Time series data storage method and device, equipment, medium and program product

Abstract

The invention provides a time sequence data storage method and device, equipment, a medium and a program product, and is applied to a vehicle end, and the method comprises the steps: obtaining time sequence data generated by vehicle operation, and determining a to-be-stored annular queue from preset multi-stage annular queues according to the time sequence data generated by vehicle operation; determining fusible time sequence data and non-fusible time sequence data according to the time sequence data generated by vehicle operation; wherein the fusible time series data represents data corresponding to the same entity existing in the to-be-stored annular queue in the time series data generated by vehicle operation; for the non-fusible time series data, judging whether the to-be-stored annular queue can store the non-fusible time series data or not; if the time sequence data to be stored in the annular queue cannot be stored and cannot be fused, performing data distillation on the time sequence data to be stored in the annular queue; storing the time sequence data which cannot be fused into a to-be-stored annular queue; and if the non-fusible time series data can be stored, storing the non-fusible time series data to the to-be-stored annular queue.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence, and in particular to a method, apparatus, device, medium, and program product for storing time series data. Background Art

[0002] The trend toward intelligent and connected vehicles generates massive amounts of time-series data during operation. This data is typically stored in fixed storage spaces. Currently, storing this massive amount of time-series data is expensive, making it difficult to retain historical data from numerous sensors. Consequently, this approach results in inefficient storage of the time-series data generated by vehicle operation. Summary of the Invention

[0003] Embodiments of the present application provide a method, apparatus, device, medium, and program product for storing time series data, so as to improve the storage effect of time series data generated by vehicle operation.

[0004] In a first aspect, an embodiment of the present application provides a method for storing time series data, which is applied to a vehicle side, and includes: obtaining time series data generated by vehicle operation, and determining a to-be-stored circular queue from a preset multi-level circular queue based on the time series data generated by vehicle operation; wherein, the higher the level of the circular queue, the larger the corresponding time period; the preset multi-level circular queue is used to store the time series data generated by vehicle operation according to the corresponding time period; determining fusible time series data and unfusible time series data based on the time series data generated by vehicle operation; wherein, the fusible time series data represents the data corresponding to the same entity in the time series data generated by vehicle operation as that in the circular queue to be stored; for unfusible time series data, determining whether the to-be-stored circular queue can store the unfusible time series data; if the unfusible time series data cannot be stored, performing data distillation on the time series data to be stored in the circular queue; storing the unfusible time series data in the to-be-stored circular queue; if the unfusible time series data can be stored, storing the unfusible time series data in the circular queue to be stored.

[0005] In some examples, based on a causal reasoning algorithm, the time series data in a preset circular queue is analyzed to obtain the causal relationship between different events; wherein, different events are caused by state changes of entities in the time series data stored in the preset circular queue; the causal relationship between different events is integrated into a time series knowledge graph; wherein, the time series knowledge graph is constructed based on historical time series data.

[0006] In some examples, after determining the fusible time series data and the non-fusible time series data based on the time series data generated by vehicle operation, it also includes: fusing the fusible time series data with the corresponding time series data to be stored in the circular queue, and storing them in the circular queue to be stored.

[0007] In some examples, data distillation is performed on the time series data to be stored in the circular queue, specifically including: determining whether the time period corresponding to the time series data to be stored in the circular queue reaches the time period corresponding to the previous-level circular queue; if it reaches the time period corresponding to the previous-level circular queue, data distillation is performed on the time series data to be stored in the circular queue to obtain new time series data and delete the time series data to be stored in the circular queue; and the new time series data is stored in the circular queue to be stored and the previous-level circular queue respectively; if it does not reach the time period corresponding to the previous-level circular queue, data distillation is performed on the time series data to be stored in the circular queue to obtain new time series data and delete the time series data in the circular queue; and the new time series data is stored in the circular queue to be stored.

[0008] In some examples, entities and relationships in new time series data are obtained; a first feature vector is obtained based on the entities and relationships in the new time series data; and the feature vector in the time series knowledge graph is updated to the first feature vector.

[0009] In some examples, storing new time series data before storing it in the circular queue and the previous level circular queue also includes: determining whether the previous level circular queue can store the new time series data; if the new time series data cannot be stored, performing data distillation on the time series data in the previous level circular queue, and then storing the new time series data in the previous level circular queue.

[0010] In some examples, entities and relationships in vehicle time series data are obtained, and a second feature vector is obtained based on the entities and relationships in the vehicle time series data; a historical feature vector corresponding to the second feature vector is obtained from the time series knowledge graph; the second feature vector and the historical feature vector are input into the diagnostic analysis model for update training to obtain an updated diagnostic analysis model.

[0011] In some examples, after obtaining the historical feature vector corresponding to the second feature vector from the temporal knowledge graph, it also includes: checking whether the historical feature vector corresponding to the second feature vector is missing; if there is a missing, generating supplementary data based on other historical feature vectors in the temporal knowledge graph; and filling the missing part of the historical feature vector corresponding to the second feature vector based on the supplementary data.

[0012] In some examples, the second feature vector is processed using a preset multi-level circular queue to extract data features; analysis is performed based on the data features combined with the time series data stored in the preset multi-level circular queue; if a new relationship is obtained, the relationship between entities in the time series knowledge graph is updated in chronological order based on the new relationship.

[0013] In some examples, when a vehicle failure is detected, vehicle time series data is obtained based on the vehicle failure data; entities and relationships in the vehicle time series data are obtained, and a third feature vector is obtained based on the entities and relationships in the vehicle time series data; the third feature vector is input into an updated diagnostic analysis model to obtain a diagnostic result.

[0014] In some examples, the diagnostic results are displayed in a visual interface so that users can repair the fault according to the diagnostic results; if the vehicle fault is not repaired successfully, the steps of obtaining entities and relationships in the vehicle time series data are executed again to obtain a new diagnostic result; if the vehicle fault is repaired successfully, the time series knowledge graph is optimized according to the diagnostic results; wherein, the time series knowledge graph represents the relationship between the historical data, diagnostic process and solution of the vehicle under different time information.

[0015] In a second aspect, an embodiment of the present application provides a storage device for time series data, comprising: an acquisition module for acquiring time series data generated by vehicle operation, and determining a to-be-stored circular queue from a preset multi-level circular queue based on the time series data generated by vehicle operation; wherein, a higher-level circular queue corresponds to a longer time period; the preset multi-level circular queue is used to store the time series data generated by vehicle operation according to the corresponding time period; a processing module for determining fused time series data and unfused time series data based on the time series data generated by vehicle operation; wherein, the fused time series data represents data corresponding to the same entity in the time series data generated by vehicle operation as that in the to-be-stored circular queue; a judgment module for judging whether the to-be-stored circular queue can store unfused time series data for unfused time series data; the processing module is further used to, if the unfused time series data cannot be stored, perform data distillation on the time series data to be stored in the circular queue; and store the unfused time series data in the to-be-stored circular queue; the processing module is further used to, if the unfused time series data can be stored, store the unfused time series data in the to-be-stored circular queue.

[0016] In a third aspect, an embodiment of the present application provides an electronic device, comprising: a memory, a processor;

[0017] Memory stores computer-executable instructions;

[0018] The processor executes the computer-executable instructions stored in the memory, so that the processor executes the above first aspect and / or various possible implementations of the first aspect.

[0019] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, in which computer-executable instructions are stored. When the computer-executable instructions are executed by a processor, they are used to implement the first aspect above and / or various possible implementation methods of the first aspect.

[0020] In a fifth aspect, an embodiment of the present application provides a computer program product, including a computer program, which, when executed by a processor, implements the above first aspect and / or various possible implementation methods of the first aspect.

