Time sequence database processing method and system based on artificial intelligence

Abstract

The invention relates to the technical field of databases, and discloses a time sequence database processing method and system based on artificial intelligence. A data acquisition module is used for acquiring multi-source heterogeneous time sequence data, and a data processing module adopts a skip list structure and an out-of-order optimization technology; the data storage module is used for optimizing fragmentation storage and compression through hardware acceleration, the query analysis module is used for multi-dimensional query of a user-defined time window, and the anomaly detection module is used for detecting an abnormal mode in time series data through a machine learning algorithm. And the data indexing module is used for generating an index based on a timestamp and a multi-dimensional condition. According to the method, real-time processing and storage of the time series data can be achieved under the conditions of high concurrency and large data volume, and by using a skip list structure and an out-of-order optimization technology, the system can still keep efficient data writing efficiency under the out-of-order and cut-off environments; and the effect of the requirement for real-time access and storage of mass data in multiple scenes is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of databases, and particularly to a method and system for processing time-series databases based on artificial intelligence. Background Art

[0002] With the rapid development of artificial intelligence technology and its extensive application in the military field, modern military systems increasingly rely on the efficient processing of real-time data and intelligent decision-making. To meet the complex and changing environmental requirements on the battlefield, improve combat efficiency, and ensure security, the demand for time-series databases in the military field has increased rapidly. Traditional relational databases have bottlenecks in processing large-scale, highly concurrent, and real-time data, making it difficult to meet the needs of military scenarios. This limitation not only leads to problems of insufficient accuracy and timeliness in situation analysis by commanders, but may even cause misjudgments, thus affecting combat effectiveness.

[0003] In actual combat scenarios, the processing efficiency of out-of-order data in traditional databases is low. Due to network latency, equipment failures, or asynchronous data collection, time-series data often arrives at the database in an out-of-order form, resulting in significant inefficiencies and data inconsistency in processing out-of-order and interrupted data, which in turn affects the timely combat judgment of commanders. Summary of the Invention

[0004] To make up for the above deficiencies, the present invention provides a method and system for processing time-series databases based on artificial intelligence, aiming to improve the problems of significant inefficiency, data inconsistency, and affecting the timely combat judgment of commanders in processing out-of-order and interrupted data.

[0005] In the first aspect, the present invention provides the following technical solution. A time-series database processing system based on artificial intelligence includes:

[0006] A data acquisition module, a data processing module, a data storage module, a query analysis module, an anomaly detection module, and a data index module;

[0007] The data acquisition module is used to collect multi-source heterogeneous time-series data, supporting asynchronous access, edge computing, and dynamic sampling;

[0008] The data processing module uses a skip list structure and out-of-order optimization technology to sort and optimize out-of-order data;

[0009] The data storage module is used to perform sharding storage and compression optimization through hardware acceleration, as well as data backup and disaster recovery functions;

[0010] The query analysis module is used for multi-dimensional queries with user-defined time windows and performs correlation analysis in combination with artificial intelligence algorithms;

[0011] The anomaly detection module is used to detect anomaly patterns in time series data through machine learning algorithms and generate real-time alerts;

[0012] The data indexing module is used to generate indexes based on timestamps and multi-dimensional conditions for retrieval and index maintenance.

[0013] Preferably, the data acquisition module includes a data access unit, an edge computing unit, and a dynamic sampling unit. The data access unit is used to access data sources supporting multiple protocols, including MQTT, HTTP, and TCP / IP protocols. The edge computing unit is used to perform denoising, aggregation, and preliminary filtering on data at the data acquisition end. The dynamic sampling unit is used to dynamically adjust the data acquisition frequency according to network bandwidth, device status, and data priority.

[0014] Preferably, the data processing module includes a skip list optimization unit, a data cleaning unit, and a time alignment unit. The skip list optimization unit is used to dynamically adjust the skip list index levels to adapt to different data traffic and improve the speed of processing and disordered data. The data cleaning unit is used to identify and remove outliers, duplicate values, and fill in missing values. The time alignment unit is used to align asynchronous data from different sources based on timestamps to ensure the temporal consistency of data. The data processing includes cleaning, time alignment, and format standardization of the collected data.

