An artificial intelligence-based data monitoring method, device, equipment and medium

Abstract

The application discloses a data monitoring method and device based on artificial intelligence, equipment and medium, which can be applied to medical health and financial bank scenes, obtains the running state parameter of the first node device currently performing the data collection task in the distributed network, determines the second node device from the multiple devices of the distributed network when the running state parameter meets the preset relay triggering condition, and sends the takeover request of the data collection task to the second node device, so that the second node device performs the data collection task, forms an overlapping collection time window, obtains the first data segment and the second data segment generated by the first node device and the second node device in the overlapping collection time window, performs time axis splicing processing on the first data segment and the second data segment by using the pre-trained collection breakpoint model, and generates a continuous data stream, so that the interruption phenomenon in the data collection process is effectively avoided, the continuity of the data stream is maintained, and the accuracy of the data monitoring is improved.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and in particular to a data monitoring method, apparatus, device, and medium based on artificial intelligence. Background Technology

[0002] With societal progress and the continuous development of science and technology, people's demands for quality of life are increasing, and data monitoring has become a focus of public attention. In the field of healthcare, continuous monitoring of vital signs such as heart rate, blood oxygen, and respiratory rate using wearable devices has become an important foundation for chronic disease management and health assessment.

[0003] However, existing data monitoring methods mostly rely on single devices such as smartwatches and smartphones for data collection. In actual use, data collection is often interrupted due to factors such as device battery depletion, removal from the device, or Bluetooth connection loss. This creates data gaps on the timeline, preventing data continuity and reducing the accuracy of data monitoring. In medical scenarios, this can potentially lead to misjudgments. Therefore, effectively avoiding interruptions in data collection to prevent reduced accuracy is a pressing technical problem that needs to be solved. Summary of the Invention

[0004] This invention provides a data monitoring method, apparatus, device, and medium based on artificial intelligence to solve the technical problem that the existing technology cannot effectively avoid interruptions in the data acquisition process, thereby reducing the accuracy of data monitoring.

[0005] Firstly, an artificial intelligence-based data monitoring method is provided, including: Obtain the operating status parameters of the first node device currently executing the data acquisition task in the distributed network, wherein the distributed network includes multiple devices; When the running status parameters meet the preset relay triggering conditions, a second node device is determined from multiple devices in the distributed network, and a takeover request for the data acquisition task is sent to the second node device so that the second node device can execute the data acquisition task, forming an overlapping acquisition time window where the first node device and the second node device acquire data simultaneously. Acquire the first data segment generated by the first node device within the overlapping acquisition time window, and the second data segment generated by the second node device within the overlapping acquisition time window; The first data segment and the second data segment are spliced ​​together using a pre-trained acquisition breakpoint model to generate a continuous data stream.

[0006] Secondly, an artificial intelligence-based data monitoring device is provided, comprising: The first acquisition module is used to acquire the operating status parameters of the first node device currently performing the data acquisition task in the distributed network, wherein the distributed network includes multiple devices; The window forming module is used to determine a second node device from multiple devices in the distributed network when the running status parameters meet the preset relay triggering conditions, and send a takeover request for the data acquisition task to the second node device so that the second node device can execute the data acquisition task, thereby forming an overlapping acquisition time window in which the first node device and the second node device acquire data simultaneously. The second acquisition module is used to acquire the first data segment generated by the first node device within the overlapping acquisition time window, and the second data segment generated by the second node device within the overlapping acquisition time window. The splicing processing module is used to perform time-axis splicing processing on the first data segment and the second data segment using a pre-trained acquisition breakpoint model to generate a continuous data stream.

[0007] Thirdly, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described data monitoring method.

[0008] Fourthly, a computer-readable storage medium is provided, which stores a computer program that, when executed by a processor, implements the steps of the above-described data monitoring method.

[0009] In summary, this invention provides a data monitoring method, apparatus, device, and medium based on artificial intelligence. By acquiring the operating status parameters of a first node device currently executing a data acquisition task in a distributed network (which includes multiple devices), when the operating status parameters meet preset relay triggering conditions, a second node device is determined from the multiple devices in the distributed network, and a data acquisition task takeover request is sent to the second node device to enable it to execute the data acquisition task. This creates an overlapping acquisition time window where the first and second node devices simultaneously acquire data. The invention acquires a first data segment generated by the first node device within the overlapping acquisition time window, and a second data segment generated by the second node device within the overlapping acquisition time window. A pre-trained acquisition breakpoint model is used to perform timeline splicing processing on the first and second data segments to generate a continuous data stream. Therefore, this application effectively avoids interruptions during data acquisition by acquiring operating status parameters and dynamically triggering a task relay mechanism, synchronously acquiring dual-node data within the overlapping acquisition time window, and combining this with timeline splicing processing using an acquisition breakpoint model. This maintains the continuity of the data stream and improves the accuracy of data monitoring. Attached Figure Description

[0010] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0011] Figure 1 This is a schematic diagram of an application environment for a data monitoring method according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating a data monitoring method according to one embodiment of the present invention; Figure 3 This is a schematic diagram of the data monitoring device in one embodiment of the present invention; Figure 4 This is a schematic diagram of the structure of a computer device according to an embodiment of the present invention. Detailed Implementation