[0021] The embodiments of the present application provide a method, device, equipment, medium and program product for storing time series data, which are applied to the vehicle side and include: obtaining time series data generated by vehicle operation, determining to be stored in a circular queue from a preset multi-level circular queue based on the time series data generated by vehicle operation; determining fused time series data and unfused time series data based on the time series data generated by vehicle operation; wherein the fused time series data represents the data corresponding to the same entity in the time series data generated by vehicle operation that exists in the circular queue to be stored; for unfused time series data, determining whether the circular queue to be stored can store unfused time series data; if the unfused time series data cannot be stored, performing data distillation on the time series data to be stored in the circular queue; storing the unfused time series data in the circular queue to be stored; if the unfused time series data can be stored, storing the unfused time series data in the circular queue to be stored. This solution uses a multi-level circular queue to store the time series data generated by vehicle operation according to corresponding different periods, and at the same time performs data distillation on the circular queue to improve the storage effect of the time series data. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0023] Figure 1 A flowchart of a method for storing time series data provided in this application;

[0024] Figure 2 A schematic diagram of a multi-stage ring queue provided by this application;

[0025] Figure 3 A schematic diagram of entities and relationships in a temporal knowledge graph provided for this application;

[0026] Figure 4 A schematic diagram of the transformation of entities and relationships in a temporal knowledge graph provided by this application;

[0027] Figure 5 A flowchart of a diagnostic method for a smart car provided in this application;

[0028] Figure 6 A schematic diagram of the structure of a time series data storage device provided by this application;

[0029] Figure 7This is a schematic diagram of the structure of an electronic device provided in this application.

[0030] The above drawings illustrate specific embodiments of the present application, which will be described in more detail below. These drawings and the textual description are not intended to limit the scope of the present application in any way, but rather to illustrate the concepts of the present application to those skilled in the art by reference to specific embodiments. DETAILED DESCRIPTION

[0031] Exemplary embodiments will be described in detail herein, with examples illustrated in the accompanying drawings. In the following description, when referring to the drawings, identical numerals in different figures represent identical or similar elements, unless otherwise indicated. The embodiments described in the following exemplary embodiments are not intended to represent all embodiments consistent with the present application. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the present application, as detailed in the appended claims.

[0032] The trend toward intelligent and connected vehicles generates massive amounts of time-series data during operation. This data is typically stored in fixed storage spaces. Currently, storing this massive amount of time-series data is expensive, making it difficult to retain historical data from numerous sensors. Consequently, this approach results in inefficient storage of the time-series data generated by vehicle operation.

[0033] The present application provides a method, device, equipment, medium and program product for storing time series data, which are applied to the vehicle side, including: obtaining time series data generated by vehicle operation, determining the time series data to be stored in the circular queue from a preset multi-level circular queue based on the time series data generated by vehicle operation; determining the fused time series data and the non-fused time series data based on the time series data generated by vehicle operation; wherein the fused time series data represents the data corresponding to the same entity in the time series data generated by vehicle operation that exists in the circular queue to be stored; for the non-fused time series data, determining whether the circular queue to be stored can store the non-fused time series data; if the non-fused time series data cannot be stored, performing data distillation on the time series data to be stored in the circular queue; storing the non-fused time series data in the circular queue to be stored; if the non-fused time series data can be stored, storing the non-fused time series data in the circular queue to be stored. This solution uses a multi-level circular queue to store the time series data generated by vehicle operation according to corresponding different periods, and at the same time performs data distillation on the circular queue to improve the storage effect of the time series data.

[0034] The following specific embodiments describe in detail the technical solution of the present application and how the technical solution of the present application solves the above-mentioned technical problems. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present application will be described below in conjunction with the accompanying drawings.

[0035] Example 1

[0036] Figure 1 A flow chart of a method for storing time series data provided in this application is shown as follows: Figure 1 Shown, including:

[0037] S101, obtaining time series data generated by vehicle operation, and determining a ring queue to be stored from a preset multi-level ring queue based on the time series data generated by the vehicle operation; wherein a higher-level ring queue corresponds to a longer time period; the preset multi-level ring queue is used to store the time series data generated by vehicle operation according to the corresponding time period;

[0038] S102, determining fusible time series data and non-fusible time series data based on the time series data generated by vehicle operation; wherein the fusible time series data represents data corresponding to the same entity in the time series data generated by vehicle operation to be stored in the circular queue;

[0039] S103, for the unfusible time series data, determining whether the ring queue to be stored can store the unfusible time series data;

[0040] S104: If the unfusible time series data cannot be stored, then after performing data distillation on the time series data to be stored in the circular queue, the unfusible time series data is stored in the circular queue;

[0041] S105: If the unfusible time series data can be stored, the unfusible time series data is stored in a ring queue to be stored.

[0042] With the increasing adoption of high-performance computers in automotive systems, these systems are becoming increasingly dependent on software. Because software architectures require rapid and flexible deployment of functions to vehicles, Service-Oriented Vehicle Diagnostics (SOVD) is being introduced to diagnose new systems and traditional sensors and actuators. SOVD is a diagnostic model based on advanced information technology and standardized architecture. Limited vehicle storage space makes storing time series data generated during operation expensive, making it difficult to store large amounts of historical and new time series data. Therefore, this solution introduces a multi-level ring queue to store large amounts of time series data.

[0043] Among them, a multi-level circular queue is a data structure composed of multiple circular queues, which are organized according to a certain hierarchical relationship. Each circular queue stores data of different time periods with a fixed capacity; wherein, the time period can be divided according to preset requirements, such as years, months, weeks, days, hours, minutes, and seconds; in addition, the larger the time period, the longer the time interval; for example, the first-level circular queue stores the comprehensive status data of key systems in units of hours, and the second-level circular queue stores detailed data of specific components in units of minutes.

[0044] Figure 2 A schematic diagram of a multi-stage ring queue provided by this application, such as Figure 2 As shown in the figure, the thick solid line is the original "edge" between the nodes in the time series knowledge graph; even in different ring queue levels, the connection between the levels will not be disconnected, and the connection represented by the thick solid line expresses the relationship between the nodes.

[0045] In this embodiment, for the unfused time series data, it is determined whether the circular queue to be stored can store the unfused time series data; specifically, for example, the storage capacity of the circular queue to be stored is 5. If the current circular queue to be stored has stored time series data to the storage capacity of 5, it means that the circular queue to be stored cannot store the unfused time series data; if the current circular queue to be stored has stored time series data to the storage capacity of 4, it means that the circular queue to be stored can store the unfused time series data, and after storage, the amount of data stored in the circular queue to be stored reaches 5.

[0046] In practical applications, time series data generated by vehicle operation is obtained, and based on this time series data, a circular queue to be stored is determined from a preset multi-level circular queue. For example, the multi-level circular queue is divided into levels based on time periods of minutes and hours, resulting in a first-level circular queue with an hourly time period and a second-level circular queue with a minutely time period. For example, if engine speed data per minute is obtained, the circular queue to be stored for this data is determined to be a second-level circular queue with a minutely time period based on the preset multi-level circular queue. A determination is made as to whether time series data for engine speed exists in the second-level circular queue. If not, a determination is made as to whether the second-level circular queue can store the engine speed data. If not, the data is first distilled before storage. If storage is possible, the data is directly stored.

[0047] By utilizing multi-level circular queues to store time-series data generated by vehicle operation, data can be stored in different time periods based on real-time requirements, enabling rapid data access and processing, and improving data query efficiency. For example, during vehicle operation, critical real-time data such as vehicle speed and engine speed require timely processing and analysis. This data can be stored in a first-level circular queue with a smaller time granularity for rapid access and processing. Less urgent data, such as comprehensive mileage data and average fuel consumption over time, can be stored in higher-level circular queues with a larger time granularity. This ensures the real-time nature of critical data while effectively managing large amounts of non-real-time data.

[0048] In some examples, this also includes:

[0049] Based on the causal reasoning algorithm, the time series data in the preset circular queue is analyzed to obtain the causal relationship between different events; wherein different events are caused by the state changes of entities in the time series data stored in the preset circular queue;

[0050] Integrate the causal relationship between different events into the time series knowledge graph; among them, the time series knowledge graph is constructed based on historical time series data.

[0051] Causal inference algorithms are a class of methods and techniques used to discover causal relationships in data. By analyzing data using causal inference algorithms, we can determine the causal relationship between variables—that is, whether a change in one variable causes a change in another, as well as the strength and direction of this causal relationship. Specifically, causal inference algorithms (such as Do-Calculus and structural causal models) are used to analyze data in a circular queue to identify causal relationships between different events. For example, analyzing time series data from engines and temperature sensors in a multi-stage circular queue reveals that higher engine speeds generate more heat, and thus, generally, higher temperature sensor readings. Therefore, the increased temperature sensor reading is caused by the increased engine speed.

[0052] By integrating causal relationships into the temporal knowledge graph, the semantic knowledge of the temporal knowledge graph is enriched, thereby improving the accuracy of automobile diagnosis.