[0015] Preferably, the data storage module includes a sharding storage unit, a compression storage unit, and a backup and disaster recovery unit. The sharding storage unit is used to shard and store data according to time windows or data source devices. The compression storage unit is used to compress and decompress data using a dynamic compression algorithm through hardware acceleration to reduce storage space occupancy. The backup and disaster recovery unit is used for real-time backup, periodic snapshots, and disaster recovery functions.

[0016] Preferably, the query analysis module includes a time window analysis unit, a correlation analysis unit, and a visualization analysis unit. The time window analysis unit is used to support user-defined fixed window, sliding window, and rolling window analysis. The correlation analysis unit is used to perform correlation analysis on multi-source data and generate a dynamic situation model by combining artificial intelligence algorithms. The visualization analysis unit is used to generate visualization analysis results in the form of time series curves and heat maps.

[0017] Preferably, the anomaly detection module includes a pattern recognition unit, a real-time alarm unit, and a dynamic threshold adjustment unit. The pattern recognition unit is used to identify anomaly patterns in data through machine learning algorithms, and the anomaly patterns include mutations, disconnections, and duplicate data. The real-time alarm unit is used to generate alarm information when detecting abnormal data and push it to the user terminal. The dynamic threshold adjustment unit is used to dynamically adjust the threshold of anomaly detection according to historical data and real-time analysis results to reduce false alarms and missed alarms.

[0018] Preferably, the data indexing module includes a time indexing unit, a multi-dimensional retrieval unit, and an index maintenance unit. The time indexing unit is used to generate multi-level indexes for data based on timestamps, and the multi-level indexes are used to improve the query efficiency. The multi-dimensional retrieval unit is used for multi-dimensional combined queries according to conditions such as time range, device ID, and data type. The index maintenance unit is used to update the index structure in real time when data is written or updated.

[0019] In a second aspect, the present invention provides the following technical solution, a time series database processing method based on artificial intelligence, including the following steps:

[0020] S1. Collect multi-source heterogeneous data through the data collection module and perform asynchronous access and dynamic sampling;

[0021] S2. In the data processing module, optimize the sorting of out-of-order data through the skip list optimization unit, and clean, time-align, and format-standardize the data;

[0022] S3. In the data storage module, store the data in slices and optimize the storage efficiency through a dynamic compression algorithm;

[0023] S4. Through the query analysis module, perform multi-dimensional queries and correlation analysis on the data based on user-defined time windows to generate visualization results;

[0024] S5. In the anomaly detection module, perform anomaly detection on the data through machine learning algorithms to generate real-time alarms;

[0025] S6. Generate timestamp indexes for the data through the data indexing module and support multi-dimensional conditional combined queries;

[0026] S7. Utilize the dynamic modeling ability of the system to combine historical data and real-time data to generate correlation analysis models or situation prediction results.

[0027] In a third aspect, the invention provides the following technical solution, a computer device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, it achieves the above-mentioned time series database processing method based on artificial intelligence.

[0028] In a fourth aspect, the present invention provides the following technical solution: a readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the above-mentioned time series database processing method based on artificial intelligence is achieved.

[0029] The present invention has the following beneficial effects:

[0030] 1. In the present invention, real-time processing and storage of time series data can be achieved under high concurrency and large data volume. By using the skip list structure and out-of-order optimization technology, the system can still maintain high data writing efficiency in out-of-order and discontinuous flow environments, achieving the effect of meeting the real-time access and storage requirements of massive data in multiple scenarios.

[0031] 2. In the present invention, through machine learning algorithms and dynamic threshold adjustment, the present invention can accurately detect abnormal patterns in data such as mutations, discontinuous flows, and duplicate data, reducing the false alarm and missed alarm rates. At the same time, combined with the real-time alarm function, the timeliness and accuracy of abnormal handling are guaranteed.

