A method, system, equipment, and medium for intelligent monitoring of equipment operating status based on multidimensional data asynchronous compensation.

Abstract

This invention relates to the field of equipment management technology, specifically a method, system, device, and medium for intelligent monitoring of equipment operating status based on multi-dimensional data asynchronous compensation. The method includes receiving and parsing monitoring requests for the operating status of a target device; extracting physical features reflecting fluctuations in the target device's operation; generating a feature fingerprint representing the current operating condition of the target device when the fluctuation rate exceeds a preset device sensitivity threshold; matching the corresponding static device ledger information; calculating the time difference offset between the trigger time of the feature fingerprint and the sampling record time of the target static device ledger and performing asynchronous compensation alignment; and calculating the operating deviation index of the target device. This solves the memory bloat problem caused by full indexing of time-series data, improves data alignment accuracy, and reduces the response time for queries involving tens of thousands of devices to less than 0.8 seconds.

Description

Technical Field

[0001] This invention relates to the field of equipment management technology, and in particular to an intelligent monitoring method, system, equipment, and medium for equipment operation status based on multi-dimensional data asynchronous compensation. Background Technology

[0002] With the deepening of the industrial internet in the power sector, regional power generation groups have generally established digital monitoring systems covering multiple energy types such as thermal power, hydropower, wind power, and photovoltaic power. These systems generate hundreds of thousands of timestamped equipment operation sequence data points per second, while also accumulating massive amounts of structured operational data such as equipment ledgers, maintenance records, and financial and marketing records. The system typically consists of a data source layer, a platform layer, and an application layer. However, in the process of refined management of massive amounts of equipment, including operation monitoring, intelligent scheduling, fault early warning, and operational decision optimization, the fusion and querying of massive amounts of time-series data (stored in time-series databases) from the data source layer and relational data (stored in relational databases) at the platform layer faces significant challenges. These two types of data are stored in time-series databases and relational databases with vastly different architectures, respectively. Intelligent equipment management and business monitoring must rely on cross-database joint queries to achieve data fusion and analysis.

[0003] Currently, mainstream heterogeneous data joint query and device management solutions suffer from three major insurmountable technical bottlenecks, severely restricting the efficiency and accuracy of business management: First, the application-layer database sharding query merging mode requires writing multiple sets of database adaptation logic, resulting in extremely high development and maintenance costs. Furthermore, in scenarios involving tens of thousands of measurement points, nested loop join operations can cause response latency to spike to 3-5 seconds, failing to meet the management requirement of within 2 seconds for real-time device operation monitoring. Second, if a full data caching index is used, the continuous high-frequency updates of time-series data will lead to a rapid increase in memory, and the cached index will quickly become dirty, failing to guarantee data consistency and thus causing deviations in business-level management decisions. If the cached index is abandoned, the cross-database query performance problem cannot be solved at its root.

[0004] More importantly, existing solutions cannot balance query real-time performance with data consistency: they either sacrifice real-time performance by waiting for full data retrieval and alignment, or sacrifice accuracy by using cached approximations. This fails to meet the requirements of both rapid response and accurate data in business scenarios such as equipment accident tracing and management, and calculation of business decision indicators, and is very likely to cause blind spots in equipment supervision or management lag. Summary of the Invention

[0005] In this section, as well as in the abstract and title of this application, some simplifications or omissions may be made to avoid obscuring the purpose of this section, the abstract, and the title of this application, and such simplifications or omissions shall not be used to limit the scope of the invention.

[0006] To address the slow response issue in existing technologies, one objective of this invention is to provide an intelligent monitoring method for equipment operation status based on multi-dimensional data asynchronous compensation. This method includes receiving and parsing monitoring requests for the operation status of a target device, obtaining static equipment ledger information for the target device; real-time acquisition of multi-source runtime sequence data streams of the target device, extracting physical features reflecting operational fluctuations of the target device, and generating a feature fingerprint representing the current operating condition of the target device when the fluctuation rate exceeds a preset equipment sensitivity threshold; matching the corresponding static equipment ledger information in a heterogeneous association index based on the feature fingerprint, calculating the time difference offset between the trigger time of the feature fingerprint and the sampling record time of the target static equipment ledger, and performing asynchronous compensation and alignment to generate a panoramic equipment ledger data matching time and attributes; inputting the panoramic equipment ledger data into a preset operation early warning rule base to calculate the operation deviation index of the target device; and generating a monitoring early warning work order containing the static equipment ledger information of the target device and the dynamic fault period when the operation deviation index exceeds a safety threshold, and dispatching it to the operation and maintenance scheduling system.

[0007] As a preferred embodiment of the intelligent monitoring method for equipment operation status based on multidimensional data asynchronous compensation of the present invention, it further includes: generating a global service identifier for each access power measurement point, and establishing a mapping relationship between the global service identifier and the unique code of the equipment in the static equipment ledger in the heterogeneous association index.

[0008] As a preferred embodiment of the intelligent monitoring method for equipment operation status based on multi-dimensional data asynchronous compensation of the present invention, the step of extracting physical characteristics reflecting the fluctuation of target equipment operation includes: calculating the instantaneous change rate of each monitoring parameter in the multi-source runtime sequence data stream in real time; when the instantaneous change rate exceeds the preset equipment sensitivity threshold, extracting a time-series sampling segment centered on the current moment as a feature fingerprint reflecting the real-time operating condition of the equipment.