[0053] Among them, the time series knowledge graph is constructed based on historical time series data. In practical applications, the entities, relationships and time information in the vehicle historical data are obtained; the types of entities and relationships in the vehicle historical data are determined; quadruple groups are constructed based on the types and time information of entities and relationships in the vehicle historical data; the initial time series knowledge graph is established based on the quadruple groups, and the diagnostic process and solution are integrated into the initial time series knowledge graph to obtain the time series knowledge graph.

[0054] Specifically, entity types refer to the classification of entities into different categories based on their characteristics and attributes. For example, "car," "truck," and "bus" all belong to the entity type "vehicle"; "oil pump," "oil filter," and "oil channel" all belong to the entity type "engine lubrication system component." Relationship types categorize and define relationships between entities. For example, "installed on" is a relationship type that represents the relationship between "engine" and "car," i.e., "engine installed on car." "Connected" is another relationship type that can describe the relationship between "drive shaft" and "transmission," i.e., "drive shaft connected to transmission." Different relationship types accurately express connections of varying nature between entities. A quaternary entity consists of a "head entity, relationship, tail entity, time" structure. For example, (temperature sensor, monitoring, engine, time) indicates that the temperature sensor was monitoring the engine at that time; (vehicle, installed on, engine, time) indicates that the engine was installed on the vehicle at that time.

[0055] Integrate diagnostic processes and solutions into the initial temporal knowledge graph. Specifically, when obtaining diagnostic processes and solutions from repair manuals, analyze how their effectiveness changes over time, based on vehicle upgrades and technological evolution. For example, repair methods that worked for older models may no longer be applicable or require adjustments for newer models. By incorporating this time-related knowledge into the temporal knowledge graph, the reasoning process can be guided to more accurately select appropriate diagnostic and solution options.

[0056] After establishing the initial temporal knowledge graph based on the quadruple, the temporal knowledge graph needs to be initialized. Specifically, an initial feature vector is obtained based on the entities and relationships in the vehicle's historical data. The initial value of the temporal knowledge graph's feature vector is set to this initial feature vector. In practical applications, a pre-trained machine algorithm model is used to analyze the entities and relationships in the vehicle's historical data to obtain the initial feature vector. For example, the engine temperature data, transmission operating status data, and coolant pump operating parameters monitored by a temperature sensor over a period of time are the engine and temperature sensor entities, and the monitoring relationship is used. The pre-trained machine algorithm model is used to analyze this data, along with the entities and relationships, to obtain a vector that reflects the entity and relationship information as the initial feature vector. The initial value of the temporal knowledge graph's feature vector is set to this initial feature vector. The initial value of the engine node's feature vector is set to the portion of the initial feature vector related to the engine, and the initial value of the temperature sensor node's feature vector is set to the portion of the initial feature vector related to the temperature sensor.

[0057] By constructing a time-series knowledge graph, we can integrate and represent vehicle entities and relationships, and clearly record their dynamic changes over time. Furthermore, by leveraging the rich information in historical vehicle data to obtain initial feature vectors, we initialize the feature vectors of the time-series knowledge graph, imbuing it with semantic and structural information. This provides a more rational and accurate foundation for subsequent analysis and reasoning tasks based on the time-series knowledge graph, such as predicting engine failures and optimizing vehicle operating parameters.

[0058] In some examples, after determining the fusionable time series data and the non-fusionable time series data based on the time series data generated by vehicle operation, the method further includes:

[0059] The fused time series data is merged with the corresponding time series data in the ring queue to be stored, and the data is stored in the ring queue to be stored.

[0060] In combination with the above example, specifically, the data of engine speed is fused with the data about engine speed stored in the circular queue, and the fused time series data is stored in the circular queue; for example, the time series data of engine speed collected with a time period of minutes is obtained, and it can be known that the circular queue to be stored is a secondary circular queue; the time series data of engine speed stored in the secondary circular queue is fused with the collected time series data of engine speed to obtain the fused time series data of engine speed, and store it in the secondary circular queue.

[0061] In some examples, data distillation is performed on time series data to be stored in a circular queue, specifically including:

[0062] Determine whether the time period corresponding to the time series data to be stored in the ring queue reaches the time period corresponding to the previous level ring queue;

[0063] If the time period corresponding to the previous level circular queue is reached, the time series data to be stored in the circular queue is distilled to obtain new time series data and the time series data to be stored in the circular queue is deleted; and the new time series data is stored in the circular queue to be stored and the previous level circular queue respectively;

[0064] If the time period corresponding to the previous level ring queue is not reached, the time series data to be stored in the ring queue is subjected to data distillation to obtain new time series data and delete the time series data to be stored in the ring queue; and the new time series data is stored in the ring queue to be stored.

[0065] Among them, data distillation is a data processing technology that aims to extract valuable information from large amounts of raw data while reducing the volume and complexity of the data for easy storage, transmission and analysis.

[0066] In combination with the above example, based on the acquired time series data generated by vehicle operation, it is determined that the circular queue to be stored is a secondary circular queue with a time period of minutes, and the upper level circular queue of the secondary circular queue is a primary circular queue with a time period of hours. In this example, it is determined whether the time period of the time series data stored in the secondary circular queue reaches the time period corresponding to the primary circular queue; for example, the amount of data currently stored in the secondary circular queue is 5, which has reached the maximum storage capacity, but the stored time series data is 50 minutes of time series data. Therefore, the secondary circular queue only stores time series data with a time period of 50 minutes, which does not reach the time period corresponding to the primary circular queue.

[0067] In actual applications, if the time period corresponding to the time series data stored in the secondary circular queue reaches one hour, the new time series data obtained after data distillation of the secondary circular queue is stored in the primary and secondary circular queues respectively. If the time period corresponding to the time series data stored in the secondary circular queue does not reach one hour, only the new time series data is stored in the secondary circular queue. At the same time, after data distillation, the original time series data stored in the secondary circular queue is deleted.

[0068] By introducing data distillation, the storage and deployment costs on vehicles can be reduced. At the same time, data distillation is used to compress the data in the circular queue, retain key data useful for causal reasoning to obtain new time series data, and delete historical time series data. This reduces storage costs and can retain key data of historical data.

[0069] In some examples, this also includes:

[0070] Obtain entities and relationships in the new time series data; obtain the first eigenvector based on the entities and relationships in the new time series data;

[0071] Update the feature vector in the time series knowledge graph to the first feature vector.

[0072] As the circular queue data is updated, the relationships represented by the edges will be adjusted according to the latest data. Combined with the previous example, the new time series data is processed, the entities and relationships in the new time series data are analyzed, and the new relationships between the entities are obtained. The entities and new relationships are embedded and transformed to obtain the first eigenvector, and the time series knowledge graph is updated to ensure that the relationships in the time series knowledge graph are consistent with the actual relationships.

[0073] Specifically, entities include vehicle components (such as engines, transmissions, batteries, and their status information at different times), sensors, error codes, and maintenance operations. Relationships describe the interactions between entities over time; relationships include but are not limited to monitoring relationships, ownership relationships, triggering relationships, repair relationships, impact relationships, and dependency relationships. Monitoring relationships refer to sensors monitoring changes in the status of vehicle components, ownership relationships refer to software modules belonging to specific vehicle systems and their version evolution, triggering relationships refer to the development of error codes from specific problem sources and taking the root causes into account, repair relationships refer to solutions used to repair specific problem sources and tracking of repair effects, impact relationships refer to how the vehicle's operating status affects the operation of components and sensors, and dependency relationships refer to the dynamic changes in a software module's dependence on certain hardware components or other software modules during its life cycle. Figure 3 A schematic diagram of entities and relationships in a temporal knowledge graph provided for this application, such as Figure 3 As shown in Figure 1, at a certain moment, entities in the temporal knowledge graph interact with each other; the nodes in the graph represent entities and the edges represent relationships.

[0074] Relationships describe the interactions between entities over time, such as "component - monitored by a sensor over a certain period of time", "error code - indicates component failure at a specific time", and "repair action - repairing a component at a certain point in time and tracking the subsequent repair results".

[0075] In some examples, storing new time series data before storing it in the ring queue and the previous ring queue also includes:

[0076] Determine whether the previous level ring queue can store new time series data;

[0077] If the new time series data cannot be stored, the time series data in the previous level ring queue is distilled and the new time series data is stored in the previous level ring queue.