[0032] 3. In the present invention, by combining historical data and real-time data to generate a dynamic association model and situation prediction, reliable prediction capabilities and intelligent analysis support are provided for complex scenarios such as equipment fault warning and situation assessment, significantly improving the application breadth and intelligent level of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 It is a system architecture diagram of the time series database processing system based on artificial intelligence proposed by the present invention;

[0034] Figure 2 It is an architecture diagram of the data acquisition module of the time series database processing system based on artificial intelligence proposed by the present invention;

[0035] Figure 3 It is a relationship architecture diagram of the data acquisition module of the time series database processing system based on artificial intelligence proposed by the present invention;

[0036] Figure 4 It is an architecture diagram of the data processing module of the time series database processing system based on artificial intelligence proposed by the present invention;

[0037] Figure 5 It is a relationship architecture diagram of the data processing module of the time series database processing system based on artificial intelligence proposed by the present invention;

[0038] Figure 6 It is an architecture diagram of the data storage module of the time series database processing system based on artificial intelligence proposed by the present invention;

[0039] Figure 7Relationship architecture diagram of the data storage module of the time-series database processing system based on artificial intelligence proposed by the present invention;

[0040] Figure 8 Query analysis module architecture diagram of the time-series database processing system based on artificial intelligence proposed by the present invention;

[0041] Figure 9 Relationship architecture diagram of the query analysis module of the time-series database processing system based on artificial intelligence proposed by the present invention;

[0042] Figure 10 Anomaly detection module architecture diagram of the time-series database processing system based on artificial intelligence proposed by the present invention;

[0043] Figure 11 Data index module architecture diagram of the time-series database processing system based on artificial intelligence proposed by the present invention;

[0044] Figure 12 Relationship architecture diagram of the data index module of the time-series database processing system based on artificial intelligence proposed by the present invention;

[0045] Figure 13 Method flow chart of the time-series database processing method based on artificial intelligence proposed by the present invention. Detailed implementation manners

[0046] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0047] Embodiment 1

[0048] Referring to Figures 1 - 12 , in the first embodiment of the present invention, the present invention provides a time-series database processing system based on artificial intelligence, including:

[0049] A data acquisition module, a data processing module, a data storage module, a query analysis module, an anomaly detection module, and a data index module;

[0050] The data acquisition module is used to collect multi-source heterogeneous time-series data and support asynchronous access, edge computing, and dynamic sampling;

[0051] The data processing module adopts a skip list structure and out-of-order optimization technology to optimize the sorting of out-of-order data;

[0052] The data storage module is used to enable sharded storage and compression optimization through hardware acceleration, as well as perform data backup and disaster recovery functions;

[0053] The query analysis module is used for multi-dimensional queries with user-defined time windows and performs correlation analysis in combination with artificial intelligence algorithms;

[0054] The anomaly detection module is used to detect anomaly patterns in time-series data through machine learning algorithms and generate real-time alerts;

[0055] The data index module is used to generate indexes based on timestamps and multi-dimensional conditions for retrieval and index maintenance.

[0056] The data acquisition module includes a data access unit, an edge computing unit, and a dynamic sampling unit. The data access unit is used to access data sources that support multiple protocols, including the MQTT, HTTP, and TCP / IP protocols. The edge computing unit is used to perform denoising, aggregation, and preliminary filtering of data at the data acquisition end. The dynamic sampling unit is used to dynamically adjust the data acquisition frequency according to network bandwidth, device status, and data priority.

[0057] Specifically, for the data access unit, it is used to access multi-source heterogeneous data and support the access of data sources including the MQTT, HTTP, and TCP / IP protocols. It standardizes data processing through a protocol adapter. For the MQTT protocol, it accesses data through topic subscription. For the HTTP protocol, it receives data through the REST API. For the TCP / IP protocol, it collects real-time data streams through persistent connections. The data access unit supports multi-threaded processing, can efficiently access data from multiple devices, and is suitable for high-concurrency environments;

[0058] For the edge computing unit, it is used to perform preliminary data processing at the acquisition end, including denoising, aggregation, and filtering. It eliminates sensor noise data through a filtering algorithm (such as mean filtering), uses an aggregation algorithm to perform statistical processing on high-frequency data according to a time window (such as calculating the mean or maximum value), and filters out abnormal data beyond the threshold through rules. For example, when the temperature sensor data exceeds the device range, the data is marked as invalid and filtered out. This unit improves the data quality at the source and reduces the processing pressure at the backend;