[0012] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0013] This invention provides an artificial intelligence-based data monitoring method that can be applied to, for example... Figure 1In this application environment, the client communicates with the server via a network. The server can obtain the operating status parameters of the first node device currently performing data acquisition tasks in the distributed network through the client. The distributed network includes multiple devices. When the operating status parameters meet preset relay trigger conditions, the server determines the second node device from among the multiple devices in the distributed network and sends a takeover request for the data acquisition task to the second node device, enabling it to execute the data acquisition task. This creates an overlapping acquisition time window where the first and second node devices simultaneously acquire data. The server acquires the first data segment generated by the first node device within the overlapping acquisition time window, and the second data segment generated by the second node device within the overlapping acquisition time window. A pre-trained acquisition breakpoint model is used to splice the first and second data segments along a timeline to generate a continuous data stream. For example, in the daily health monitoring of patients with chronic heart disease, the smartwatch worn by the patient acts as the first node device, continuously monitoring heart rate and blood oxygen saturation. When the patient enters an elevator or underground parking lot, the Bluetooth communication link is interrupted due to signal shielding, and the device's remaining battery power drops to a critical value, triggering automatic shutdown. In this scenario, the data acquisition task is completely interrupted during a critical time period, resulting in missing heart rate data. This makes subsequent analysis unable to accurately identify potential arrhythmia events, increasing the possibility of medical misdiagnosis. Therefore, this application can obtain the operating status parameters of the wearable medical device and dynamically trigger a task relay mechanism to synchronously acquire dual-node data within the overlapping acquisition time window. Combined with the acquisition breakpoint model, timeline splicing is performed, effectively avoiding interruptions in the data acquisition process, maintaining the continuity of the data stream, and improving the accuracy of data monitoring. The client can be, but is not limited to, various personal computers, laptops, smartphones, tablets, and portable wearable devices. The server can be implemented using a standalone server or a server cluster consisting of multiple servers.

[0014] To illustrate the technical solution of the present invention, the present invention will be described in detail below through specific embodiments.

[0015] See Figure 2 This is a flowchart illustrating a data monitoring method based on artificial intelligence provided in an embodiment of the present invention, as shown below. Figure 2 As shown, this data monitoring method can be implemented through the following steps.

[0016] S210: Obtain the operating status parameters of the first node device currently executing the data acquisition task in the distributed network, wherein the distributed network includes multiple devices.

[0017] In one implementation, the operating status parameters of the first node device currently performing data acquisition tasks in a distributed network are acquired in real time. This distributed network consists of multiple devices, which are interconnected systems that can work collaboratively to complete tasks such as data acquisition, processing, and transmission. Within this network, each device can operate independently or assign and relay tasks as needed. These devices can be smartwatches (primary acquisition devices), smartphones (auxiliary acquisition and central control devices), or other health monitoring devices such as heart rate monitors, blood pressure monitors, or temperature sensors. Real-time acquisition can be achieved in several ways. For example, one method is for the first node device to periodically report its own operating status parameters (such as battery level, signal strength, CPU utilization, etc.) to a central management server in the network. The server receives these parameters, stores and analyzes them. Another method is for the first node device to be configured with a local monitoring agent, which immediately triggers a reporting mechanism when it detects a significant change in the operating status parameters. By following the steps above, the operational status of the first node device can be continuously obtained, providing a basis for subsequent task relay and ensuring the continuity and integrity of subsequent data collection.

[0018] For example, in a medical health monitoring system, a user wears a smart bracelet as the first node device. This bracelet is responsible for collecting the user's vital signs data such as heart rate and blood oxygen in real time and uploading the data to a cloud-based distributed network. This distributed network also includes other devices such as the user's smartphone and smartwatch, all of which have data collection capabilities and can serve as potential second node devices. By monitoring the smart bracelet's operating status parameters in real time, it can determine whether the operating status parameters meet preset relay trigger conditions. Alternatively, in the financial and banking transaction field, the bank's Tiandun risk control system can be deployed on edge computing nodes across the country to collect real-time data on the fund flow and transaction behavior of a key enterprise, enabling millisecond-level assessment and interception of transaction risks and strengthening the financial security defense line.

[0019] S220: When the running status parameters meet the preset relay triggering conditions, a second node device is determined from the multiple devices in the distributed network, and a takeover request for the data acquisition task is sent to the second node device so that the second node device can execute the data acquisition task, forming an overlapping acquisition time window where the first node device and the second node device acquire data simultaneously.

[0020] In one implementation, under the current operating scenario, when the system detects that relevant operating status parameters have reached or meet preset relay trigger conditions, it will initiate the corresponding relay mechanism. This process first requires filtering and judging from multiple selectable device nodes in the connected distributed network environment to determine a qualified second node device. Subsequently, the system will automatically generate and send a detailed request to the second node device regarding the takeover of the data acquisition task, enabling the second node device to execute the data acquisition task. This creates an overlapping acquisition time window where the first and second node devices simultaneously acquire data. This overlapping acquisition time window refers to the time window formed during the data acquisition task relay process, before the first node device stops its acquisition task, during which both the first and second node devices simultaneously execute the data acquisition task. The purpose of setting this window is to ensure the continuity of data acquisition and avoid data interruption.

[0021] Furthermore, the relay trigger condition can be determined based on various strategies, which can be set according to actual conditions, and this application does not impose any limitations on this. Specifically, a simple threshold can be set, triggering the relay when the battery level of the first node device falls below a certain percentage. Alternatively, a composite condition can be set, such as triggering the relay only when the battery level falls below a threshold and the communication signal quality deteriorates to a certain extent. For example, a smart bracelet might report its remaining battery level and Bluetooth connection signal strength to the cloud every minute. When the system detects that the smart bracelet's remaining battery level is below 20%, this operating status parameter meets the preset relay trigger condition (e.g., battery level below 20% is set as the trigger condition). When determining the second node device, a device can be selected from the distributed network according to priority. Alternatively, it can be selected according to a pre-defined device list order. When sending a takeover request, a request message containing task identifiers, acquisition parameters, and other information can be sent to the selected second node device via a network communication protocol, such as TCP / IP or UDP. After receiving the request, the second node device starts its data acquisition module and begins executing the task. By following the steps above, the continuity and reliability of data collection can be effectively improved, and data loss caused by the depletion of power of a single device or communication interruption can be avoided, thereby ensuring the continuity and integrity of subsequent data monitoring processes.