[0078] In combination with the above example, before storing new time series data in the first-level circular queue, determine whether the amount of time series data stored in the first-level circular queue reaches the storage capacity. For example, the storage capacity of the first-level circular queue is 8. If the amount of data stored in the first-level circular queue is 8 at this time, the first-level circular queue will be distilled before storing the new time series data.

[0079] In practical applications, the levels of a multi-level ring queue are configured based on specific needs. For example, a multi-level ring queue can be constructed with ring queues for hourly, daily, weekly, and monthly time periods. When data distillation is performed on any level of a ring queue and needs to be stored in the previous level of the ring queue, the previous level of the ring queue must first determine whether it can store the time series data obtained after the data distillation.

[0080] In some examples, this also includes:

[0081] Obtaining entities and relationships in the vehicle time series data, and obtaining a second eigenvector based on the entities and relationships in the vehicle time series data;

[0082] Obtain the historical feature vector corresponding to the second feature vector from the time series knowledge graph;

[0083] The second eigenvector and the historical eigenvector are input into the diagnosis and analysis model for update training to obtain an updated diagnosis and analysis model.

[0084] The trend toward intelligent and connected vehicles generates massive amounts of time-series data during vehicle operation. Traditional automotive diagnostic methods struggle to effectively process this complex time-series information and are unable to fully exploit the underlying patterns within the data. Time-series knowledge graphs, as powerful knowledge representation and reasoning tools, can integrate diverse information within the automotive field. Specifically, time tags are added to entities and relationships within the time-series knowledge graph to represent the timeliness and dynamic nature of knowledge. Diagnostic analysis models, such as the Recurrent Event Network (RE-NET) model, are examples.

[0085] In this example, the diagnostic analysis model deployed on the vehicle side is obtained by pre-training the machine algorithm model on the cloud side using historical time series data through knowledge distillation. The diagnostic analysis model is deployed on the vehicle side, and the time series data generated by the vehicle operation is used to update the diagnostic analysis model, thereby improving the diagnostic analysis model's ability to diagnose the vehicle.

[0086] In practical applications, entities are not simply recorded for their basic operational states, such as starting and driving. Rather, the different states of the entities over different time periods, the numerical changes in performance indicators at different points in time, and the relationship between these changes and vehicle mileage and operating environment (such as temperature and altitude) are recorded. For example, as a battery in an electric vehicle ages, the normal monitoring thresholds of sensors may change. A time series knowledge graph can accurately record and reflect these changes, enabling more precise judgment of component abnormalities. For example, entities and relationships in vehicle time series data are obtained, such as the monitoring relationship between the engine and the sensor. Embedding transformations are performed on the entities and relationships in the vehicle time series data to obtain a second feature vector. The corresponding historical feature vector for the second feature vector is obtained from the time series knowledge graph, i.e., the historical feature vector for the engine. The second feature vector and the historical feature vector are then input into a diagnostic analysis model for training and update, resulting in an updated diagnostic analysis model. By using the time series knowledge graph to obtain the historical feature vector and inputting it into the diagnostic analysis model along with the second feature vector, the analysis data becomes more comprehensive, thereby improving the accuracy of vehicle diagnostic analysis results.

[0087] In some examples, after obtaining the historical feature vector corresponding to the second feature vector from the temporal knowledge graph, the method further includes:

[0088] Check whether the historical eigenvector corresponding to the second eigenvector is missing;

[0089] If there is any missing data, supplementary data is generated based on other historical feature vectors in the temporal knowledge graph;

[0090] The missing part of the historical feature vector corresponding to the second feature vector is filled in according to the supplementary data.

[0091] Specifically, supplementary data is generated based on other historical feature vectors in the time series knowledge graph; for example, the second feature vector refers to a vector composed of the features of engine temperature, vehicle speed, and fuel consumption, and the historical feature vector refers to a vector composed of the features of engine temperature, vehicle speed, and fuel consumption recorded at various moments in the past; after checking the historical feature vectors, it is found that the engine temperature data is missing in the historical feature vector at time t5, and supplementary data is generated based on the engine temperature data at time t4.

[0092] Supplementary data is generated by checking historical feature vectors to ensure the continuity of the diagnostic analysis process and to improve the accuracy of the diagnostic results.

[0093] In some examples, this also includes:

[0094] Processing the second feature vector using a preset multi-level ring queue to extract data features; analyzing the data features in combination with the time series data stored in the preset multi-level ring queue;

[0095] If a new relationship is obtained, the relationship between entities in the temporal knowledge graph is updated in chronological order according to the new relationship.

[0096] Specifically, the first-in-first-out feature of the multi-level circular queue is used to store data in chronological order, and the data features related to the time series are extracted from the second feature vector. The extracted data features are combined with the time series data stored in the multi-level circular queue, and analyzed using a causal reasoning algorithm. If new relationships are found, they are integrated into the time series knowledge graph in chronological order. Figure 4 A schematic diagram of the transformation of entities and relationships in a temporal knowledge graph provided in this application, such as Figure 4 As shown in the figure, nodes (such as fault point, problem A, etc.) represent entities, and the lines between nodes represent the relationships between entities. The connection relationships between entities change over time. For example, at time t1, fault B is connected to problem A and fault C. At time t2, fault B is only connected to problem A, and problem A is also connected to fault C. At time tn, new problems D and E appear. Problem D is connected to fault B, and problem E is connected to fault C.

[0097] By integrating new relationships into the time series knowledge graph in chronological order, the real-time and comprehensiveness of the time series knowledge graph is guaranteed to improve the accuracy of diagnostic analysis.

[0098] In some examples, Figure 5 A flow chart of a diagnostic method for a smart car provided in this application is shown as follows: Figure 5 As shown, after obtaining the historical feature vector corresponding to the first feature vector from the time series knowledge graph, it also includes:

[0099] S501, when a vehicle failure is detected, obtaining vehicle time series data based on vehicle failure data;

[0100] S502, obtaining entities and relationships in the vehicle time series data, and obtaining a third eigenvector based on the entities and relationships in the vehicle time series data;

[0101] S503: Input the third eigenvector into the updated diagnosis analysis model to obtain a diagnosis result.

[0102] In the above example, when a vehicle fault is detected, vehicle time series data is obtained based on the vehicle fault data, and the entities and relationships in the vehicle time series data are obtained. Based on the entities and relationships in the vehicle time series data, a third eigenvector is obtained. This third eigenvector is input into the updated diagnostic analysis model to obtain a diagnostic result. The analysis of the time series data using a causal inference algorithm can reveal the causal relationship between the fault cause and the fault manifestation, thereby improving the accuracy of the fault diagnosis and analysis results.

[0103] In some examples, this also includes:

[0104] Display the diagnostic results on a visual interface so that users can repair the faults based on the diagnostic results;

[0105] If the vehicle fault is not repaired successfully, the step of obtaining entities and relationships in the vehicle time series data is executed again to obtain a new diagnosis result;

[0106] If the vehicle fault is successfully repaired, the time series knowledge graph is optimized based on the diagnosis results; among them, the time series knowledge graph represents the relationship between the vehicle's historical data, diagnosis process and solution under different time information.

[0107] The diagnostic results are displayed on a visual interface. Specifically, the diagnostic results are displayed in chronological order. The diagnostic results include a basic fault diagnosis report, solution steps and operating prompts, the change curve of key vehicle parameters (such as engine speed, vehicle speed, oil temperature, and water temperature) in a period of time before and after the fault occurs, and a comparison of these parameters with the normal operating range. By displaying in chronological order, users can freely choose to view data and analysis results of different time periods on the timeline, and gain a deeper understanding of the occurrence and development of the fault. For example, by clicking on a specific point on the timeline, you can view the detailed operating status and related sensor data of the vehicle at that moment. At the same time, according to the implementation steps of the maintenance recommendations, the time sequence and estimated execution time of each step are displayed to help users better determine the maintenance process.

[0108] After the vehicle is repaired, if the vehicle still has a fault, the vehicle's time series data is obtained to re-analyze the cause of the fault; if the vehicle is successfully repaired, the correlation in the time series knowledge graph and the parameters of the diagnostic analysis model are optimized based on the diagnosis results to improve the accuracy of future fault diagnosis.