[0059] The dynamic sampling unit adjusts the sampling frequency dynamically according to network bandwidth, device status, and data priority. When the network bandwidth is insufficient, it automatically reduces the sampling frequency of low-priority data to ensure the transmission of critical data. When the device is in a low-battery or high-load state, it reduces the sampling frequency to save energy consumption. High-priority data maintains a high sampling frequency, such as battlefield monitoring sensors sampling every second, while environmental data can be reduced to sampling every 10 seconds. This mechanism optimizes the utilization rate of system resources, ensures the real-time nature of core data, and the acquisition module of the present invention can efficiently access various protocol data, improve data quality, and maintain stability in complex network and multi-device environments.

[0060] The data processing module includes a skip list optimization unit, a data cleaning unit, and a time alignment unit. The skip list optimization unit is used to dynamically adjust the number of skip list index levels to adapt to different data traffic, and is used to process and improve the speed of out-of-order data. The data cleaning unit is used to identify and eliminate outliers, duplicate values, and fill in missing values. The time alignment unit is used to align asynchronous data from different sources based on timestamps to ensure the temporal consistency of data. Data processing includes cleaning, time alignment, and format standardization processing of the acquired data.

[0061] Specifically, the skip list optimization unit is used to process out-of-order data by dynamically adjusting the number of skip list index levels to adapt to different data traffic. The dynamic adjustment of the skip list index is optimized according to the size and distribution of real-time data traffic. When the data traffic is large, the number of index levels is increased to improve the efficiency of insertion and search. When the data traffic is small, the number of index levels is reduced to save storage resources and computational overhead. This unit effectively solves the efficiency problem of sorting out-of-order data and significantly improves the processing speed of temporal data;

[0062] The data cleaning unit is used to identify and process outliers, duplicate values, and missing values in the original data, and the accuracy and consistency of the data. By setting predefined rules or algorithms (such as range threshold detection), abnormal data outside the reasonable range is automatically eliminated. Through the duplicate value detection mechanism, redundant data collected repeatedly is removed. For missing data, it is reasonably filled by interpolation or default value filling strategies. This unit ensures the quality of subsequent data processing and analysis;

[0063] The time alignment unit is used to perform timestamp-based alignment processing on asynchronous data from different sources, ensuring the temporal consistency of multi-source data. By comparing the timestamp information of different data, data with a large deviation in the acquisition time point is redistributed to the corresponding time period, and placeholder data is inserted or the time order is adjusted when necessary. This unit is particularly applicable to the data processing scenarios of asynchronous acquisition devices, effectively eliminating the time differences between multi-source asynchronous data. The data processing module also performs format standardization processing on the data, converting the data into a unified structured format (such as JSON or tabular structure) for subsequent storage and analysis. Through data cleaning, time alignment, and format standardization processing, this module can significantly improve the quality and temporal consistency of the data, providing a reliable data foundation for subsequent storage and analysis. Experimental verification shows that this module can significantly improve the data processing speed and accuracy in high-concurrency and out-of-order environments.

[0064] The data storage module includes a sharding storage unit, a compression storage unit, and a backup and disaster recovery unit. The sharding storage unit is used to shard and store data based on time windows or data source devices; the compression storage unit is used to compress and decompress data using a dynamic compression algorithm through hardware acceleration to reduce storage space occupancy; the backup and disaster recovery unit is used for real-time backup, periodic snapshot, and disaster recovery functions.

