this mathematical expression, The update lag of a single piece of knowledge. For the current moment, This refers to the moment when this piece of knowledge was most recently updated. For the duration of the knowledge benchmark in effect, For the frequency of knowledge source iteration, This represents the maximum frequency of knowledge source iterations for all knowledge of the same category as this piece of knowledge. The knowledge baseline effective duration is the set baseline effective duration for a single piece of knowledge within the in-vehicle AI knowledge base. This baseline effective duration is the average update interval of similar knowledge in history, which is the average time span between two adjacent updates of this knowledge. The actual knowledge residence time is the duration of a single piece of knowledge in the in-vehicle AI knowledge base from the end of its last update to the current moment. The knowledge scenario matching frequency is the cumulative number of times a single piece of knowledge in the in-vehicle AI knowledge base has been called, matched, and used for decision-making by the in-vehicle AI system. The knowledge source iteration frequency is the number of times a single piece of knowledge in the in-vehicle AI knowledge base has been updated by knowledge in the cloud server up to the current moment. The knowledge effective matching deviation is the deviation between a single piece of knowledge in the vehicle-mounted AI knowledge base and the latest knowledge used to update that knowledge in the cloud server.

2. The data processing method for vehicle-mounted AI knowledge base according to claim 1, characterized in that, The standardized collection method for knowledge metadata of in-vehicle AI knowledge base includes: For each piece of knowledge in the vehicle AI knowledge base, five types of objective metadata are collected. These five types of objective metadata are the knowledge metadata, which include the effective duration of the knowledge benchmark, the actual residence time of the knowledge, the frequency of knowledge scenario matching, the frequency of knowledge source iteration, and the deviation of effective knowledge matching.

3. The data processing method for vehicle-mounted AI knowledge base according to claim 2, characterized in that, The knowledge in the in-vehicle AI knowledge base includes traffic regulations, road network and traffic conditions, vehicle operation and maintenance, in-vehicle interaction, and scenario service knowledge.

4. The data processing method for vehicle-mounted AI knowledge base according to claim 3, characterized in that, Methods for calculating knowledge metadata using a knowledge obsolescence metric computational model include: The obsolescence of knowledge is obtained by calculating the knowledge metadata using the mathematical expression of the knowledge obsolescence quantification calculation model.

5. The data processing method for vehicle-mounted AI knowledge base according to claim 4, characterized in that, Methods for calculating the knowledge update lag using a quantitative computational model include: The mathematical expression for constructing a quantitative calculation model of knowledge update lag is used to quantify the lag degree of a single knowledge update, that is, to calculate the knowledge update lag degree.

6. The data processing method for vehicle-mounted AI knowledge base according to claim 5, characterized in that, The specific criteria for determining the level of knowledge obsolescence and lag include: In terms of the obsolescence of knowledge At that time, the obsolescence level of the knowledge was determined to be slightly obsolete; In terms of the obsolescence of knowledge At that time, the obsolescence level of the knowledge was determined to be moderately obsolete; In terms of the obsolescence of knowledge When this happens, the level of obsolescence of the knowledge is determined to be severely obsolete; Lag in knowledge updates In this case, the lag level of the knowledge update is determined to be slightly lagging; Lag in knowledge updates When this happens, the knowledge update lag level is determined to be moderate lag; Lag in knowledge updates When this happens, the knowledge update lag level is determined to be severely lagging.

7. The data processing method for vehicle-mounted AI knowledge base according to claim 6, characterized in that, The criteria for classifying knowledge as outdated and lagging also include: When the obsolescence level of knowledge is determined to be slightly obsolete or the update lag level of knowledge is determined to be slightly lagging, periodic inspection and updates are performed on the knowledge. When the knowledge is determined to be moderately outdated or moderately lagging in terms of update lag, an incremental partial update is performed on the knowledge. When the obsolescence level of knowledge is determined to be severely obsolete or the update lag level is determined to be seriously lagging, the knowledge will be either fully replaced and updated or directly cleaned up and deleted.

Images