[0109] A method for storing time series data provided by an embodiment of the present application is applied to the vehicle side, including: obtaining time series data generated by vehicle operation, determining to-be-stored in a preset multi-level circular queue based on the time series data generated by vehicle operation; determining fused time series data and unfused time series data based on the time series data generated by vehicle operation; wherein, the fused time series data represents the data corresponding to the same entity in the time series data generated by vehicle operation as that in the circular queue to be stored; for unfused time series data, determining whether the circular queue to be stored can store unfused time series data; if the unfused time series data cannot be stored, performing data distillation on the time series data to be stored in the circular queue; storing the unfused time series data in the circular queue to be stored; if the unfused time series data can be stored, storing the unfused time series data in the circular queue to be stored. This solution improves the storage effect of time series data by using a multi-level circular queue to store the time series data generated by vehicle operation according to corresponding different periods, and performing data distillation on the circular queue at the same time.

[0110] In practical applications, when a vehicle breaks down, it quickly searches for relevant nodes and relationships in the time-series knowledge graph, infers the possible cause of the failure through a diagnostic analysis model, and performs in-depth analysis based on information in the time dimension, outputting a diagnostic result regarding the vehicle failure. Furthermore, when the vehicle encounters a new fault or abnormal state, it can analyze the diagnostic model and the time-series knowledge graph to predict links and infer the cause of the failure, while also taking time into account, thus achieving truly intelligent diagnosis. The diagnostic results can then be dynamically adjusted and optimized based on data at different time points.

[0111] Example 2

[0112] Figure 6 This is a schematic diagram of a storage device for time series data provided by this application. Figure 6 Shown, including:

[0113] An acquisition module 10 is configured to acquire time series data generated by vehicle operation and determine a ring queue to be stored from a preset multi-level ring queue based on the time series data generated by vehicle operation; wherein a higher level ring queue corresponds to a longer time period; the preset multi-level ring queue is configured to store the time series data generated by vehicle operation according to the corresponding time period;

[0114] The processing module 20 is configured to determine, based on the time series data generated by the vehicle operation, time series data that can be fused and time series data that cannot be fused; wherein the time series data that can be fused represents data corresponding to the same entity in the time series data generated by the vehicle operation that is to be stored in the circular queue;

[0115] A judgment module 30 is used to judge whether the ring queue to be stored can store the unfused time series data;

[0116] The processing module 20 is further configured to, if the unfusible time series data cannot be stored, perform data distillation on the time series data to be stored in the circular queue; and store the unfusible time series data in the circular queue to be stored;

[0117] The processing module 20 is further configured to store the unfused time series data in a to-be-stored circular queue if the unfused time series data can be stored.

[0118] With the increasing adoption of high-performance computers in automotive systems, these systems are becoming increasingly dependent on software. Because software architectures require rapid and flexible deployment of functions to vehicles, Service-Oriented Vehicle Diagnostics (SOVD) is being introduced to diagnose new systems and traditional sensors and actuators. SOVD is a diagnostic model based on advanced information technology and standardized architecture. Limited vehicle storage space makes storing time series data generated during operation expensive, making it difficult to store large amounts of historical and new time series data. Therefore, this solution introduces a multi-level ring queue to store large amounts of time series data.

[0119] Among them, a multi-level circular queue is a data structure composed of multiple circular queues, which are organized according to a certain hierarchical relationship. Each circular queue stores data of different time periods with a fixed capacity; wherein, the time period can be divided according to preset requirements, such as years, months, weeks, days, hours, minutes, and seconds; in addition, the larger the time period, the longer the time interval; for example, the first-level circular queue stores the comprehensive status data of key systems in units of hours, and the second-level circular queue stores detailed data of specific components in units of minutes.

[0120] The thick solid lines in the figure represent the original "edges" between nodes in the time series knowledge graph. Even in different ring queue levels, the connections between levels will not be disconnected, and the connections represented by the thick solid lines express the association between nodes.

[0121] In this embodiment, for the unfused time series data, it is determined whether the circular queue to be stored can store the unfused time series data; specifically, for example, the storage capacity of the circular queue to be stored is 5. If the current circular queue to be stored has stored time series data to the storage capacity of 5, it means that the circular queue to be stored cannot store the unfused time series data; if the current circular queue to be stored has stored time series data to the storage capacity of 4, it means that the circular queue to be stored can store the unfused time series data, and after storage, the amount of data stored in the circular queue to be stored reaches 5.

[0122] In practical applications, time series data generated by vehicle operation is obtained, and based on this time series data, a circular queue to be stored is determined from a preset multi-level circular queue. For example, the multi-level circular queue is divided into levels based on time periods of minutes and hours, resulting in a first-level circular queue with an hourly time period and a second-level circular queue with a minutely time period. For example, if engine speed data per minute is obtained, the circular queue to be stored for this data is determined to be a second-level circular queue with a minutely time period based on the preset multi-level circular queue. A determination is made as to whether time series data for engine speed exists in the second-level circular queue. If not, a determination is made as to whether the second-level circular queue can store the engine speed data. If not, the data is first distilled before storage. If storage is possible, the data is directly stored.

[0123] By utilizing multi-level circular queues to store time-series data generated by vehicle operation, data can be stored in different time periods based on real-time requirements, enabling rapid data access and processing, and improving data query efficiency. For example, during vehicle operation, critical real-time data such as vehicle speed and engine speed require timely processing and analysis. This data can be stored in a first-level circular queue with a smaller time granularity for rapid access and processing. Less urgent data, such as comprehensive mileage data and average fuel consumption over time, can be stored in higher-level circular queues with a larger time granularity. This ensures the real-time nature of critical data while effectively managing large amounts of non-real-time data.

[0124] In some examples, the processing module 20 is further configured to:

[0125] Based on the causal reasoning algorithm, the time series data in the preset circular queue is analyzed to obtain the causal relationship between different events; wherein different events are caused by the state changes of entities in the time series data stored in the preset circular queue;

[0126] Integrate the causal relationship between different events into the time series knowledge graph; among them, the time series knowledge graph is constructed based on historical time series data.

[0127] Causal inference algorithms are a class of methods and techniques used to discover causal relationships in data. By analyzing data using causal inference algorithms, we can determine the causal relationship between variables—that is, whether a change in one variable causes a change in another, as well as the strength and direction of this causal relationship. Specifically, causal inference algorithms (such as Do-Calculus and structural causal models) are used to analyze data in a circular queue to identify causal relationships between different events. For example, analyzing time series data from engines and temperature sensors in a multi-stage circular queue reveals that higher engine speeds generate more heat, and thus, generally, higher temperature sensor readings. Therefore, the increased temperature sensor reading is caused by the increased engine speed.

[0128] By integrating causal relationships into the temporal knowledge graph, the semantic knowledge of the temporal knowledge graph is enriched, thereby improving the accuracy of automobile diagnosis.

[0129] Among them, the time series knowledge graph is constructed based on historical time series data. In practical applications, the entities, relationships and time information in the vehicle historical data are obtained; the types of entities and relationships in the vehicle historical data are determined; quadruple groups are constructed based on the types and time information of entities and relationships in the vehicle historical data; the initial time series knowledge graph is established based on the quadruple groups, and the diagnostic process and solution are integrated into the initial time series knowledge graph to obtain the time series knowledge graph.

[0130] Specifically, entity types refer to the classification of entities into different categories based on their characteristics and attributes. For example, "car," "truck," and "bus" all belong to the entity type "vehicle"; "oil pump," "oil filter," and "oil channel" all belong to the entity type "engine lubrication system component." Relationship types categorize and define relationships between entities. For example, "installed on" is a relationship type that represents the relationship between "engine" and "car," i.e., "engine installed on car." "Connected" is another relationship type that can describe the relationship between "drive shaft" and "transmission," i.e., "drive shaft connected to transmission." Different relationship types accurately express connections of varying nature between entities. A quaternary entity consists of a "head entity, relationship, tail entity, time" structure. For example, (temperature sensor, monitoring, engine, time) indicates that the temperature sensor was monitoring the engine at that time; (vehicle, installed on, engine, time) indicates that the engine was installed on the vehicle at that time.