[0065] Specifically, the sharding storage unit logically shards and stores data according to time windows or data source devices. The sharding storage strategy divides the data into independent shards by setting time windows (such as by day, hour, or minute), and further refines the storage partitions based on the data source devices (such as sensor IDs or device types). This sharding method not only improves the organization and retrievability of the data, but also enables quick location of target data during the query process, reducing the retrieval latency of the storage system. In addition, sharding storage supports a distributed storage architecture, which can evenly distribute the data load among multiple nodes, enhancing the scalability and reliability of the system;

[0066] The compression storage unit compresses and decompresses data using a dynamic compression algorithm (such as LZ4 or ZSTD) through hardware acceleration technology to reduce storage space occupancy and improve storage performance. During the data writing process, a suitable compression algorithm is selected according to the data type and compression rate requirements, achieving fast and efficient data compression; during the data reading process, decompression is performed through hardware acceleration, ensuring low latency and high throughput of data access. Through this compression technology, the storage space occupancy can be significantly reduced while maintaining the data access performance, providing an effective storage solution for processing massive amounts of time-series data;

[0067] The backup and disaster recovery unit is used to provide real-time backup, regular snapshots and disaster recovery functions, ensuring the high reliability and security of data. Through synchronous replication technology, it backs up critical data to a redundant storage system in real time to prevent data loss due to sudden failures or equipment damage. It regularly generates snapshots of the stored data and saves them on independent storage media so that in case of accidental deletion or tampering of data, it can be quickly restored to a specified historical state. In case of major hardware failures or catastrophic events, it quickly restores the storage system through multi-node backup and off-site disaster recovery technologies to ensure the business continuity of the system. The data storage module can efficiently organize and manage massive amounts of time-series data, while reducing storage costs and enhancing the reliability and fault tolerance of the system.

[0068] The query and analysis module includes a time window analysis unit, a correlation analysis unit and a visualization analysis unit. The time window analysis unit is used to support fixed window, sliding window and rolling window analysis defined by users. The correlation analysis unit is used to perform correlation analysis on multi-source data and generate a dynamic situation model by combining artificial intelligence algorithms. The visualization analysis unit is used to generate visualization analysis results in the form of time-series curves and heat maps.

[0069] Specifically, data is divided at fixed time intervals such as 1 hour or 1 day for statistical or query operations, which is suitable for regular summarization and trend analysis. It slides with a fixed time span and supports the analysis of data changes in continuous time periods in a dynamic scenario. Similar to the sliding window, but the start and end points of the rolling window move forward in real time with the data stream and are used for real-time monitoring of streaming data. The time window analysis unit allows users to flexibly define the time window size and type through parameterized configuration to meet various complex time series analysis requirements. The correlation analysis unit combines artificial intelligence algorithms (such as association rule mining, classification algorithms or clustering algorithms) to perform in-depth correlation analysis on multi-source data and generate a dynamic situation model;

[0070] The visualization analysis unit is used to generate a visual display of the analysis results in the form of time-series curves, heat maps, etc., to help users intuitively understand the analyzed data. For time-series data, it generates trend charts, fluctuation charts, etc., which intuitively reflect the changes in data over time. It is used to display the aggregation characteristics of multi-dimensional data, such as the hot spots or time period distributions of sensor data. Combined with dynamic interaction functions, users can zoom in, zoom out or filter data for specific time periods in the chart, so as to quickly obtain the analysis results. Through visual display, users can more intuitively identify data trends, abnormal patterns and complex relationships between multi-dimensional data.

[0071] The anomaly detection module includes a pattern recognition unit, a real-time alarm unit, and a dynamic threshold adjustment unit. The pattern recognition unit is used to identify abnormal patterns in the data through a machine learning algorithm. Abnormal patterns include mutations, interruptions, and duplicate data. The real-time alarm unit is used to generate alarm information when abnormal data is detected and push it to the user end. The dynamic threshold adjustment unit is used to dynamically adjust the threshold of anomaly detection based on historical data and real-time analysis results to reduce false alarms and missed alarms.

[0072] Specifically, the pattern recognition unit uses a time series analysis algorithm based on machine learning (such as support vector machines, LSTM neural networks or clustering algorithms) to detect abnormal patterns in the data. For data mutation anomalies, the system calculates the rate of change of data in a short period of time through a sliding window. When the rate exceeds the set threshold, it is judged as a mutation anomaly; for interruption anomalies, the time interval in the continuous data stream is detected. If the time interval exceeds the normal collection cycle, it is judged as an interruption; for repeated data anomalies, the system identifies repeated records that appear multiple times through content comparison, especially in network jitter or equipment failure scenarios, duplicate data can be efficiently eliminated, and the pattern recognition unit can quickly and accurately locate various types of anomalies, data quality and the reliability of subsequent analysis;