[0009] As a preferred embodiment of the intelligent monitoring method for equipment operation status based on multidimensional data asynchronous compensation of the present invention, the step of generating equipment panoramic ledger data that matches time and attributes includes: performing business logic divide-and-conquer parsing on the operation status monitoring request, identifying whether the monitoring dimension involved in the monitoring request is a dynamic operation dimension or a static asset dimension; based on the identification result, parallelly searching heterogeneous association indexes and relational databases, extracting dynamic historical operating condition data associated with feature fingerprints and corresponding static equipment parameters respectively, and performing multidimensional data aggregation.

[0010] As a preferred embodiment of the intelligent monitoring method for equipment operation status based on multi-dimensional data asynchronous compensation of the present invention, the step of calculating the time difference offset between the trigger time of the feature fingerprint and the sampling record time of the target static equipment ledger and performing asynchronous compensation and alignment includes: locating the anchor point of the corresponding feature fingerprint in the heterogeneous association index, and calculating the time difference offset between the trigger time of the feature fingerprint and the sampling record time of the target static equipment ledger; determining whether the time difference offset exceeds a preset consistency tolerance threshold; selecting to execute either the incremental pull mode for the underlying database or the interpolation completion mode based on the feature fingerprint according to the determination result; if it exceeds the threshold, executing the incremental pull mode and initiating a pull request for a specific missing segment to the underlying time series database; if it does not exceed the threshold, executing the interpolation completion mode and performing interpolation processing using the feature fingerprint.

[0011] Another objective of this invention is to provide an intelligent monitoring system for equipment operation status based on multi-dimensional data asynchronous compensation, comprising: a request parsing module for receiving and parsing monitoring requests for the operation status of a target device and obtaining static equipment ledger information of the target device; a working condition identification module for real-time acquisition of multi-source runtime sequence data streams of the target device and extraction of physical features reflecting the operational fluctuations of the target device, generating a feature fingerprint representing the current operating condition of the device when the fluctuation rate exceeds a preset equipment sensitivity threshold; a data fusion module for matching the corresponding static equipment ledger information in a heterogeneous association index based on the feature fingerprint, calculating the time difference offset between the trigger time of the feature fingerprint and the sampling record time of the target static equipment ledger and performing asynchronous compensation alignment to generate equipment panoramic ledger data that matches time and attributes; and a monitoring decision module for inputting the equipment panoramic ledger data into a preset operation early warning rule base, calculating the operation deviation index of the target device, and generating a monitoring early warning work order containing equipment static ledger information and dynamic fault time periods when the operation deviation index exceeds a safety threshold, and dispatching it to the operation and maintenance scheduling system.

[0012] As a preferred embodiment of the intelligent monitoring system for equipment operation status based on multi-dimensional data asynchronous compensation of the present invention, it further includes: a closed-loop feedback optimization module, used to receive the work order execution status and on-site measured deviation data fed back by the operation and maintenance scheduling system, and dynamically correct the safety threshold in the operation early warning rule base according to the on-site measured deviation data.

[0013] Another object of the present invention is to provide a computer device, a memory for storing a computer program, and a processor for executing the computer program to implement the above-described intelligent monitoring method for device operating status based on multi-dimensional data asynchronous compensation.

[0014] Another object of the present invention is to provide a computer-readable storage medium having a computer program stored thereon, characterized in that the program, when executed by a processor, implements the above-described intelligent monitoring method for device operating status based on multidimensional data asynchronous compensation.

[0015] Another object of the present invention is to provide a computer program product, including a computer program, characterized in that, when the program is executed by a processor, it implements the above-mentioned intelligent monitoring method for device operating status based on multi-dimensional data asynchronous compensation.

[0016] The beneficial effects of this invention are as follows: This invention utilizes a non-full feature fingerprint generation mechanism triggered by fluctuation characteristics to build an index only for critical periods when device operational fluctuations exceed a threshold. This ensures that the cache index size is only related to the frequency of abnormal device fluctuations, rather than linearly related to sampling duration or the number of measurement points. This reduces memory bandwidth usage during stable system operation and solves the memory bloat problem caused by full indexing of time-series data. Furthermore, through an asynchronous compensation alignment mechanism based on time difference offset, precise timeline alignment between time-series streaming data and static relational data is achieved, improving data alignment accuracy and reducing the response time for querying associations of tens of thousands of devices from 3-5 seconds to less than 0.8 seconds. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying 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.

[0018] Figure 1 This is a flowchart (I) of the intelligent monitoring method for equipment operation status based on multi-dimensional data asynchronous compensation according to the present invention.

[0019] Figure 2 This is a flowchart (II) of the intelligent monitoring method for equipment operating status based on multi-dimensional data asynchronous compensation according to the present invention.

[0020] Figure 3 This is a schematic diagram of a sub-process of the intelligent monitoring method for equipment operating status based on multi-dimensional data asynchronous compensation according to the present invention (I).

[0021] Figure 4 This is a schematic diagram of a sub-process of the intelligent monitoring method for equipment operating status based on multi-dimensional data asynchronous compensation according to the present invention (II).

[0022] Figure 5 This is a framework diagram of the intelligent monitoring system for equipment operation status based on multi-dimensional data asynchronous compensation, as described in this invention.

[0023] In the diagram, 100 is the request parsing module; 200 is the working condition identification module; 300 is the data fusion module; 400 is the regulatory decision-making module; and 500 is the closed-loop feedback optimization module. Detailed Implementation

[0024] To make the objectives, features, and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0025] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0026] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0027] Example 1, see Figure 1 and Figure 2 This is the first embodiment of the present invention, which provides an intelligent monitoring method for equipment operating status based on multi-dimensional data asynchronous compensation. The specific implementation steps are as follows: Step S101: Receive and parse the operation status monitoring request for the target device, and obtain the static equipment ledger information of the target device.