[0022] S230: Obtain the first data segment generated by the first node device within the overlapping acquisition time window, and the second data segment generated by the second node device within the overlapping acquisition time window.

[0023] In one implementation, during collaborative data acquisition, it is necessary to first obtain the first data segment generated by the first node device within the overlapping acquisition time window, and simultaneously obtain the second data segment generated by the second node device within the same overlapping acquisition time window. The overlapping acquisition time window refers to a specific period of time before the first node device finally stops its data acquisition task, during which both the first and second node devices are simultaneously performing data acquisition tasks. This period of parallel acquisition activities is called the overlapping acquisition time window. Specifically, just before ceasing its acquisition task, the first node device proactively uploads the data segments it has collected and cached within the overlapping acquisition time window to a pre-configured shared storage area accessible to multiple parties via a predetermined network channel. Alternatively, it can choose to directly send them to the central management server responsible for coordination for centralized processing. Simultaneously, after the second node device successfully receives the data acquisition task takeover request issued by the system and then officially begins to execute the data acquisition task, it also needs to upload the data segments it has collected within the same overlapping acquisition time window to the same shared storage area mentioned above, or choose to send them directly to the same central management server, so as to ensure that the two overlapping data segments can be gathered in the same place for subsequent comparison or fusion processing.

[0024] For example, assuming the system sets the overlapping data acquisition time window to 5 minutes, the first node device will continue to complete the data acquisition within this 5-minute period before stopping its task, and will package and send out the data generated during this period. Simultaneously, the second node device, after starting its acquisition task, will also immediately begin acquiring data covering the same 5-minute period, generating the corresponding data segments and submitting them, thus achieving a seamless transition of acquisition responsibilities and complete data continuity. By acquiring the first data segment generated by the first node device within the overlapping data acquisition time window, and the second data segment generated by the second node device within the same overlapping data acquisition time window, the continuity and integrity of data during node switching can be effectively guaranteed, avoiding data loss or time gaps caused by task handover, ensuring the accuracy of data acquisition during task handover, and thus laying the foundation for the continuity of subsequent data monitoring processes.

[0025] S240: Use a pre-trained acquisition breakpoint model to splice the first data segment and the second data segment along the time axis to generate a continuous data stream.

[0026] In one implementation, we utilize a pre-trained acquisition breakpoint model to perform precise timeline splicing of the first and second data segments, generating a continuous and complete data stream in the time dimension. This acquisition breakpoint model is essentially a specialized data processing module. Through specific training, it can perform efficient timeline alignment and seamless splicing of data from different acquisition devices that overlap or have gaps in their time windows. Its core objective is to effectively solve the data breakage problem caused by device switching or acquisition interruptions by integrating multiple data segments to generate a continuous and smooth data stream. Specifically, the model takes the first and second data segments as input. During the splicing process, a simple timestamp-based alignment strategy can be adopted. For example, the model first identifies and extracts the end timestamp of the first data segment and the start timestamp of the second data segment. Then, based on these two key time points, it directly concatenates the second data segment sequentially after the first data segment on the timeline, forming a coherent data whole. If analysis reveals a slight time interval between two data segments, the model can compensate for this gap by interpolation or filling in blank data. Conversely, if the two data segments slightly overlap in time, the model can truncate or merge the overlapping areas based on timestamps to eliminate redundant information and ensure the accuracy and consistency of the timeline. Through these steps, even if the data acquisition task involves switching or transferring between different devices, the final integrated output data stream maintains a high degree of continuity and consistency on the timeline, thus completely avoiding data breaks or missing information, improving the accuracy of data monitoring, and providing a complete and reliable foundation for subsequent data analysis and applications.

[0027] For example, in the scenario of high-frequency trading risk control systems in finance, the distributed networks of exchanges or banks need to collect massive amounts of trading data continuously at millisecond levels to identify risks such as insider trading and abnormal fund transfers. In the scenario of ICU wards in smart hospitals, patients' vital signs data (such as heart rate, blood oxygen, respiratory rate, etc.) need to be continuously collected 24 / 7 through bedside monitors and other equipment so that the system can assess the risk of complications such as deep vein thrombosis and acute myocardial infarction in real time.

[0028] In summary, this invention provides a data monitoring method, apparatus, device, and medium based on artificial intelligence. By acquiring the operating status parameters of a first node device currently executing a data acquisition task in a distributed network (which includes multiple devices), when the operating status parameters meet preset relay triggering conditions, a second node device is determined from the multiple devices in the distributed network, and a data acquisition task takeover request is sent to the second node device to enable it to execute the data acquisition task. This creates an overlapping acquisition time window where the first and second node devices simultaneously acquire data. The invention acquires a first data segment generated by the first node device within the overlapping acquisition time window, and a second data segment generated by the second node device within the overlapping acquisition time window. A pre-trained acquisition breakpoint model is used to perform timeline splicing processing on the first and second data segments to generate a continuous data stream. Therefore, this application effectively avoids interruptions during data acquisition by acquiring operating status parameters and dynamically triggering a task relay mechanism, synchronously acquiring dual-node data within the overlapping acquisition time window, and combining this with timeline splicing processing using an acquisition breakpoint model. This maintains the continuity of the data stream and improves the accuracy of data monitoring.

[0029] In one embodiment, specifically step S220, which involves the step of "when the running state parameters meet the preset relay triggering conditions", the following steps are included: A comprehensive status score is obtained by weighted summation of the power status, communication signal quality, and device operating status of the first node device. Determine whether the overall status score is less than or equal to a preset score threshold; If the overall status score is less than or equal to the preset score threshold, then the running status parameter is determined to meet the preset relay triggering condition.