[0073] After the pattern recognition unit detects an anomaly, the real-time alarm unit automatically generates an alarm message, which includes the anomaly type (sudden change, interruption or duplicate data), the time when the anomaly occurred, the device ID, the anomaly value and the recommended treatment measures. The alarm information is pushed to the user in the form of SMS, email or system notification through the system's built-in message service module, and the user end can receive and view the alarm details in real time. In addition, the alarm information can also be pushed to external systems such as monitoring platforms or command systems through the API interface, achieving automated abnormal response and linkage processing. The real-time alarm unit significantly shortens the time interval between abnormal discovery and response, effectively improving the system's operational reliability;

[0074] The dynamic threshold adjustment unit combines historical data statistics and real-time data analysis to optimize the threshold parameters of anomaly detection in real time. Combined with real-time data analysis, the system can dynamically modify the detection threshold according to the current network environment, device status and load conditions. The threshold is adapted to the actual scenario. The dynamic threshold adjustment unit effectively reduces false positives and false negatives in anomaly detection and improves the accuracy and robustness of detection. In complex and changing scenarios, the unit can adaptively adjust the detection strategy, and the stability and reliability of the system in a high-concurrency, dynamic environment.

[0075] The data indexing module includes a time indexing unit, a multi-dimensional retrieval unit, and an index maintenance unit. The time indexing unit is used to generate multi-level indexes for data based on timestamps, and the multi-level indexes are used to improve the query efficiency. The multi-dimensional retrieval unit is used to perform multi-dimensional combined queries according to conditions such as time range, device ID, and data type. The index maintenance unit is used to update the index structure in real time when data is written or updated.

[0076] Specifically, the time indexing unit generates multi-level indexes for data based on timestamps, and efficiently organizes and manages time-series data through a hierarchical structure (such as year, month, day, hour). The multi-level indexes are based on a tree structure, divide the data into multiple subsets according to the time dimension, and identify and locate on each level of the index. The multi-level indexes quickly locate the data in the target time period through layering, significantly reducing the scanning range during data retrieval. Especially in the scenario of large-scale time-series data, it can quickly filter out the data subset within the specified time range, reducing query latency.

[0077] The multi-dimensional retrieval unit supports combined queries based on multiple conditions such as time range, device ID, and data type. Users can flexibly obtain the required data by combining conditions to query specific data types of a certain device within a specific time range. By establishing a composite index structure based on multi-dimensional conditions such as B+ tree or inverted index, this unit can effectively solve the performance bottleneck in complex query scenarios and provide support for multi-dimensional data analysis. For example, users can quickly query the temperature changes of multiple devices in the same time period, or analyze the cross-time trend of specific types of data. The multi-dimensional retrieval unit supports large-scale concurrent query requirements, significantly reducing query complexity.

[0078] The index maintenance unit is used to update the index structure in real time when data is written, updated, or deleted. The index always remains consistent with the underlying data. The index maintenance adopts an incremental update strategy, that is, only the newly added or modified data part is updated, rather than reconstructing the entire index. Updating the index structure in real time avoids the problem of index invalidation, ensuring the accuracy and timeliness of queries. Even in the case of high-frequency data writing or modification, the index maintenance unit can reduce the time overhead of index maintenance by optimizing the update algorithm. For example, during the real-time writing of device data, the system can dynamically add new time index nodes without reconstructing the existing index. The index maintenance unit shows good stability and performance in high-concurrency writing scenarios, greatly improving the real-time performance of the system.