[0028] Target equipment refers to various types of equipment in the power system that require monitoring of their operational status, including but not limited to boilers, steam turbines, and generators of thermal power units, turbines and generators of hydropower units, wind turbines and converters of wind power units, and inverters and combiner boxes of photovoltaic power plants.

[0029] An operational status monitoring request refers to a request initiated by a user through the visual interface or API of an industrial internet platform to query the operational status of a target device. The request includes information such as the unique identifier of the target device, the query time range, and monitoring indicators.

[0030] Static equipment ledger information refers to basic equipment information stored in a relational database that does not change frequently over time, including equipment model, rated parameters, commissioning time, installation location, manufacturer, maintenance records, spare parts information, etc.

[0031] By uniformly receiving and parsing operational status monitoring requests, standardized processing of monitoring requests from different sources and in different formats is achieved, providing a foundation for subsequent data fusion and analysis; at the same time, static equipment ledger information of target devices is directly obtained, avoiding repeated queries to relational databases and improving processing efficiency.

[0032] Step S102: Collect multi-source runtime sequence data streams of the target device in real time, extract physical characteristics that reflect the fluctuations in the operation of the target device, and generate a feature fingerprint representing the current operating condition of the target device when the fluctuation rate exceeds the preset device sensitivity threshold.

[0033] Multi-source runtime sequence data streams refer to numerical sequences with precise timestamps that are continuously collected from various physical measurement points of the target device. The sampling frequency is usually from 1Hz to 100Hz, and includes various physical parameters such as voltage, current, temperature, pressure, speed, vibration, and flow rate.

[0034] Physical characteristics refer to characteristic parameters that can reflect changes in the operating status of the target equipment, including but not limited to the instantaneous value, average value, maximum value, minimum value, rate of change, and variance of the measurement point values.

[0035] The fluctuation rate refers to the range of change of the measured value per unit time, and is used to measure the degree of fluctuation in the operating status of equipment.

[0036] Equipment sensitivity threshold refers to a pre-set critical value used to determine whether abnormal fluctuations occur in the operating status of equipment. Different types of equipment and measuring points correspond to different sensitivity thresholds.

[0037] Method for determining the equipment sensitivity threshold: The equipment sensitivity threshold is calculated based on the equipment's rated parameters and safety factor. The calculation formula is: S = k × (R / T), where R is the rated parameter value of the corresponding measuring point, T is the time base (1 second), and k is the safety factor. The safety factor values ​​for different types of measuring points are: 0.05 for voltage and current measuring points, 0.03 for temperature measuring points, 0.02 for pressure measuring points, and 0.01 for speed measuring points.

[0038] Feature fingerprints refer to a segment of time-series sampled data centered on the current moment when the device experiences abnormal fluctuations in operation. This data segment contains time window information, core feature values, and timestamps, and can uniquely represent the device's operating condition at that moment.

[0039] The feature fingerprint is stored in a structured data format and consists of the following interrelated fields, all of which are required: Device Identity Field: Stores the unique device code of the target device, used to uniquely identify the device to which this feature fingerprint belongs. Measurement Point Identity Field: Stores the global business identifier of the corresponding power measurement point, using a 16-bit fixed-length string format, used to uniquely identify the physical measurement point that generated this feature fingerprint. Time Base Field: Contains three timestamps, all using millisecond-level Unix time format. Anchor Time: The center time point of the feature fingerprint data segment, serving as the time alignment benchmark. Window Start Time: The earliest time point of the time-series sampling segment. Window End Time: The latest time point of the time-series sampling segment. Sampled Data Field: Stores a continuous sequence of sampled point values ​​centered at the anchor time; the number of sampled points corresponds to the measurement point sampling frequency and window length. Feature Extraction Field: Stores the core statistical features calculated from the sampled data, including the maximum, minimum, arithmetic mean, instantaneous rate of change, and overall slope within the sampled segment. Generation Time Field: Stores the timestamp of the feature fingerprint generated in the system, used for index updates and data validity checks. For example, the feature fingerprint generated by the main steam temperature measuring point of a thermal power unit has the equipment identity field "FD01-GT-0001", the measuring point identity field "0101-0203-00000001", the anchor time is, for example, 10:00:00 on May 12, 2024, the window length is 1 second, it contains 100 sampling points, and the core features include a maximum value of 541.8℃, a minimum value of 531.9℃, and an average value of 536.2℃.

[0040] By collecting multi-source runtime sequence data streams in real time, the real-time operating status of the target device can be fully grasped; feature fingerprints are generated only when the fluctuation rate exceeds the device sensitivity threshold, reducing the amount of data that needs to be processed and stored, and avoiding the problem of excessive system load caused by full data processing.

[0041] Step S103: Based on the feature fingerprint, match the corresponding static equipment ledger information in the heterogeneous association index, calculate the time difference offset between the trigger time of the feature fingerprint and the sampling record time of the target static equipment ledger, and perform asynchronous compensation and alignment to generate equipment panoramic ledger data that matches time and attributes.

[0042] Heterogeneous relational indexes refer to index tables stored in the server's high-speed cache area, which pre-record the mapping relationship between each power measurement point and the corresponding equipment ledger primary key, and are the core foundation for realizing fast cross-database data association.