[0030] In one implementation, the operational status parameters include battery status, communication signal quality, and device operating status. Battery status refers to the remaining battery power or power supply status of the device, reflecting its endurance. This status can be obtained by reading the battery percentage data provided by the device's built-in battery management chip (BMS), or assessed by monitoring the stability of the external power supply or the presence of interruption risks. Communication signal quality refers to the signal strength and stability when the device transmits data with other nodes or a central server in a distributed network; it is a crucial factor in ensuring data transmission reliability. This quality can be assessed by monitoring metrics such as RSSI, signal-to-noise ratio (SNR), or bit error rate (BER), or by evaluating packet loss rate or retransmission count. Device operating status refers to whether the device is currently operating normally, whether there are any faults or overloads, reflecting the device's overall health and processing capacity. This status can be obtained by monitoring CPU utilization, memory usage, storage space availability, abnormal sensor operation, or error information in system logs, or by periodically reporting its health status through the device's internal self-test program.

[0031] Furthermore, a weighted summation is performed on the power status, communication signal quality, and device operating status of the first node device. Specifically, different weights are assigned to different parameters, allowing for flexible adjustment of the importance of each parameter in the overall evaluation based on actual application scenarios and task requirements. This more accurately reflects the true health status of the device, resulting in a comprehensive status score. For example, power status might have a weight of 0.4, communication signal quality 0.3, and device operating status 0.3. The standardized values ​​of each parameter are then multiplied by their corresponding weights and summed. Alternatively, a machine learning model can be used to dynamically learn and adjust the weights of each parameter based on historical data and task relay results. The comprehensive status score is then determined to be less than or equal to a preset score threshold. This preset threshold is a pre-defined value used to define whether the device's operating status has reached the critical point requiring task relay. A fixed threshold can be set based on experience or historical data analysis; this application does not impose any limitations on this. For example, a comprehensive score below 60 points triggers relay; or the threshold can be dynamically adjusted based on current network load, task importance, etc. When the overall status score falls below a preset threshold, the system can promptly and accurately determine that the operating status of the first node device no longer meets the requirements for continuously executing the data acquisition task, thereby triggering task relay. Through these steps, the limitations of relying on a single indicator are avoided, significantly improving the accuracy and scientific rigor of task relay triggering. This effectively prevents potential data loss or task interruption risks caused by improper device status assessment, ensuring the continuity and reliability of subsequent data acquisition tasks in the distributed network.

[0032] In one embodiment, that is, when the operating status parameters meet the preset relay triggering conditions, determining the second node device from multiple devices in the distributed network includes the following steps: When the operating status parameters meet the preset relay triggering conditions, the device characteristic information of multiple devices in the distributed network is obtained, wherein the device characteristic information includes device type, device remaining power and device load status; Determine whether there are target devices that meet preset access conditions based on the device type, remaining power, and load status of multiple devices. If there is a target device among the multiple devices that meets the preset admission conditions, then the comprehensive priority of each target device is calculated based on the device feature information, and the device with the highest comprehensive priority is selected as the second node device.

[0033] In one implementation, when the operating status parameters of the first node device meet the preset relay trigger conditions, to ensure that the selected second node device can reliably take over the data acquisition task, the device characteristic information of all devices in the distributed network is first obtained. This device characteristic information records key information such as the device type, remaining battery power, and load status of each device. This information is updated through dynamic status data reported in real time during device operation, or through learning and prediction using historical task execution data. Device type refers to the classification identifier of devices in the distributed network, such as sensor nodes, edge computing devices, gateway devices, servers, etc. Different types of devices typically have different processing capabilities, storage capacity, communication bandwidth, and power consumption characteristics. Device types can be pre-configured and registered in the system, or classified through an automatic identification mechanism when the device first connects to the network. Remaining battery power refers to the device's currently available battery power or power reserve. For battery-powered mobile or IoT devices, remaining battery power is a key indicator of their continuous working capability. This information can be monitored and reported in real time through the device's built-in power management module, or obtained through periodic queries of the device status interface. Device load status refers to the current resource utilization of a device, such as CPU utilization, memory usage, and network bandwidth utilization. A high load status may indicate that the device's processing capacity is nearing saturation, making it difficult to handle new data acquisition tasks. Device load status can be collected and analyzed in real time through performance monitoring interfaces provided by the operating system or device firmware.

[0034] Furthermore, the system will evaluate the device capability profiles of these devices one by one based on pre-set conditions to filter out candidate devices that meet at least one preset condition. The preset conditions refer to the filtering rules or thresholds set for capability profile parameters such as device type, remaining battery power, and load status when selecting second-node devices. These conditions can be flexibly configured according to the needs of actual application scenarios, such as requiring a specific device type, remaining battery power above a certain percentage, or CPU utilization below a certain threshold. Preset conditions can exist in the form of rule sets, logical expressions, or machine learning model outputs. For example, preset conditions might require candidate devices to have remaining battery power above a certain threshold or a load status below a certain upper limit. This can initially exclude devices that do not have basic takeover capabilities. After filtering out qualified candidate devices, the system will further comprehensively evaluate the various capability parameters of these devices and calculate the overall priority of each device. This overall priority reflects the overall suitability of the device in the current distributed network environment and can be achieved using various methods such as weighted summation, fuzzy logic reasoning, analytic hierarchy process (AHP), or multi-objective optimization algorithms. For example, different weights can be assigned to device type, remaining battery power, and load status, and then a weighted score can be calculated. The device with the highest score is selected as the device with the highest overall priority. Finally, the system will select the device with the highest overall priority from these candidate devices as the second node device and send it a data acquisition task takeover request. Through the above steps, the most suitable device can be intelligently selected to undertake the task, thereby effectively avoiding task interruption or data quality degradation caused by insufficient device capacity or poor status, ensuring the continuity and stability of data acquisition, and thus improving the accuracy of subsequent data monitoring.