[0079] Embodiment 2:

[0080] Refer to Figure 13 , in the second embodiment of the present invention, the present invention provides a time-series database processing method based on artificial intelligence, including the following steps:

[0081] S1. Collect multi-source heterogeneous data through the data acquisition module, and perform asynchronous access and dynamic sampling;

[0082] S2. In the data processing module, optimize the sorting of out-of-order data through the skip list optimization unit, and clean, time-align, and standardize the format of the data;

[0083] S3. In the data storage module, store the data in slices, and optimize the storage efficiency through the dynamic compression algorithm;

[0084] S4. Through the query analysis module, perform multi-dimensional query and correlation analysis on the data based on the user-defined time window, and generate a visual result;

[0085] S5. In the anomaly detection module, perform anomaly detection on the data through machine learning algorithms, and generate real-time alarms;

[0086] S6. Through the data indexing module, generate a timestamp index for the data, and support multi-dimensional conditional combination queries;

[0087] S7. Utilize the dynamic modeling ability of the system to generate a correlation analysis model or a situation prediction result by combining historical data and real-time data.

[0088] Specifically, through the data acquisition module, perform asynchronous access and dynamic sampling on multi-source heterogeneous data, and combine the skip list optimization unit, data cleaning, and time alignment techniques in the data processing module to optimize the sorting and standardize the out-of-order data; in the data storage module, utilize slice storage and dynamic compression algorithms to efficiently manage the data, and at the same time ensure data security through backup and disaster recovery functions; the query analysis module supports user-defined time windows, multi-dimensional conditional combination queries, and correlation analysis, and generates visual results to intuitively display data patterns; the anomaly detection module identifies anomaly patterns such as mutations and disconnections through machine learning algorithms, and combines real-time alarms and dynamic threshold adjustment to reduce false alarms and missed alarms; the data indexing module improves the retrieval efficiency through multi-level time indexing and real-time indexing maintenance, and supports complex multi-dimensional queries. In addition, the system generates a correlation analysis model or a situation prediction by combining historical and real-time data through its dynamic modeling ability, which is widely applicable to intelligent monitoring, anomaly handling, and data analysis scenarios, significantly improving the processing efficiency, query accuracy, and system stability of time series data;

[0089] Through the full-process optimization of data acquisition, processing, storage, query analysis, anomaly detection, and index management, efficient processing, intelligent analysis, and real-time response to multi-source heterogeneous time series data are achieved. At the same time, the storage space occupancy and false alarm rate are significantly reduced, the robustness and scalability of the system are enhanced, providing reliable technical support for complex application scenarios.

[0090] Embodiment III

[0091] The third embodiment of the present invention, based on the same inventive concept, provides a computer-readable storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the time series database processing method based on artificial intelligence in the above embodiments are achieved.

[0092] Embodiment 4

[0093] The fourth embodiment of the present invention, based on the same inventive concept, provides a computer device, including: a processor and a memory; the processor and the memory communicate with each other; the memory is used to store instructions; the processor is used to execute the instructions in the memory to execute the time series database processing method based on artificial intelligence in the above embodiments.

[0094] It should be understood that various parts of the present invention can be implemented by hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one or a combination of the following well-known techniques in the art can be used: discrete logic circuits with logic gate circuits for performing logical functions on data signals, application-specific integrated circuits with appropriate combinational logic gate circuits, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), etc.

[0095] Finally, it should be noted that the above are only the preferred embodiments of the present invention and are not used to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some of the technical features. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A time series database processing system based on artificial intelligence, characterized in that It includes: a data acquisition module, a data processing module, a data storage module, a query and analysis module, an anomaly detection module, and a data indexing module; The data acquisition module is used to collect multi-source heterogeneous time-series data, supporting asynchronous access, edge computing, and dynamic sampling; The data processing module adopts a skip list structure and out-of-order optimization technology to sort and optimize out-of-order data; The data storage module is used to perform sharded storage and compression optimization through hardware acceleration, as well as data backup and disaster recovery functions; The query and analysis module is used for multi-dimensional queries with user-defined time windows and performs correlation analysis in combination with artificial intelligence algorithms; The anomaly detection module is used to detect anomaly patterns in time-series data through machine learning algorithms and generate real-time alarms; The data indexing module is used to generate indexes based on timestamps and multi-dimensional conditions for retrieval and index maintenance.

2. The time series database processing system based on artificial intelligence according to claim 1, wherein The data acquisition module includes a data access unit, an edge computing unit, and a dynamic sampling unit. The data access unit is used to access data sources that support multiple protocols, and the protocols include MQTT, HTTP, and TCP / IP protocols. The edge computing unit is used to perform denoising, aggregation, and preliminary filtering on data at the data acquisition end. The dynamic sampling unit is used to dynamically adjust the data acquisition frequency according to network bandwidth, device status, and data priority.