[0043] The heterogeneous association index is implemented using a distributed high-speed caching architecture, with an overall key-value pair storage structure. The index key is a globally unique business identifier for the power measurement point, unique throughout the entire system. The index value is a structured index entry containing four core pieces of information: a unique device code, a unique identifier for the latest feature fingerprint, the last update time of the index entry, and the measurement point's operating status. The index employs an asynchronous update mechanism: when a new feature fingerprint is generated, an update notification is sent to a dedicated index update service via an internal message passing mechanism. The index update service executes all update operations using a single-threaded serial processing method, fundamentally avoiding concurrent write conflicts. Simultaneously, the system sets an automatic expiration time of 7 days for each index entry; offline measurement point entries that have not been updated for more than 7 days will be automatically cleaned up, freeing up cache space.

[0044] The time difference offset refers to the difference between the timestamp of the feature fingerprint and the recording timestamp of the static equipment ledger information, in milliseconds.

[0045] Asynchronous compensation alignment refers to adjusting the time axis of time-series data and static data based on the time difference offset, eliminating the time misalignment between the two and ensuring data consistency in the time dimension. Specifically, the trigger time of the feature fingerprint is set to Ta, and the times of its two nearest ledger record points are T1 and T2, with corresponding static parameter values ​​P1 and P2. The formula for calculating the compensated alignment value Palign is: Palign = P1 + (P2 - P1) × (Ta - T1) / (T2 - T1). Through linear compensation, the analysis error caused by inconsistent sampling periods is eliminated.

[0046] Equipment panoramic ledger data refers to the complete data generated by integrating the static equipment ledger information and dynamic operation data of the target equipment. It includes all basic information and real-time operation information of the equipment and can comprehensively reflect the operating status of the equipment.

[0047] Heterogeneous relational indexes enable rapid association between time-series and static data, avoiding time-consuming cross-database table join operations; asynchronous compensation alignment resolves time misalignment issues caused by inconsistent update frequencies between time-series and static data, ensuring the accuracy and consistency of equipment overview ledger data.

[0048] Step S104: Input the equipment panoramic ledger data into the preset operation early warning rule library, calculate the operation deviation index of the target equipment; when the operation deviation index exceeds the safety threshold, generate a supervision and early warning work order containing the static equipment ledger information of the target equipment and the dynamic fault period, and send it to the operation and maintenance scheduling system.

[0049] The operation early warning rule base refers to a pre-stored set of rules used to determine whether the operating status of equipment is abnormal. It includes information such as the normal operating range, fault judgment criteria, and early warning levels of various types of equipment.

[0050] Operating deviation index refers to the difference between the actual operating parameters of equipment and the rated parameters or normal operating range, and is used to measure the degree to which the operating status of equipment deviates from the normal state.

[0051] A safety threshold is a pre-set critical value used to determine whether the operating status of equipment is safe. When the operating deviation index exceeds the safety threshold, the equipment is deemed to have a risk of failure.

[0052] A monitoring and early warning work order is an electronic work order used to notify maintenance personnel to handle equipment failures. It includes information such as the faulty equipment, the time of failure, the fault symptoms, the warning level, and handling suggestions.

[0053] The operation and maintenance dispatch system refers to the management system used by power companies to uniformly dispatch operation and maintenance personnel and allocate operation and maintenance tasks.

[0054] By automatically calculating operational deviation indicators using a rule base for early warning, automated identification and early warning of equipment failures are achieved. By generating and dispatching monitoring and early warning work orders, rapid response and closed-loop processing of equipment failures are realized, improving operation and maintenance efficiency and equipment reliability.

[0055] Example 2, see Figures 1-4 This is the second embodiment of the present invention. This embodiment is a preferred implementation based on embodiment 1.

[0056] Specifically, this also includes generating a global service identifier for each power measurement point that is connected, and establishing a mapping relationship between the global service identifier and the unique device code in the static device ledger in the heterogeneous association index.

[0057] A power measurement point refers to a sensor or measuring device installed on a target device to collect specific physical parameters. Each power measurement point corresponds to a unique physical address.

[0058] The global business identifier is a unique identity generated for each power measurement point within the entire power industrial internet platform. It adopts a 16-bit three-segment coding rule of "power plant code (4 digits), equipment type code (4 digits), and measurement point channel code (8 digits)," and is permanently bound to the physical address of the power measurement point, and does not change with changes in the data acquisition channel or the migration of the storage location.

[0059] The unique equipment code refers to the code used to uniquely identify the target equipment in the static equipment ledger information.

[0060] The mapping relationship refers to the one-to-one correspondence between global business identifiers and device unique codes, which is stored in the heterogeneous association index.

[0061] By generating a global business identifier, a unified identity is provided for power measurement points across systems and databases, avoiding association errors caused by inconsistent identifiers between different systems. By establishing a mapping relationship, the association between power measurement points and target equipment is realized, eliminating the need to perform complex cross-database table join operations during queries, further reducing query time.

[0062] Preferably, the step of extracting physical characteristics reflecting the fluctuations in the operation of the target equipment includes: calculating the instantaneous change rate of each monitoring parameter in the multi-source runtime sequence data stream in real time; when the instantaneous change rate exceeds the preset equipment sensitivity threshold, extracting a time-series sampling segment centered on the current moment as a feature fingerprint reflecting the real-time operating condition of the equipment.

[0063] The instantaneous rate of change is calculated using a sliding time window. The duration of the sliding time window is configured according to the sampling frequency of the measurement point. For example, for a measurement point sampled at 100Hz, the window duration is set to 100ms, and the sliding step size is set to 20ms. The formula for calculating the instantaneous rate of change is: Instantaneous rate of change = |Current measurement point value - Window start measurement point value| / (Current timestamp - Window start timestamp).