[0131] Integrate diagnostic processes and solutions into the initial temporal knowledge graph. Specifically, when obtaining diagnostic processes and solutions from repair manuals, analyze how their effectiveness changes over time, based on vehicle upgrades and technological evolution. For example, repair methods that worked for older models may no longer be applicable or require adjustments for newer models. By incorporating this time-related knowledge into the temporal knowledge graph, the reasoning process can be guided to more accurately select appropriate diagnostic and solution options.

[0132] After establishing the initial temporal knowledge graph based on the quadruple, the temporal knowledge graph needs to be initialized. Specifically, an initial feature vector is obtained based on the entities and relationships in the vehicle's historical data. The initial value of the temporal knowledge graph's feature vector is set to this initial feature vector. In practical applications, a pre-trained machine algorithm model is used to analyze the entities and relationships in the vehicle's historical data to obtain the initial feature vector. For example, the engine temperature data, transmission operating status data, and coolant pump operating parameters monitored by a temperature sensor over a period of time are the engine and temperature sensor entities, and the monitoring relationship is used. The pre-trained machine algorithm model is used to analyze this data, along with the entities and relationships, to obtain a vector that reflects the entity and relationship information as the initial feature vector. The initial value of the temporal knowledge graph's feature vector is set to this initial feature vector. The initial value of the engine node's feature vector is set to the portion of the initial feature vector related to the engine, and the initial value of the temperature sensor node's feature vector is set to the portion of the initial feature vector related to the temperature sensor.

[0133] By constructing a time-series knowledge graph, we can integrate and represent vehicle entities and relationships, and clearly record their dynamic changes over time. Furthermore, by leveraging the rich information in historical vehicle data to obtain initial feature vectors, we initialize the feature vectors of the time-series knowledge graph, imbuing it with semantic and structural information. This provides a more rational and accurate foundation for subsequent analysis and reasoning tasks based on the time-series knowledge graph, such as predicting engine failures and optimizing vehicle operating parameters.

[0134] In some examples, after determining the fusible time series data and the unfusible time series data based on the time series data generated by the vehicle operation, the processing module 20 is further configured to:

[0135] The fused time series data is merged with the corresponding time series data in the ring queue to be stored, and the data is stored in the ring queue to be stored.

[0136] In combination with the above example, specifically, the data of engine speed is fused with the data about engine speed stored in the circular queue, and the fused time series data is stored in the circular queue; for example, the time series data of engine speed collected with a time period of minutes is obtained, and it can be known that the circular queue to be stored is a secondary circular queue; the time series data of engine speed stored in the secondary circular queue is fused with the collected time series data of engine speed to obtain the fused time series data of engine speed, and store it in the secondary circular queue.

[0137] In some examples, when performing data distillation on the time series data to be stored in the circular queue, the determination module 30 is further configured to:

[0138] Determine whether the time period corresponding to the time series data to be stored in the ring queue reaches the time period corresponding to the previous level ring queue;

[0139] If the time period corresponding to the previous level circular queue is reached, the time series data to be stored in the circular queue is distilled to obtain new time series data and the time series data to be stored in the circular queue is deleted; and the new time series data is stored in the circular queue to be stored and the previous level circular queue respectively;

[0140] If the time period corresponding to the previous level ring queue is not reached, the time series data to be stored in the ring queue is subjected to data distillation to obtain new time series data and delete the time series data to be stored in the ring queue; and the new time series data is stored in the ring queue to be stored.

[0141] Among them, data distillation is a data processing technology that aims to extract valuable information from large amounts of raw data while reducing the volume and complexity of the data for easy storage, transmission and analysis.

[0142] In combination with the above example, based on the acquired time series data generated by vehicle operation, it is determined that the circular queue to be stored is a secondary circular queue with a time period of minutes, and the upper level circular queue of the secondary circular queue is a primary circular queue with a time period of hours. In this example, it is determined whether the time period of the time series data stored in the secondary circular queue reaches the time period corresponding to the primary circular queue; for example, the amount of data currently stored in the secondary circular queue is 5, which has reached the maximum storage capacity, but the stored time series data is 50 minutes of time series data. Therefore, the secondary circular queue only stores time series data with a time period of 50 minutes, which does not reach the time period corresponding to the primary circular queue.

[0143] In actual applications, if the time period corresponding to the time series data stored in the secondary circular queue reaches one hour, the new time series data obtained after data distillation of the secondary circular queue is stored in the primary and secondary circular queues respectively. If the time period corresponding to the time series data stored in the secondary circular queue does not reach one hour, only the new time series data is stored in the secondary circular queue. At the same time, after data distillation, the original time series data stored in the secondary circular queue is deleted.

[0144] By introducing data distillation, the storage and deployment costs on vehicles can be reduced. At the same time, data distillation is used to compress the data in the circular queue, retain key data useful for causal reasoning to obtain new time series data, and delete historical time series data. This reduces storage costs and can retain key data of historical data.

[0145] In some examples, the acquisition module 10 is further configured to:

[0146] Obtain entities and relationships in the new time series data; obtain the first eigenvector based on the entities and relationships in the new time series data;

[0147] Update the feature vector in the time series knowledge graph to the first feature vector.

[0148] As the circular queue data is updated, the relationships represented by the edges will be adjusted according to the latest data. Combined with the previous example, the new time series data is processed, the entities and relationships in the new time series data are analyzed, and the new relationships between the entities are obtained. The entities and new relationships are embedded and transformed to obtain the first eigenvector, and the time series knowledge graph is updated to ensure that the relationships in the time series knowledge graph are consistent with the actual relationships.

[0149] Specifically, entities include vehicle components (such as engines, transmissions, batteries, and their status information at different times), sensors, error codes, and maintenance operations. Relationships describe the interactions between entities over time; relationships include but are not limited to monitoring relationships, ownership relationships, triggering relationships, repair relationships, impact relationships, and dependency relationships. Monitoring relationships refer to sensors monitoring changes in the status of vehicle components, ownership relationships refer to software modules belonging to specific vehicle systems and their version evolution, triggering relationships refer to the development of error codes from specific problem sources and taking the root causes into account, repair relationships refer to solutions used to repair specific problem sources and tracking of repair effects, impact relationships refer to how the vehicle's operating status affects the operation of components and sensors, and dependency relationships refer to the dynamic changes in a software module's dependence on certain hardware components or other software modules during its life cycle. Figure 3 A schematic diagram of entities and relationships in a temporal knowledge graph provided for this application, such as Figure 3As shown in Figure 1, at a certain moment, entities in the temporal knowledge graph interact with each other; the nodes in the graph represent entities and the edges represent relationships.

[0150] Relationships describe the interactions between entities over time, such as "component - monitored by a sensor over a certain period of time", "error code - indicates component failure at a specific time", and "repair action - repairing a component at a certain point in time and tracking the subsequent repair results".

[0151] In some examples, before storing the new time series data in the to-be-stored ring queue and the previous-level ring queue, the determination module 30 is further configured to:

[0152] Determine whether the previous level ring queue can store new time series data;

[0153] If the new time series data cannot be stored, the time series data in the previous level ring queue is distilled and the new time series data is stored in the previous level ring queue.

[0154] In combination with the above example, before storing new time series data in the first-level circular queue, determine whether the amount of time series data stored in the first-level circular queue reaches the storage capacity. For example, the storage capacity of the first-level circular queue is 8. If the amount of data stored in the first-level circular queue is 8 at this time, the first-level circular queue will be distilled before storing the new time series data.

[0155] In practical applications, the levels of a multi-level ring queue are configured based on specific needs. For example, a multi-level ring queue can be constructed with ring queues for hourly, daily, weekly, and monthly time periods. When data distillation is performed on any level of a ring queue and needs to be stored in the previous level of the ring queue, the previous level of the ring queue must first determine whether it can store the time series data obtained after the data distillation.

[0156] In some examples, the acquisition module 10 is further configured to:

[0157] Obtaining entities and relationships in the vehicle time series data, and obtaining a second eigenvector based on the entities and relationships in the vehicle time series data;

[0158] Obtain the historical feature vector corresponding to the second feature vector from the time series knowledge graph;

[0159] The second eigenvector and the historical eigenvector are input into the diagnosis and analysis model for update training to obtain an updated diagnosis and analysis model.