3. The time series database processing system based on artificial intelligence according to claim 1, characterized in that, The data processing module includes a skip list optimization unit, a data cleaning unit, and a time alignment unit. The skip list optimization unit is used to dynamically adjust the skip list index level to adapt to different data traffic and improve the speed of processing and out-of-order data. The data cleaning unit is used to identify and remove outliers, duplicate values, and fill in missing values. The time alignment unit is used to align asynchronous data from different sources based on timestamps to ensure the temporal consistency of data. The data processing includes cleaning, time alignment, and format standardization processing of the collected data.

4. The time series database processing system based on artificial intelligence according to claim 1, wherein The data storage module includes a sharded storage unit, a compressed storage unit, and a backup and disaster recovery unit. The sharded storage unit is used to shard and store data according to time windows or data source devices; The compressed storage unit is used to compress and decompress data through hardware acceleration using a dynamic compression algorithm to reduce storage space occupancy; The backup and disaster recovery unit is used for real-time backup, periodic snapshots, and disaster recovery functions.

5. The time series database processing system based on artificial intelligence according to claim 1, characterized in that, The query and analysis module includes a time window analysis unit, a correlation analysis unit, and a visualization analysis unit. The time window analysis unit is used to support fixed window, sliding window, and rolling window analysis defined by users. The correlation analysis unit is used to perform correlation analysis on multi-source data in combination with artificial intelligence algorithms and generate a dynamic situation model. The visualization analysis unit is used to generate visualization analysis results in the form of time-series curves and heat maps.

6. The time series database processing system based on artificial intelligence according to claim 1, wherein The anomaly detection module includes a pattern recognition unit, a real-time alert unit, and a dynamic threshold adjustment unit. The pattern recognition unit is used to identify anomaly patterns in data through machine learning algorithms. The anomaly patterns include mutations, data breaks, and duplicate data. The real-time alert unit is used to generate alert information when detecting anomaly data and push it to the client. The dynamic threshold adjustment unit is used to dynamically adjust the threshold of anomaly detection according to historical data and real-time analysis results to reduce false alarms and missed alarms.

7. The time series database processing system based on artificial intelligence according to claim 1, wherein The data indexing module includes a time indexing unit, a multi-dimensional retrieval unit, and an index maintenance unit. The time indexing unit is used to generate multi-level indexes for data based on timestamps. The multi-level indexes are used to improve the query efficiency. The multi-dimensional retrieval unit is used to perform multi-dimensional combination queries by time range, device ID, and data type conditions. The index maintenance unit is used to update the index structure in real time when data is written or updated.

8. A method for processing a time series database based on artificial intelligence, characterized in that, For the artificial intelligence-based time series database processing system according to any one of claims 1-7, it includes the following steps: S1. Collect multi-source heterogeneous data through the data collection module and perform asynchronous access and dynamic sampling; S2. In the data processing module, optimize the sorting of out-of-order data through the skip list optimization unit, and clean, time-align, and standardize the format of the data; S3. In the data storage module, store the data in slices and optimize the storage efficiency through a dynamic compression algorithm; S4. Through the query analysis module, perform multi-dimensional queries and correlation analysis on the data based on the user-defined time window to generate visual results; S5. In the anomaly detection module, perform anomaly detection on the data through machine learning algorithms to generate real-time alerts; S6. Generate timestamp indexes for the data through the data indexing module and support multi-dimensional conditional combination queries; S7. Utilize the dynamic modeling ability of the system to generate a correlation analysis model or a situation prediction result by combining historical data and real-time data.

9. A computer device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, When the processor executes the computer program, it achieves the artificial intelligence-based time series database processing method according to any one of claims 1 to 7.

10. A readable storage medium, characterized in that, A computer program is stored on the readable storage medium. When the computer program is executed by the processor, it achieves the artificial intelligence-based time series database processing method according to any one of claims 1 to 7.

Images