[0064] The sensitivity threshold of the equipment is pre-configured according to the equipment type and the physical characteristics of the measuring point. For example, the voltage measuring point is ±5% / s of the rated value, the temperature measuring point is ±3℃ / s, the pressure measuring point is ±2% / s, and the speed measuring point is ±10r / min / s.

[0065] The length of the timing sampling segment is pre-configured according to the device type and measurement point characteristics. It is usually a time period centered on the current time, with 500ms before and after, for a total duration of 1 second.

[0066] By quantifying and calculating the instantaneous rate of change, the system achieves automated and accurate identification of abnormal equipment fluctuations. Only time-series sampling segments of abnormal fluctuations are extracted as feature fingerprints to further compress the amount of indexed data, while ensuring that the abnormal operating conditions most relevant to the power scenario are fully indexed.

[0067] Preferably, the steps for generating equipment panoramic ledger data that matches time and attributes include: performing business logic divide-and-conquer parsing on the operation status supervision request, identifying whether the supervision dimension involved in the supervision request is a dynamic operation dimension or a static asset dimension; based on the identification results, parallel searching of heterogeneous association indexes and relational databases, extracting dynamic historical operating condition data associated with feature fingerprints and corresponding static equipment parameters respectively, and performing multi-dimensional data aggregation.

[0068] The business logic divide-and-conquer approach follows a standardized process: The input operational status monitoring request is structured and parsed, extracting all query fields and filtering conditions. Data is dimensionally divided based on its inherent attributes and access characteristics: Dynamic operation dimension: Contains all numerical data that changes continuously over time; its access characteristics typically include time range limitations or require statistical aggregation calculations, such as operational parameters like equipment temperature, pressure, and current. Static asset dimension: Contains all attribute data that does not change frequently over time; its access characteristics typically involve equality matching or enumeration filtering, such as ledger information like equipment model, installation location, and commissioning time. The original monitoring request is split into two independent sub-query tasks: a dynamic data sub-query and a static data sub-query. These two sub-query tasks are executed in parallel; the dynamic data sub-query points to a heterogeneous relational index, while the static data sub-query points to a relational database. The results of the two sub-queries are matched using the equipment's unique code, and finally aggregated to generate a unified query result.

[0069] Business logic divide-and-conquer parsing refers to using a built-in SQL semantic parser to perform abstract syntax tree parsing on runtime status monitoring requests, extracting the source, data type, and access characteristics of query fields.

[0070] Dynamic operation dimension refers to the regulatory dimension related to the real-time operating status of the equipment, which corresponds to high-flow data. Its access characteristics include time range filtering and aggregation calculation.

[0071] The static asset dimension refers to the regulatory dimension related to basic equipment information. It corresponds to strongly static data, and its access characteristics include attribute condition filtering and enumeration matching.

[0072] Relational databases are databases used to store static equipment ledger information, such as MySQL, Oracle, and MPP databases.

[0073] Dynamic historical operating data refers to the historical operating data of devices that are stored in a heterogeneous association index and associated with feature fingerprints.

[0074] Static device parameters refer to the basic parameter information of the target device stored in a relational database.

[0075] Multidimensional data aggregation refers to linking and integrating dynamic historical operating condition data and static equipment parameters according to the unique equipment code to generate unified panoramic equipment ledger data.

[0076] By using business logic divide-and-conquer parsing, the system can intelligently split regulatory requests and allocate different types of query tasks to the most suitable database for execution. By using parallel retrieval of heterogeneous relational indexes and relational databases, the traditional serial query time can be reduced by more than 50%, thus improving query response speed.

[0077] Furthermore, the step of calculating the time difference offset between the trigger time of the feature fingerprint and the sampling record time of the target static equipment ledger and performing asynchronous compensation and alignment includes: locating the anchor point of the corresponding feature fingerprint in the heterogeneous association index and calculating the time difference offset between the trigger time of the feature fingerprint and the sampling record time of the target static equipment ledger; determining whether the time difference offset exceeds a preset consistency tolerance threshold; selecting either the incremental pull mode for the underlying time series database or the interpolation completion mode based on the feature fingerprint based on the determination result; if it exceeds the threshold, executing the incremental pull mode and initiating a pull request for a specific missing segment to the underlying time series database; if it does not exceed the threshold, executing the interpolation completion mode and performing interpolation processing using the feature fingerprint.

[0078] The anchor point of a feature fingerprint refers to the center timestamp of a feature fingerprint data segment.

[0079] The consistency tolerance threshold is pre-configured based on the business scenario: 100 milliseconds for real-time monitoring, 10 milliseconds for incident tracing, and 1 second for operational statistics. The threshold is determined according to the different real-time and accuracy requirements of each business scenario: Real-time monitoring: Prioritize response speed, with a tolerance threshold of 100 milliseconds. Incident tracing: Prioritize data accuracy, with a tolerance threshold of 10 milliseconds. Operational statistics: Prioritize data integrity, with a tolerance threshold of 1 second.

[0080] Underlying time-series databases refer to databases used to store raw, multi-source runtime time-series data streams, such as InfluxDB and TimescaleDB.

[0081] Incremental fetch mode refers to fetching only the original data of the specific missing time segment corresponding to the time difference offset from the underlying time series database, rather than querying the entire data.

[0082] Interpolation completion mode refers to using the extreme points and start and end points in the feature fingerprint, and employing linear interpolation, Lagrange interpolation or spline interpolation algorithms to calculate the measurement point values ​​corresponding to the target timestamp, thereby completing the time axis alignment.