[0035] In one embodiment, specifically in step S240, which involves using a pre-trained acquisition breakpoint model to perform time-axis splicing on the first data segment and the second data segment to generate a continuous data stream, the following steps are included: The end timestamp and end feature vector of the first data segment, and the start timestamp and start feature vector of the second data segment are obtained using a pre-trained acquisition breakpoint model. Calculate the similarity between the end feature vector and the start feature vector, and calculate the time deviation between the end timestamp and the start timestamp; When the similarity is greater than a preset similarity threshold and the time deviation is less than a preset deviation threshold, the data consistency check is deemed to have passed, and the second data segment is appended to the first data segment. When the similarity is greater than a preset similarity threshold and the time deviation is greater than or equal to a preset deviation threshold, the data consistency check is determined to have failed, and the second data segment is processed as abnormal data based on the data quality of the first data segment.

[0036] In one implementation, when the first node device is about to stop its data acquisition task and the second node device takes over, simultaneously acquiring data within the overlapping acquisition time window, the system first utilizes a pre-established acquisition breakpoint model to accurately obtain the end timestamp and end feature vector of the first data segment, and the start timestamp and start feature vector of the second data segment. The end and start feature vectors are high-dimensional mathematical vectors output by the pre-trained acquisition breakpoint model after performing forward propagation calculations on the last frame (end portion) of the first data segment and the first frame (start portion) of the second data segment, respectively. The end timestamp indicates the end time of acquisition for the first data segment, and the start timestamp indicates the start time of acquisition for the second data segment. The acquisition breakpoint model can predefine the structure and metadata of the data segments, thereby accurately parsing and extracting this timestamp and feature vector information. When encapsulating data segments, the acquisition breakpoint model stores the start and end timestamps of the data segment as metadata at the beginning or end of the data segment, or in an index structure associated with the data segment, for rapid retrieval. The data acquisition breakpoint model can scan data segments, extract feature vectors, and determine the end and beginning feature vectors of the data segment according to preset time field formats and sorting rules. The system calculates the similarity between these two feature vectors and the time deviation between these two timestamps. The similarity can be calculated using methods such as cosine similarity and Euclidean distance. The time deviation can be calculated directly as the absolute time difference between the two timestamps, or it can be calculated based on the data content within the overlapping acquisition time window. For example, it can compare the differences in the mean square error, correlation coefficient, or specific feature values ​​of the data values ​​at the same time point in the overlapping region of two data segments.

[0037] Furthermore, the system compares the calculated similarity with a preset similarity threshold and the calculated time deviation with a preset difference threshold. The preset similarity threshold and preset difference threshold are configurable parameters. These thresholds can be manually configured based on historical data analysis, sensor accuracy, or business needs, or they can be adaptively determined or adjusted using machine learning algorithms by learning from a large number of normal and abnormal data switching scenarios. This application does not impose any limitations on this. This comparison is the core step in data consistency verification. If the similarity is greater than the preset similarity threshold and the time deviation is less than the preset deviation threshold, it indicates that the two data segments are highly consistent in time or data content at the junction, and the data consistency verification passes. In this case, the system smoothly continues the data collected by the second node device within the overlapping acquisition time window to the first data segment, thereby generating a logically seamless continuous data stream. Specifically, all data from the second data segment can be directly appended to the end of the first data segment, and the metadata of the data stream, such as the total length and time range, can be updated. Before continuation, simple deduplication or merging operations can be performed on the data within the overlapping acquisition time window to ensure that there is no data redundancy. However, if the similarity exceeds a preset similarity threshold and the time deviation is greater than or equal to a preset deviation threshold, it indicates a significant inconsistency between the two data segments, and the data consistency check fails. In this case, the system will not simply concatenate the data; instead, it will process the second data segment for anomalies based on the data quality of the first data segment. Specifically, it can evaluate the data quality of the first data segment according to preset rules or algorithms, such as checking data integrity, the proportion of missing values, and the number of outliers. Through these steps, the integrity and availability of the data stream are ensured, preventing data breaks or inconsistencies caused by switching, and avoiding the direct mixing of potentially problematic data into the continuous data stream, thereby maintaining the overall quality of the data stream.

[0038] In one embodiment, specifically in the process of handling abnormal data in the second data segment based on the data quality of the first data segment, the following steps are included: If the data quality of the first data segment is determined to meet the preset integrity requirements, the second data segment is discarded, and interpolation or smoothing compensation is performed on the missing data in the subsequent overlapping acquisition time window based on the data characteristics of the first data segment. If the data quality of the first data segment is determined to be inconsistent with the preset integrity requirements, the second data segment is marked as abnormal data, and a breakpoint record is generated in the continuous data stream.

[0039] In one implementation, when the data collected by the first node device and the second node device within the overlapping acquisition time window differs significantly, causing data consistency verification to fail, the system first evaluates the data quality of the first data segment to determine whether it meets preset integrity requirements. The data quality of the first data segment refers to the reliability, accuracy, and completeness of the data collected by the first node device within the overlapping acquisition time window. This can include, but is not limited to, data point density, the proportion of missing values, noise level, the number of outliers, and whether the data conforms to the expected pattern or trend. For example, data quality can be evaluated based on preset statistical indicators (such as standard deviation, mean, and rate of change), or by comparing it with historical data or sensor calibration data. The preset integrity requirements refer to the minimum standards set for the data quality of the first data segment. These requirements can be specifically manifested as a data missing rate below a certain percentage, data point continuity reaching a certain threshold, or no significant abnormal fluctuations in the data within a specific time period. These requirements can be set according to actual circumstances, and this application does not impose any limitations on them. For example, it can be set that within the overlapping acquisition time window, the number of valid data points in the first data segment must reach more than 95% of the total number of data points, or the time of continuous missing data cannot exceed 5 seconds.