[0160] The trend toward intelligent and connected vehicles generates massive amounts of time-series data during vehicle operation. Traditional automotive diagnostic methods struggle to effectively process this complex time-series information and are unable to fully exploit the underlying patterns within the data. Time-series knowledge graphs, as powerful knowledge representation and reasoning tools, can integrate diverse information within the automotive field. Specifically, time tags are added to entities and relationships within the time-series knowledge graph to represent the timeliness and dynamic nature of knowledge. Diagnostic analysis models, such as the Recurrent Event Network (RE-NET) model, are examples.

[0161] In this example, the diagnostic analysis model deployed on the vehicle side is obtained by pre-training the machine algorithm model on the cloud side using historical time series data through knowledge distillation. The diagnostic analysis model is deployed on the vehicle side, and the time series data generated by the vehicle operation is used to update the diagnostic analysis model, thereby improving the diagnostic analysis model's ability to diagnose the vehicle.

[0162] In practical applications, entities are not simply recorded for their basic operational states, such as starting and driving. Rather, the different states of the entities over different time periods, the numerical changes in performance indicators at different points in time, and the relationship between these changes and vehicle mileage and operating environment (such as temperature and altitude) are recorded. For example, as a battery in an electric vehicle ages, the normal monitoring thresholds of sensors may change. A time series knowledge graph can accurately record and reflect these changes, enabling more precise judgment of component abnormalities. For example, entities and relationships in vehicle time series data are obtained, such as the monitoring relationship between the engine and the sensor. Embedding transformations are performed on the entities and relationships in the vehicle time series data to obtain a second feature vector. The corresponding historical feature vector for the second feature vector is obtained from the time series knowledge graph, i.e., the historical feature vector for the engine. The second feature vector and the historical feature vector are then input into a diagnostic analysis model for training and update, resulting in an updated diagnostic analysis model. By using the time series knowledge graph to obtain the historical feature vector and inputting it into the diagnostic analysis model along with the second feature vector, the analysis data becomes more comprehensive, thereby improving the accuracy of vehicle diagnostic analysis results.

[0163] In some examples, after obtaining the historical feature vector corresponding to the second feature vector from the temporal knowledge graph, the processing module 20 is further configured to:

[0164] Check whether the historical eigenvector corresponding to the second eigenvector is missing;

[0165] If there is any missing data, supplementary data is generated based on other historical feature vectors in the temporal knowledge graph;

[0166] The missing part of the historical feature vector corresponding to the second feature vector is filled in according to the supplementary data.

[0167] Specifically, supplementary data is generated based on other historical feature vectors in the time series knowledge graph; for example, the second feature vector refers to a vector composed of the features of engine temperature, vehicle speed, and fuel consumption, and the historical feature vector refers to a vector composed of the features of engine temperature, vehicle speed, and fuel consumption recorded at various moments in the past; after checking the historical feature vectors, it is found that the engine temperature data is missing in the historical feature vector at time t5, and supplementary data is generated based on the engine temperature data at time t4.

[0168] Supplementary data is generated by checking historical feature vectors to ensure the continuity of the diagnostic analysis process and to improve the accuracy of the diagnostic results.

[0169] In some examples, the processing module 20 is further configured to:

[0170] Processing the second feature vector using a preset multi-level ring queue to extract data features; analyzing the data features in combination with the time series data stored in the preset multi-level ring queue;

[0171] If a new relationship is obtained, the relationship between entities in the temporal knowledge graph is updated in chronological order according to the new relationship.

[0172] Specifically, the first-in-first-out feature of the multi-level circular queue is used to store data in chronological order, and the data features related to the time series are extracted from the second feature vector. The extracted data features are combined with the time series data stored in the multi-level circular queue, and analyzed using a causal reasoning algorithm. If new relationships are found, they are integrated into the time series knowledge graph in chronological order. Figure 4 A schematic diagram of the transformation of entities and relationships in a temporal knowledge graph provided in this application, such as Figure 4 As shown in the figure, nodes (such as fault point, problem A, etc.) represent entities, and the lines between nodes represent the relationships between entities. The connection relationships between entities change over time. For example, at time t1, fault B is connected to problem A and fault C. At time t2, fault B is only connected to problem A, and problem A is also connected to fault C. At time tn, new problems D and E appear. Problem D is connected to fault B, and problem E is connected to fault C.

[0173] By integrating new relationships into the time series knowledge graph in chronological order, the real-time and comprehensiveness of the time series knowledge graph is guaranteed to improve the accuracy of diagnostic analysis.

[0174] In some examples, the processing module 20 is further configured to:

[0175] When a vehicle failure is detected, vehicle time series data is obtained based on the vehicle failure data;

[0176] Obtaining entities and relationships in the vehicle time series data, and obtaining a third eigenvector based on the entities and relationships in the vehicle time series data;

[0177] The third eigenvector is input into the updated diagnosis analysis model to obtain a diagnosis result.

[0178] In the above example, when a vehicle fault is detected, vehicle time series data is obtained based on the vehicle fault data, and the entities and relationships in the vehicle time series data are obtained. Based on the entities and relationships in the vehicle time series data, a third eigenvector is obtained. This third eigenvector is input into the updated diagnostic analysis model to obtain a diagnostic result. The analysis of the time series data using a causal inference algorithm can reveal the causal relationship between the fault cause and the fault manifestation, thereby improving the accuracy of the fault diagnosis and analysis results.

[0179] In some examples, the processing module 20 is further configured to:

[0180] Display the diagnostic results on a visual interface so that users can repair the faults based on the diagnostic results;

[0181] If the vehicle fault is not repaired successfully, the step of obtaining entities and relationships in the vehicle time series data is executed again to obtain a new diagnosis result;

[0182] If the vehicle fault is successfully repaired, the time series knowledge graph is optimized based on the diagnosis results; among them, the time series knowledge graph represents the relationship between the vehicle's historical data, diagnosis process and solution under different time information.

[0183] The diagnostic results are displayed on a visual interface. Specifically, the diagnostic results are displayed in chronological order. The diagnostic results include a basic fault diagnosis report, solution steps and operating prompts, the change curve of key vehicle parameters (such as engine speed, vehicle speed, oil temperature, and water temperature) in a period of time before and after the fault occurs, and a comparison of these parameters with the normal operating range. By displaying in chronological order, users can freely choose to view data and analysis results of different time periods on the timeline, and gain a deeper understanding of the occurrence and development of the fault. For example, by clicking on a specific point on the timeline, you can view the detailed operating status and related sensor data of the vehicle at that moment. At the same time, according to the implementation steps of the maintenance recommendations, the time sequence and estimated execution time of each step are displayed to help users better determine the maintenance process.

[0184] After the vehicle is repaired, if the vehicle still has a fault, the vehicle's time series data is obtained to re-analyze the cause of the fault; if the vehicle is successfully repaired, the correlation in the time series knowledge graph and the parameters of the diagnostic analysis model are optimized based on the diagnosis results to improve the accuracy of future fault diagnosis.

[0185] In practical applications, when a vehicle breaks down, it quickly searches for relevant nodes and relationships in the time-series knowledge graph, infers the possible cause of the failure through a diagnostic analysis model, and performs in-depth analysis based on information in the time dimension, outputting a diagnostic result regarding the vehicle failure. Furthermore, when the vehicle encounters a new fault or abnormal state, it can analyze the diagnostic model and the time-series knowledge graph to predict links and infer the cause of the failure, while also taking time into account, thus achieving truly intelligent diagnosis. The diagnostic results can then be dynamically adjusted and optimized based on data at different time points.

[0186] Figure 7 This is a schematic diagram of the structure of an electronic device provided by this application. Figure 7 As shown, the electronic device 50 provided in this embodiment includes: at least one processor 501 and a memory 502. Optionally, the device 50 further includes a communication component 503. The processor 501, the memory 502 and the communication component 503 are connected via a bus 504.

[0187] In a specific implementation process, at least one processor 501 executes the computer-executable instructions stored in the memory 502, so that the at least one processor 501 performs the above method.

[0188] The specific implementation process of the processor 501 can be found in the above method embodiment. Its implementation principle and technical effects are similar and will not be repeated here in this embodiment.

[0189] In the above embodiments, it should be understood that the processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASICs), etc. A general-purpose processor may be a microprocessor or any conventional processor. The steps of the method disclosed in the present invention may be directly executed by a hardware processor or by a combination of hardware and software modules within the processor.