[0083] The interpolation completion mode automatically matches the optimal interpolation algorithm according to the overall slope feature of the feature fingerprint. The overall slope (unit: corresponding physical quantity / second), and the specific rules are as follows: When the absolute value of the overall slope of the feature fingerprint ≤ 0.1, it is determined that the device is in a stable operation state, and the linear interpolation algorithm is used. When 0.1 < absolute value of the overall slope ≤ 1.0, it is determined that the device is in a slow change state, and the quadratic Lagrange interpolation algorithm is used. When the absolute value of the overall slope > 1.0, it is determined that the device is in a severe fluctuation state, and the cubic spline interpolation algorithm is used. The specific calculation process of linear interpolation is as follows: Given that the times of two adjacent sampling points in the feature fingerprint are t1 and t2 respectively, and the corresponding values are v1 and v2 respectively, and the target time t satisfies t1 < t < t2, then the calculation formula for the value v(t) corresponding to the target time is: v(t)=v1+(v2 - v1) / (t2 - t1)×(t - t1). For example: Given that the times of two adjacent sampling points are 10:00:00.000 and 10:00:00.010 respectively, and the corresponding temperature values are 536.2°C and 536.5°C respectively, then the interpolation result at 10:00:00.005 is 536.35°C.

[0084] Through the hierarchical compensation mechanism, on the premise that the data accuracy meets the business requirements, cache interpolation completion is preferred to maximize the query speed; only perform precise incremental pulling when necessary to ensure data accuracy, achieving a dynamic balance between query efficiency and data accuracy, and can adapt to the differentiated requirements of different power business scenarios.

[0085] This experiment was carried out in a standard industrial Internet platform environment, and the experimental configuration is as follows: Hardware environment: 2 32-core high-performance servers, 256GB of memory, and 4TB of high-speed solid-state drives. Data environment: Simulate 10,000 power measurement points, with a sampling frequency of 10Hz, and time-series data continuously running for 7 days. Comparison scheme: The industry-wide application layer sub-library query and merging scheme, and the experimental results are shown in Table 1.

[0086] Table 1:

[0087] Example 3, refer to Figures 1-5 , an intelligent supervision system for device operation status based on multi-dimensional data asynchronous compensation.

[0088] Specifically, an intelligent supervision system for device operation status based on multi-dimensional data asynchronous compensation includes: a request parsing module, which is used to receive and parse the operation status supervision request for the target device to obtain the static device ledger information of the target device. The request parsing module is deployed in the application layer of the industrial Internet platform, communicates with the user interface and the API gateway, and is built with a request parser and a parameter validator, and can perform standardized parsing on supervision requests in different formats and verify the legality of the request parameters.

[0089] This enables unified access and standardized processing of regulatory requests, providing standardized input data for subsequent modules and improving the system's compatibility and stability.

[0090] The modules interact with each other using a standardized internal communication protocol. All interactive data is transmitted in a structured format. The core interaction process is as follows: After parsing the regulatory request, the request parsing module 100 sends a regulatory condition identification instruction to the operating condition identification module 200. The instruction includes the unique code of the target equipment, the query time range, and a list of operating indicators to be monitored. After generating a feature fingerprint, the operating condition identification module 200 sends the complete feature fingerprint data to the data fusion module 300. The data includes equipment identity, measurement point identity, time information, sampling data, and core feature values. After generating a panoramic equipment ledger, the data fusion module 300 sends the complete fused data to the regulatory decision-making module 400. The data includes static equipment ledger information, dynamic operating data, and time alignment identifiers. After generating a supervision and early warning work order, the regulatory decision-making module 400 sends the work order data to the operation and maintenance scheduling system. The work order includes faulty equipment information, the time period of the fault, a description of the fault phenomenon, the early warning level, and preliminary handling suggestions.

[0091] The operating condition identification module is used to collect multi-source runtime sequence data streams of the target equipment in real time and extract physical features reflecting equipment operating fluctuations. When the fluctuation rate exceeds a preset equipment sensitivity threshold, a feature fingerprint representing the current operating condition of the equipment is generated. Deployed in the stream processing layer of the industrial internet platform, the operating condition identification module communicates with the time-series data access bus and the associated index module. It incorporates a sliding window calculation unit, a threshold judgment unit, a feature extraction unit, and an asynchronous update unit. It achieves real-time processing of time-series data streams and automatic identification of abnormal operating conditions, enabling timely capture of changes in equipment operating status and providing accurate basis for subsequent fault warnings.

[0092] The data fusion module is used to match corresponding static equipment ledger information in the heterogeneous association index based on the feature fingerprint, calculate the time difference offset between the trigger time of the feature fingerprint and the sampling record time of the target static equipment ledger, and perform asynchronous compensation and alignment to generate equipment panoramic ledger data that matches time and attributes. The data fusion module is deployed at the platform layer of the industrial internet platform and communicates with the heterogeneous association index, relational database, and underlying time-series database. It has built-in index retrieval unit, time difference calculation unit, hierarchical compensation execution unit, and data aggregation unit. It achieves efficient fusion and accurate alignment of time-series data and static data, and the generated equipment panoramic ledger data can comprehensively and accurately reflect the operating status of the equipment, providing reliable data support for regulatory decisions.