[0040] Furthermore, if the data quality of the first data segment meets the preset integrity requirements, it indicates that the data from the first node device before stopping the acquisition task is reliable. In this case, to avoid introducing erroneous data from the second node device that may be inconsistent with the first data segment, the system will discard the second data segment. Subsequently, to maintain the continuity of the data stream, the system will perform interpolation or smoothing compensation on the data after the overlapping acquisition time window based on the reliable data from the first data segment. Interpolation methods can include linear interpolation, polynomial interpolation, spline interpolation, etc., to fill data gaps. Smoothing compensation methods can include moving average, exponential smoothing, etc., to reduce data fluctuations and make the data stream more stable. This processing method ensures that the data stream can transition smoothly under the premise that the data from the first node device is reliable, avoiding the overall data quality degradation caused by abnormal data from the second node device. Conversely, if the data quality of the first data segment does not meet the preset integrity requirements, it means that even if there are differences between the first and second data segments, the first data segment itself may have problems and cannot be used as a reliable benchmark for interpolation or compensation. In this scenario, the system does not blindly discard or compensate for the second data segment. Instead, it marks the second data segment as anomalous and generates a breakpoint record in the continuous data stream. This marking can include adding an anomaly flag to the data record, storing the data in a specific anomalous data area, or displaying it with different colors or symbols in the data visualization interface. The breakpoint record can include the breakpoint's timestamp, breakpoint type, involved device information, and a description of possible causes of the anomaly. This approach aims to retain all possible data information while clearly informing subsequent data users that the data within that time period is anomalous or discontinuous, thus providing accurate contextual information for subsequent data analysis and decision-making, and avoiding misleading conclusions. Through these steps, the most appropriate data processing method can be flexibly selected based on the actual data quality, significantly improving data integrity and reliability during the data acquisition task relay process in a distributed network, and reducing risks caused by data inconsistency or quality issues.

[0041] In one embodiment, specifically in step S240, the breakpoint model is established as follows: Acquire breakpoint sample data generated during node switching in historical data acquisition tasks, wherein the breakpoint sample data includes data fragments before and after the breakpoint, device attribute information, and environmental status information; Based on the breakpoint sample data, a breakpoint object containing an index field, a time domain field, and an integrity field is constructed. The initial neural network model is trained under supervision using the breakpoint object until the output loss value of the initial neural network model converges, thus obtaining the acquisition breakpoint model.

[0042] In one implementation, when a node switching operation is performed, it is necessary to obtain breakpoint sample data generated by this change in historical acquisition tasks. Specifically, if a switching or transfer of acquisition nodes occurs during task execution, the sample data information at the resulting breakpoints needs to be captured and organized in a timely manner. This breakpoint sample data refers to the original data fragments collected within a certain period before and after the node switching event. It includes data fragments before and after the breakpoint, device attribute information, and environmental status information. The data fragments before and after the breakpoint are time-series data (such as sensor readings and log streams) at the last moment before the interruption and the first moment after the recovery. The device attribute information includes static characteristics such as device ID, hardware model, and firmware version at the time of the switch. The environmental status information includes dynamic environmental characteristics such as network latency, CPU / memory load, and temperature at that time. Then, based on the breakpoint sample data, a breakpoint object containing index fields, time domain fields, and integrity fields is constructed by using a standardized JSON / dictionary document or encapsulating it with a time-series database or dedicated storage model. It should be noted that this breakpoint object is a comprehensive description of a specific data acquisition breakpoint and its preceding and following data context. For example, the breakpoint object can be a record or data structure containing multiple fields.

[0043] Furthermore, the breakpoint object is mapped to a pre-defined initial neural network model. The temporal field from the breakpoint object is input into the encoding layer of the initial neural network model to obtain a hidden state vector. The index field is then concatenated and fused with the hidden state vector, and input into the prediction layer of the initial neural network model to obtain a predicted and reconstructed sequence for the data segment after the breakpoint. The integrity field is used as a weighting coefficient to calculate the difference between the predicted and reconstructed sequence and the actual sequence of the data segment after the breakpoint, resulting in a loss value. This process continues until the output loss value of the initial neural network model converges, yielding the data acquisition breakpoint model. This initial neural network model provides a unified framework for storing, querying, and analyzing all information related to data acquisition breakpoints, ensuring data consistency and operability. For example, breakpoint objects can be stored in a specific table in a database or serialized into a format conforming to a specific protocol. The breakpoint object contains an index field, a temporal field, and an integrity field. The index field uniquely identifies each breakpoint object, such as a globally unique ID. The temporal field records the time information of the data acquisition interval, such as the start timestamp, end timestamp, and duration, which is crucial for timeline concatenation. The integrity field describes the quality and integrity of the data within the data collection period, such as whether the data is complete, whether there are any anomalies, and the data quality score. This helps in data consistency verification and anomaly handling during data stitching. This step effectively avoids data loss or inconsistency caused by data breakpoints, thereby improving the model's ability to quickly and accurately stitch data along the timeline, ensuring the reliability and accuracy of data monitoring.

[0044] In one embodiment, an artificial intelligence-based data monitoring device is provided, which corresponds one-to-one with the artificial intelligence-based data monitoring method described in the above embodiments. For example... Figure 3 As shown, the data monitoring device 30 includes a first acquisition module 31, a window forming module 32, a second acquisition module 33, and a splicing processing module 34. Detailed descriptions of each functional module are as follows: The first acquisition module 31 is used to acquire the operating status parameters of the first node device currently performing the data acquisition task in the distributed network, wherein the distributed network includes multiple devices; The window forming module 32 is used to determine a second node device from multiple devices in the distributed network when the running status parameters meet the preset relay triggering conditions, and send a takeover request for the data acquisition task to the second node device so that the second node device can execute the data acquisition task, thereby forming an overlapping acquisition time window in which the first node device and the second node device acquire data simultaneously. The second acquisition module 33 is used to acquire the first data segment generated by the first node device within the overlapping acquisition time window, and the second data segment generated by the second node device within the overlapping acquisition time window. The splicing processing module 34 is used to perform time-axis splicing processing on the first data segment and the second data segment using a pre-trained acquisition breakpoint model to generate a continuous data stream.