[0190] The memory may include random access memory (RAM) and may also include non-volatile memory (NVM), such as at least one disk storage.

[0191] A bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus. Buses can be categorized as address buses, data buses, and control buses. For ease of illustration, the buses in the drawings of this application are not limited to just one bus or just one type of bus.

[0192] The present application also provides a computer program product, including a computer program, which implements the above method when executed by a processor.

[0193] The present application also provides a computer-readable storage medium, in which computer-executable instructions are stored. When a processor executes the computer-executable instructions, the above method is implemented.

[0194] The readable storage medium may be implemented by any type of volatile or non-volatile memory device, or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The readable storage medium may be any available medium that can be accessed by a general-purpose or special-purpose computer.

[0195] An exemplary readable storage medium is coupled to a processor so that the processor can read information from the readable storage medium and write information to the readable storage medium. Of course, the readable storage medium can also be an integral part of the processor. The processor and the readable storage medium can be located in an application specific integrated circuit (ASIC). Of course, the processor and the readable storage medium can also exist in the device as discrete components.

[0196] The division of units is merely a logical functional division; actual implementations may employ alternative divisions, such as combining or integrating multiple units or components into another system, or omitting or disabling certain features. Furthermore, any direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection between devices or units, either through an interface, electrical, mechanical, or other means.

[0197] Units described as separate components may or may not be physically separate, and components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0198] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0199] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the various embodiments of the method of the present invention. The aforementioned storage medium includes various media that can store program code, such as USB flash drives, mobile hard drives, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical disks.

[0200] Those skilled in the art will appreciate that all or part of the steps in the above-described method embodiments can be implemented using hardware associated with program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0201] Finally, it should be noted that those skilled in the art will readily identify other embodiments of the present invention after considering the specification and practicing the invention disclosed herein. The present invention is intended to cover any variations, uses, or adaptations of the present invention that follow the general principles of the present invention and include common knowledge or customary techniques in the art not disclosed herein. The present invention is not limited to the precise structure described above and illustrated in the accompanying drawings, and various modifications and variations may be made without departing from the scope thereof. The scope of the present invention is limited solely by the appended claims.

Claims

1. A method for storing time series data, applied to a vehicle, characterized in that: include: acquiring time series data generated by vehicle operation, and determining a ring queue to be stored from a preset multi-level ring queue based on the time series data generated by the vehicle operation; wherein a higher-level ring queue corresponds to a longer time period; the preset multi-level ring queue is used to store the time series data generated by the vehicle operation according to the corresponding time period; Determining fusible time series data and non-fusible time series data based on the time series data generated by the vehicle operation; wherein the fusible time series data represents data corresponding to the same entity in the time series data generated by the vehicle operation as that in the circular queue to be stored; For the unfusible time series data, determining whether the to-be-stored circular queue can store the unfusible time series data; If the unfused time series data cannot be stored, performing data distillation on the time series data to be stored in the circular queue; storing the unfused time series data in the circular queue to be stored; If the unfused time series data can be stored, the unfused time series data is stored in the to-be-stored circular queue.

2. The method according to claim 1, characterized in that The method further comprises: Based on a causal reasoning algorithm, the time series data in the preset multi-level circular queue is analyzed to obtain a causal relationship between different events; wherein the different events are caused by state changes of entities in the time series data stored in the preset circular queue; Integrate the causal relationship between the different events into a time series knowledge graph; wherein the time series knowledge graph is constructed based on historical time series data.

3. The method according to claim 1, characterized in that After determining the time series data that can be fused and the time series data that cannot be fused based on the time series data generated by the vehicle operation, the method further includes: The fusible time series data is fused with the corresponding time series data in the to-be-stored circular queue, and the data is stored in the to-be-stored circular queue.

4. The method according to claim 1, wherein The performing data distillation on the time series data to be stored in the ring queue specifically includes: Determine whether the time period corresponding to the time series data to be stored in the circular queue reaches the time period corresponding to the previous level circular queue; If the time period corresponding to the previous level circular queue is reached, the time series data to be stored in the circular queue is subjected to data distillation to obtain new time series data and the time series data to be stored in the circular queue is deleted; and the new time series data is stored in the circular queue to be stored and the previous level circular queue respectively; If the time period corresponding to the previous level circular queue is not reached, the time series data to be stored in the circular queue is subjected to data distillation to obtain new time series data and the time series data to be stored in the circular queue is deleted; and the new time series data is stored in the circular queue to be stored.

5. The method according to claim 4, characterized in that The method further comprises: Acquire entities and relationships in the new time series data; obtain a first feature vector based on the entities and relationships in the new time series data; Update the feature vector in the temporal knowledge graph to the first feature vector.

6. The method according to claim 4, characterized in that Before storing the new time series data into the to-be-stored circular queue and the upper-level circular queue, the method further includes: Determine whether the previous level ring queue can store the new time series data; If the new time series data cannot be stored, the time series data in the previous level circular queue is subjected to data distillation, and then the new time series data is stored in the previous level circular queue.

7. The method according to any one of claims 1 to 6, characterized in that Also includes: Acquire entities and relationships in the vehicle time series data, and obtain a second feature vector based on the entities and relationships in the vehicle time series data; Obtaining a historical feature vector corresponding to the second feature vector from the time series knowledge graph; The second feature vector and the historical feature vector are input into the diagnosis and analysis model for update training to obtain an updated diagnosis and analysis model.

8. The method according to claim 7, characterized in that After obtaining the historical feature vector corresponding to the second feature vector from the time series knowledge graph, the method further includes: Checking whether a historical feature vector corresponding to the second feature vector is missing; If there is any missing data, supplementary data is generated based on other historical feature vectors in the temporal knowledge graph; The missing part of the historical feature vector corresponding to the second feature vector is filled in according to the supplementary data.

9. The method according to claim 7, characterized in that The method further comprises: Processing the second feature vector using the preset multi-level circular queue to extract data features; analyzing the data features in combination with the time series data stored in the preset multi-level circular queue; If a new relationship is obtained, the relationship between entities in the temporal knowledge graph is updated in chronological order according to the new relationship.

10. The method according to claim 9, characterized in that The method further comprises: When a vehicle failure is detected, obtaining the vehicle time series data according to the vehicle failure data; Obtaining entities and relationships in the vehicle time series data, and obtaining a third eigenvector based on the entities and relationships in the vehicle time series data; The third eigenvector is input into the updated diagnosis analysis model to obtain a diagnosis result.

11. The method according to claim 10, characterized in that The method further comprises: Displaying the diagnostic results on a visual interface so that the user can perform fault repair according to the diagnostic results; If the vehicle fault is not repaired successfully, the step of obtaining entities and relationships in the vehicle time series data is performed again to obtain a new diagnosis result; If the vehicle fault is successfully repaired, the time series knowledge graph is optimized based on the diagnosis result; wherein, the time series knowledge graph represents the relationship between the vehicle's historical data, diagnosis process and solution under different time information.

12. A storage device for time series data, characterized in that: include: an acquisition module, configured to acquire time series data generated by vehicle operation, and determine a ring queue to be stored from a preset multi-level ring queue based on the time series data generated by the vehicle operation; wherein a higher level ring queue corresponds to a longer time period; the preset multi-level ring queue is configured to store the time series data generated by the vehicle operation according to the corresponding time period; a processing module, configured to determine, based on the time series data generated by the vehicle operation, time series data that can be fused and time series data that cannot be fused; wherein the time series data that can be fused represents data corresponding to the same entity in the time series data generated by the vehicle operation as that in the circular queue to be stored; a judgment module, configured to judge, with respect to the unfusible time series data, whether the circular queue to be stored can store the unfusible time series data; The processing module is further configured to, if the unfused time series data cannot be stored, perform data distillation on the time series data to be stored in the circular queue; and store the unfused time series data in the circular queue to be stored; The processing module is further configured to store the unfused time series data in the to-be-stored circular queue if the unfused time series data can be stored.

13. An electronic device, characterized in that: include: Memory, processor; The memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory, so that the processor performs the method according to any one of claims 1 to 11.

14. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer-executable instructions, which are used to implement the method according to any one of claims 1 to 11 when executed by a processor.

15. A computer program product, characterized in that The invention comprises a computer program, which implements the method according to any one of claims 1 to 11 when being executed by a processor.

Images