[0093] The monitoring and decision-making module is used to input equipment overview data into a preset operation early warning rule base, calculate the operation deviation index of the target equipment, and generate a monitoring and early warning work order containing static equipment ledger information and dynamic fault period when the operation deviation index exceeds the safety threshold, and dispatch it to the operation and maintenance scheduling system. The monitoring and decision-making module is deployed at the application layer of the industrial internet platform, communicating with the operation early warning rule base and the operation and maintenance scheduling system. It has a built-in rule engine, deviation calculation unit, work order generation unit, and work order dispatch unit. This enables automatic early warning of equipment faults and automatic dispatch of work orders, improving fault response speed and operation and maintenance efficiency, reducing manual intervention, and lowering the probability of human error.

[0094] Preferably, the system also includes a closed-loop feedback optimization module 500, used to receive work order execution status and on-site measured deviation data from the operation and maintenance scheduling system, and dynamically correct the safety thresholds in the operation early warning rule base based on the on-site measured deviation data. The closed-loop feedback optimization module 500 is deployed at the application layer of the industrial internet platform, communicating with the operation and maintenance scheduling system and the operation early warning rule base, and includes a built-in data receiving unit, data analysis unit, and rule update unit. Work order execution status includes work order reception, work order processing, work order completion, and work order rejection. On-site measured deviation data refers to the difference between the equipment parameters actually measured by operation and maintenance personnel when handling faults on-site and the operation deviation index calculated by the system when issuing an early warning. Achieving closed-loop optimization of the monitoring system allows for continuous adjustment and improvement of early warning rules based on actual on-site conditions, continuously reducing false alarm and false negative rates, and improving the accuracy and reliability of the system.

[0095] The closed-loop feedback optimization module 500 uses a weighted moving average method to dynamically correct the safety threshold. The specific steps are as follows: It continuously receives on-site measured deviation data from the operation and maintenance scheduling system. Each feedback deviation is the difference between the actual on-site measurement and the calculated value at the time of system warning. The 10 most recent valid feedback deviation data are selected and assigned different weights according to their time proximity: the most recent feedback has a weight of 10, the previous one 9, and so on, with the earliest feedback having a weight of 1. The weighted average deviation value is calculated by multiplying each deviation data by its corresponding weight, summing the results, and then dividing by the sum of all weights. The safety threshold is corrected based on the weighted average deviation value: the new safety threshold equals the original safety threshold plus the product of the weighted average deviation value and a correction coefficient. The correction coefficient ranges from 0.1 to 0.3, with a default value of 0.2. The threshold update operation is performed once daily at 2:00 AM. An update log is automatically generated after each update, retaining the threshold history for the most recent 30 days for traceability.

[0096] A computer device includes: a memory for storing computer programs; and a processor for executing the computer programs to implement the aforementioned intelligent monitoring method for device operating status based on multi-dimensional data asynchronous compensation. The computer device is a server cluster of an industrial internet platform, including at least one application server, a database server, and a cache server. Each server is equipped with an independent memory and processor. The memory includes a high-speed cache area where heterogeneous relational indexes reside. The high-speed cache area is a DDR5 memory cache configured in the cache server, with a read / write latency of less than 100ns and a capacity of not less than 256GB. The processor uses a multi-core high-performance processor, with a minimum of 32 cores per server and a clock speed of not less than 2.5GHz. Through server cluster deployment, it can support the access of massive amounts of power data and large-scale concurrent queries, ensuring high availability and high stability of the system; by using a memory-level high-speed cache to reside heterogeneous relational indexes, ultra-low latency access to the indexes is achieved, improving query response speed.

[0097] A computer-readable storage medium stores a computer program thereon, which, when executed by a processor, implements the aforementioned intelligent monitoring method for device operating status based on multi-dimensional data asynchronous compensation. The computer-readable storage medium includes, but is not limited to, media capable of storing computer program code such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks. This allows the method of the present invention to be stored and distributed in the form of a software product, facilitating users to deploy and use the technical solution of the present invention on different computer devices.

[0098] A computer program product includes a computer program that, when executed by a processor, implements the aforementioned intelligent monitoring method for device operating status based on asynchronous compensation of multi-dimensional data. The computer program product can exist in the form of an installation package, image file, etc., and can run on various operating systems and hardware platforms. It provides a more convenient deployment and usage method; users only need to install the corresponding computer program product to use the technical solution of this invention, without the need for complex configuration and development.

[0099] To facilitate understanding of the technical solution of this invention, its working principle is explained in detail below: The first stage is the preprocessing stage: the system generates a unique global service identifier for each access power measurement point and establishes a mapping relationship between the global service identifier and the unique code of the device in the heterogeneous association index; at the same time, the operating condition identification module 200 collects multi-source runtime sequence data streams of the target device in real time, extracts physical features and calculates the fluctuation rate of change. Only when the fluctuation rate of change exceeds the device sensitivity threshold is a feature fingerprint generated and asynchronously updated to the heterogeneous association index.

[0100] The second stage is the regulatory execution stage: When a request for operational status supervision is received, the request parsing module 100 parses the request and obtains the static equipment ledger information of the target device; the data fusion module 300 matches the corresponding static equipment ledger information in the heterogeneous association index based on the feature fingerprint, calculates the time difference offset and performs asynchronous compensation alignment, and generates equipment panoramic ledger data; the regulatory decision module 400 inputs the equipment panoramic ledger data into the operation early warning rule base, calculates the operation deviation index, and if it exceeds the safety threshold, generates a supervision early warning work order and dispatches it to the operation and maintenance scheduling system.

[0101] The third stage is the closed-loop optimization stage: The closed-loop feedback optimization module 500 receives the work order execution status and on-site measured deviation data from the operation and maintenance scheduling system, and dynamically corrects the safety thresholds in the operation early warning rule base based on the on-site measured deviation data, thereby continuously improving the system's early warning accuracy.