[0045] In one embodiment, the window forming module 32 is specifically used for: The operating status parameters include power status, communication signal quality, and device operating status; A comprehensive status score is obtained by weighted summation of the power status, communication signal quality, and device operating status of the first node device. Determine whether the overall status score is less than or equal to a preset score threshold; If the overall status score is less than or equal to the preset score threshold, then the running status parameter is determined to meet the preset relay triggering condition.

[0046] In one embodiment, the window forming module 32 is further configured to: When the operating status parameters meet the preset relay triggering conditions, the device characteristic information of multiple devices in the distributed network is obtained, wherein the device characteristic information includes device type, device remaining power and device load status; Determine whether there are target devices that meet preset access conditions based on the device type, remaining power, and load status of multiple devices. If there is a target device among the multiple devices that meets the preset admission conditions, then the comprehensive priority of each target device is calculated based on the device feature information, and the device with the highest comprehensive priority is selected as the second node device.

[0047] In one embodiment, the splicing processing module 34 is specifically used for: The end timestamp and end feature vector of the first data segment, and the start timestamp and start feature vector of the second data segment are obtained using a pre-trained acquisition breakpoint model. Calculate the similarity between the end feature vector and the start feature vector, and calculate the time deviation between the end timestamp and the start timestamp; When the similarity is greater than a preset similarity threshold and the time deviation is less than a preset deviation threshold, the data consistency check is deemed to have passed, and the second data segment is appended to the first data segment. When the similarity is greater than a preset similarity threshold and the time deviation is greater than or equal to a preset deviation threshold, the data consistency check is determined to have failed, and the second data segment is processed as abnormal data based on the data quality of the first data segment.

[0048] In one embodiment, the splicing processing module 34 is further configured to: If the data quality of the first data segment is determined to meet the preset integrity requirements, the second data segment is discarded, and interpolation or smoothing compensation is performed on the missing data in the subsequent overlapping acquisition time window based on the data characteristics of the first data segment. If the data quality of the first data segment is determined to be inconsistent with the preset integrity requirements, the second data segment is marked as abnormal data, and a breakpoint record is generated in the continuous data stream.

[0049] In one embodiment, the splicing processing module 34 is further configured to: Acquire breakpoint sample data generated during node switching in historical data acquisition tasks, wherein the breakpoint sample data includes data fragments before and after the breakpoint, device attribute information, and environmental status information; Based on the breakpoint sample data, a breakpoint object containing an index field, a time domain field, and an integrity field is constructed. The initial neural network model is trained under supervision using the breakpoint object until the output loss value of the initial neural network model converges, thus obtaining the acquisition breakpoint model.

[0050] For specific limitations regarding the data monitoring device, please refer to the limitations of the data monitoring method above. Its specific functions and resulting technical effects are detailed in the method implementation section and will not be repeated here. Each module in the aforementioned data monitoring device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the operations corresponding to each module.

[0051] Figure 4 This is a schematic diagram of the structure of a computer device provided in an embodiment of the present invention. Figure 4 As shown, the computer device of this embodiment includes: at least one processor ( Figure 4 Only one is shown in the diagram), a memory, and a computer program stored in the memory and executable on at least one processor, wherein the processor executes the computer program to perform the following steps: Obtain the operating status parameters of the first node device currently executing the data acquisition task in the distributed network, wherein the distributed network includes multiple devices; When the running status parameters meet the preset relay triggering conditions, a second node device is determined from multiple devices in the distributed network, and a takeover request for the data acquisition task is sent to the second node device so that the second node device can execute the data acquisition task, forming an overlapping acquisition time window where the first node device and the second node device acquire data simultaneously. Acquire the first data segment generated by the first node device within the overlapping acquisition time window, and the second data segment generated by the second node device within the overlapping acquisition time window; The first data segment and the second data segment are spliced ​​together using a pre-trained acquisition breakpoint model to generate a continuous data stream.

[0052] This computer device may include, but is not limited to, a processor and memory. Those skilled in the art will understand that... Figure 3 The examples of computer devices are merely examples and do not constitute a limitation on computer devices. Computer devices may include more or fewer components than shown in the illustration, or combinations of certain components, or different components, such as network interfaces, displays, and input systems.

[0053] In one embodiment, a computer-readable storage medium is provided that, when the instructions in the computer-readable storage medium are executed by a processor in a computer device, enables the computer device to perform the steps of any embodiment of the data monitoring method disclosed in this invention, which will not be repeated here. The computer-readable storage medium may be non-volatile or volatile.

[0054] It should be noted that the functions or steps that can be implemented by the computer-readable storage medium or computer device described above can be referred to the relevant descriptions on the server side and client side in the foregoing method embodiments. To avoid repetition, they will not be described one by one here.

[0055] The processor referred to can be a CPU, but it can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor.

[0056] Memory includes readable storage media, internal memory, etc., wherein internal memory can be the RAM of a computer device, providing an environment for the operation of the operating system and computer-readable instructions stored in the readable storage media. The readable storage media can be the hard drive of the computer device, or in other embodiments, it can be an external storage device of the computer device, such as a plug-in hard drive, Smart Media Card (SMC), Secure Digital (SD) card, or Flash Card. Furthermore, memory can include both internal storage units and external storage devices of the computer device. Memory is used to store the operating system, cooperative applications, boot loader, data, and other programs, such as program code of computer programs. Memory can also be used to temporarily store data that has been output or will be output.

[0057] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0058] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0059] It should be noted that any AI models, software tools, or components not belonging to this company appearing in the embodiments of this application are merely illustrative examples and do not represent actual use. All user personal information involved in the embodiments of this application has been authorized (with the knowledge and consent) by the relevant parties or has been fully authorized by all parties, and the executing entity may obtain it through various legal and compliant means. The collection, storage, use, processing, transmission, provision, and disclosure of the information, data, and signals involved all comply with relevant laws and regulations and do not violate public order and good morals.

[0060] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A data monitoring method based on artificial intelligence, characterized in that, include: Obtain the operating status parameters of the first node device currently executing the data acquisition task in the distributed network, wherein the distributed network includes multiple devices; When the running status parameters meet the preset relay triggering conditions, a second node device is determined from multiple devices in the distributed network, and a takeover request for the data acquisition task is sent to the second node device so that the second node device can execute the data acquisition task, forming an overlapping acquisition time window where the first node device and the second node device acquire data simultaneously. Acquire the first data segment generated by the first node device within the overlapping acquisition time window, and the second data segment generated by the second node device within the overlapping acquisition time window; The first data segment and the second data segment are spliced ​​together on a time axis using a pre-trained acquisition breakpoint model to generate a continuous data stream.

2. The data monitoring method as described in claim 1, characterized in that, The step of using a pre-trained acquisition breakpoint model to perform time-axis splicing processing on the first data segment and the second data segment to generate a continuous data stream includes: The end timestamp and end feature vector of the first data segment, and the start timestamp and start feature vector of the second data segment are obtained using a pre-trained acquisition breakpoint model. Calculate the similarity between the end feature vector and the start feature vector, and calculate the time deviation between the end timestamp and the start timestamp; When the similarity is greater than a preset similarity threshold and the time deviation is less than a preset deviation threshold, the data consistency check is deemed to have passed, and the second data segment is appended to the first data segment. When the similarity is greater than a preset similarity threshold and the time deviation is greater than or equal to a preset deviation threshold, the data consistency check is determined to have failed, and the second data segment is processed as abnormal data based on the data quality of the first data segment.

3. The data monitoring method as described in claim 2, characterized in that, The step of processing abnormal data in the second data segment based on the data quality of the first data segment includes: If the data quality of the first data segment is determined to meet the preset integrity requirements, the second data segment is discarded, and interpolation or smoothing compensation is performed on the missing data in the subsequent overlapping acquisition time window based on the data characteristics of the first data segment. If the data quality of the first data segment is determined to be inconsistent with the preset integrity requirements, the second data segment is marked as abnormal data, and a breakpoint record is generated in the continuous data stream.

4. The data monitoring method as described in claim 1, characterized in that, When the operating status parameters meet the preset relay triggering conditions, determining the second node device from multiple devices in the distributed network includes: When the operating status parameters meet the preset relay triggering conditions, the device characteristic information of multiple devices in the distributed network is obtained, wherein the device characteristic information includes device type, device remaining power and device load status; Determine whether there are target devices that meet preset access conditions based on the device type, remaining power, and load status of multiple devices. If there is a target device among the multiple devices that meets the preset admission conditions, then the comprehensive priority of each target device is calculated based on the device feature information, and the device with the highest comprehensive priority is selected as the second node device.

5. The data monitoring method as described in claim 1, characterized in that, The operating status parameters include battery status, communication signal quality, and device operating status. When the operating status parameters meet the preset relay trigger conditions, the following conditions apply: A comprehensive status score is obtained by weighted summation of the power status, communication signal quality, and device operating status of the first node device. Determine whether the overall status score is less than or equal to a preset score threshold; If the overall status score is less than or equal to the preset score threshold, then the running status parameter is determined to meet the preset relay triggering condition.

6. The data monitoring method as described in claim 1, characterized in that, The data acquisition breakpoint model is established in the following way: Acquire breakpoint sample data generated during node switching in historical data acquisition tasks, wherein the breakpoint sample data includes data fragments before and after the breakpoint, device attribute information, and environmental status information; Based on the breakpoint sample data, a breakpoint object containing an index field, a time domain field, and an integrity field is constructed. The initial neural network model is trained under supervision using the breakpoint object until the output loss value of the initial neural network model converges, thus obtaining the acquisition breakpoint model.

7. A data monitoring device based on artificial intelligence, characterized in that, include: The first acquisition module is used to acquire the operating status parameters of the first node device currently performing the data acquisition task in the distributed network, wherein the distributed network includes multiple devices; The window forming module is used to determine a second node device from multiple devices in the distributed network when the running status parameters meet the preset relay triggering conditions, and send a takeover request for the data acquisition task to the second node device so that the second node device can execute the data acquisition task, thereby forming an overlapping acquisition time window in which the first node device and the second node device acquire data simultaneously. The second acquisition module is used to acquire the first data segment generated by the first node device within the overlapping acquisition time window, and the second data segment generated by the second node device within the overlapping acquisition time window. The splicing processing module is used to perform time-axis splicing processing on the first data segment and the second data segment using a pre-trained acquisition breakpoint model to generate a continuous data stream.

8. The data monitoring device as described in claim 7, characterized in that, The splicing processing module is specifically used for: The end timestamp and end feature vector of the first data segment, and the start timestamp and start feature vector of the second data segment are obtained using a pre-trained acquisition breakpoint model. Calculate the similarity between the end feature vector and the start feature vector, and calculate the time deviation between the end timestamp and the start timestamp; When the similarity is greater than a preset similarity threshold and the time deviation is less than a preset deviation threshold, the data consistency check is deemed to have passed, and the second data segment is appended to the first data segment. When the similarity is greater than a preset similarity threshold and the time deviation is greater than or equal to a preset deviation threshold, the data consistency check is determined to have failed, and the second data segment is processed as abnormal data based on the data quality of the first data segment.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the data monitoring method as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the data monitoring method as described in any one of claims 1 to 7.

Images