[0102] Through the collaborative work of the above three stages, this invention solves the problems of slow cross-database query response, memory expansion, inability to balance real-time performance and consistency, and lack of closed-loop optimization capabilities in existing equipment monitoring systems, and achieves efficient, accurate, and intelligent monitoring of equipment operating status.

[0103] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for intelligent monitoring of equipment operating status based on multi-dimensional data asynchronous compensation, characterized in that, include: Receive and parse operational status monitoring requests for target devices to obtain static equipment ledger information for the target devices; The system collects multi-source runtime sequence data streams of the target device in real time, extracts physical characteristics that reflect the fluctuations in the operation of the target device, and generates a feature fingerprint representing the current operating condition of the target device when the fluctuation rate exceeds the preset device sensitivity threshold. Based on the feature fingerprint, the corresponding static equipment ledger information is matched in the heterogeneous association index. The time difference offset between the trigger time of the feature fingerprint and the sampling record time of the target static equipment ledger is calculated and asynchronously compensated and aligned to generate equipment panoramic ledger data that matches time and attributes. The equipment overview ledger data is input into the preset operation early warning rule library to calculate the operation deviation index of the target equipment. When the operation deviation index exceeds the safety threshold, a supervision and early warning work order containing the static equipment ledger information of the target equipment and the dynamic fault period is generated and dispatched to the operation and maintenance scheduling system. The steps for extracting physical characteristics reflecting the operational fluctuations of the target equipment include: Real-time calculation of the instantaneous change rate of each monitored parameter in the multi-source runtime sequence data stream; When the instantaneous rate of change exceeds the preset device sensitivity threshold, a time-series sampling segment centered on the current moment is extracted as a feature fingerprint reflecting the real-time operating condition of the device. The steps for generating equipment overview ledger data that matches time and attributes include: The operational status monitoring requests are analyzed using business logic divide-and-conquer, identifying the monitoring dimensions involved in the monitoring requests as dynamic operational dimensions and static asset dimensions; Based on the identification results, heterogeneous association indexes and relational databases are searched in parallel to extract dynamic historical operating condition data and corresponding static equipment parameters associated with the feature fingerprints, and multidimensional data aggregation is performed. The steps of calculating the time difference offset between the trigger time of the feature fingerprint and the sampling record time of the target static device ledger, and performing asynchronous compensation and alignment, include: Locate the anchor point of the corresponding feature fingerprint in the heterogeneous association index, and calculate the time difference offset between the trigger time of the feature fingerprint and the sampling record time of the target static device ledger. Determine whether the time difference offset exceeds the preset consistency tolerance threshold; Based on the judgment result, either the incremental fetch mode targeting the underlying database or the interpolation completion mode based on feature fingerprints is executed. If the limit is exceeded, the incremental fetch mode will be executed, and a fetch request for the specific missing segment will be sent to the underlying time series database. If the value is not exceeded, the interpolation completion mode is executed, and the interpolation is performed using the feature fingerprint.

2. The intelligent monitoring method for equipment operating status based on multi-dimensional data asynchronous compensation according to claim 1, characterized in that, Also includes: A global service identifier is generated for each power measurement point connected to the system, and a mapping relationship is established between the global service identifier and the unique device code in the static device ledger in the heterogeneous association index.

3. An intelligent monitoring system for equipment operating status based on multi-dimensional data asynchronous compensation, characterized in that, The system is used to execute the intelligent monitoring method for equipment operating status based on multi-dimensional data asynchronous compensation as described in claim 1 or 2, and the system includes: The request parsing module (100) is used to receive and parse the operation status monitoring request for the target device and obtain the static device ledger information of the target device. The operating condition identification module (200) is used to collect multi-source runtime sequence data streams of the target device in real time and extract physical features that reflect the operating fluctuations of the target device. When the fluctuation rate exceeds the preset device sensitivity threshold, a feature fingerprint representing the current operating condition of the device is generated. The data fusion module (300) is used to match the corresponding static equipment ledger information in the heterogeneous association index based on the feature fingerprint, calculate the time difference offset between the trigger time of the feature fingerprint and the sampling record time of the target static equipment ledger and perform asynchronous compensation and alignment to generate equipment panoramic ledger data that matches time and attributes. The monitoring decision module (400) is used to input the equipment panoramic ledger data into the preset operation early warning rule library, calculate the operation deviation index of the target equipment, and when the operation deviation index exceeds the safety threshold, generate a monitoring early warning work order containing static ledger information of the equipment and dynamic fault period, and dispatch it to the operation and maintenance scheduling system.

4. The intelligent monitoring system for equipment operation status based on multi-dimensional data asynchronous compensation according to claim 3, characterized in that, Also includes: The closed-loop feedback optimization module (500) is used to receive the work order execution status and on-site measured deviation data from the operation and maintenance scheduling system, and dynamically correct the safety threshold in the operation early warning rule base according to the on-site measured deviation data.

5. A computer device, characterized in that, include: Memory, used to store computer programs; A processor is used to execute the computer program to implement the intelligent monitoring method for device operating status based on multidimensional data asynchronous compensation as described in claim 1 or 2.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by the processor, the program implements the intelligent monitoring method for device operating status based on multidimensional data asynchronous compensation as described in claim 1 or 2.

7. A computer program product, comprising a computer program, characterized in that, When executed by the processor, the program implements the intelligent monitoring method for device operating status based on multidimensional data asynchronous compensation as described in claim 1 or 2.

Images