Equipment operation and maintenance management system based on intelligent diagnosis

Abstract

The embodiment of the invention provides an equipment operation and maintenance management system based on intelligent diagnosis, and the system comprises a data collection and transmission module which collects operation data from target equipment and uploads the operation data to an edge computing node or a cloud server; the multi-model diagnosis engine is configured to receive the operation data and call a diagnosis model to perform parallel analysis on the operation data so as to generate a preliminary diagnosis result at least comprising a fault type and a severity level; the decision fusion and early warning center is configured to perform confidence weighted fusion on the preliminary diagnosis result, generate a target diagnosis conclusion and trigger a corresponding early warning operation according to the severity level of the target diagnosis conclusion, and the early warning operation comprises prompt information, notification information or automatic triggering of an emergency control instruction; and the linkage control interface is configured to automatically execute preset emergency operation when receiving the emergency control instruction, and the emergency operation comprises switching of standby equipment or adjustment of system operation parameters. The system forms a complete technical closed loop, and the operation and maintenance safety of equipment is improved.

Description

Technical Field

[0001] This application relates to the field of equipment testing technology, and in particular to an equipment operation and maintenance management system based on intelligent diagnosis. Background Technology

[0002] In the fields of industrial automation, intelligent manufacturing, and critical equipment monitoring, the stable and reliable operation of equipment is crucial for ensuring production efficiency and economic benefits. With the widespread adoption of IoT technology, modern equipment monitoring systems typically rely on data transmission modules to upload equipment operating data to the cloud or local servers for processing via wired or wireless means. Building upon this, analyzing data using fault diagnosis models has become an important technical approach. These models have evolved from traditional rule-based expert systems to today's machine learning and deep learning algorithms, aiming to automatically identify various faults such as electrical short circuits and mechanical wear. Furthermore, the system integrates early warning and alarm mechanisms, notifying maintenance personnel via SMS, email, etc., based on the severity of the fault, and providing a remote operation interface for viewing equipment status and remote control, thus forming a relatively complete equipment health management and maintenance system.

[0003] However, existing technical solutions of this type still have significant shortcomings in practical applications. First, the various modules of the system often exhibit loose coupling or operate in isolation, with insufficient coordination among multiple functional units such as data transmission, fault diagnosis, early warning, and control, resulting in overall sluggish response and difficulty in achieving rapid closed-loop handling of sudden severe faults. Second, existing fault diagnosis models mostly employ single algorithms, which have limited adaptability and diagnostic accuracy when facing complex and ever-changing industrial field conditions, and are prone to misjudgment or missed diagnosis due to weak model generalization ability. Third, although the early warning mechanism is tiered, it is usually limited to information notification and lacks deep linkage with the equipment control system. It cannot automatically execute critical operations such as switching to backup equipment or activating emergency plans in emergency situations, still requiring manual intervention and missing the best handling opportunity. In addition, existing solutions still need to be integrated and improved in terms of the real-time performance of data processing, the accuracy of diagnostic results, and the automation and effectiveness of system response actions. Summary of the Invention

[0004] In view of this, embodiments of this application provide a device operation and maintenance management system and method based on intelligent diagnostics. One or more embodiments of this application also relate to a computing device, a computer-readable storage medium, and a computer program to address the technical deficiencies existing in the prior art.

[0005] In a first aspect, embodiments of this application provide a device operation and maintenance management system based on intelligent diagnostics, including: The data acquisition and transmission module is configured to collect operational data from the target device and upload it to the edge computing node or cloud server via wired or wireless communication. A multi-model diagnostic engine, deployed on the edge computing node or cloud server, is configured to receive the runtime data and call one or more diagnostic models among the diagnostic unit, machine learning classifier, and deep learning model to perform parallel analysis on the runtime data to generate preliminary diagnostic results that include at least the fault type and severity level. The decision fusion and early warning center is configured to perform confidence-weighted fusion on the preliminary diagnostic results, generate a target diagnostic conclusion, and trigger a corresponding early warning operation based on the severity level of the target diagnostic conclusion. The early warning operation includes prompt information, notification information, or automatic triggering of emergency control instructions. The linkage control interface is configured to automatically execute preset emergency operations upon receiving the emergency control command. These emergency operations include switching to backup equipment or adjusting system operating parameters.

[0006] In one possible implementation, the data acquisition and transmission module dynamically selects one of the following communication protocols—4G, 5G, LoRa, or NB-IoT—based on the communication link quality and data priority during wireless communication.

[0007] In one possible implementation, the deep learning model in the multi-model diagnostic engine is a recurrent neural network used to model the time sequence of equipment operation data to identify progressive mechanical faults.

[0008] In one possible implementation, the specific steps by which the decision fusion and early warning center performs the confidence-weighted fusion include: Calculate the overall confidence score of the target diagnostic conclusion; When the overall confidence score is greater than or equal to the confidence threshold, the early warning operation is triggered. The overall confidence score is determined using the following formula: in, The overall confidence score is given. The weight factor for the i-th diagnostic model is preset in the multi-model diagnostic engine, and satisfies the following conditions: , is the confidence value output by the i-th diagnostic model for the current diagnostic result.

[0009] In one possible implementation, the weighting factor is dynamically adjusted based on the historical diagnostic accuracy of each invoked diagnostic model.

[0010] In one possible implementation, the system further includes a device health assessment module, which is configured to: The real-time health index of the target device is calculated periodically and sent to the multi-model diagnostic engine. The multi-model diagnostic engine combines the operating data of the target device and the real-time health index to analyze and output preliminary diagnostic results. The calculation process of the real-time health index includes: Obtain parameters of vibration, temperature, and load of the target equipment; Calculate the normalized deviations of vibration, temperature, and load respectively, where vibration deviation is the square of the ratio of real-time vibration amplitude to maximum permissible vibration amplitude, temperature deviation is the ratio of the absolute deviation of real-time temperature value relative to rated temperature value to maximum permissible temperature deviation, and load deviation is the ratio of real-time load value to maximum permissible load value. These normalized deviations are multiplied by their respective preset weighting coefficients and summed to obtain a comprehensive degradation index. The real-time health index is obtained by subtracting the baseline health value from the comprehensive deterioration index.

[0011] In one possible implementation, the real-time health index is calculated using the following formula: in, The real-time health index, For real-time vibration amplitude, For the maximum permissible vibration amplitude, This is the real-time temperature value. This is the rated temperature value. For the maximum allowable temperature deviation, This is the real-time load value. The maximum allowable load value, , These are the weighting coefficients for vibration, temperature, and load, respectively, and they satisfy... .

[0012] In one possible implementation, the linkage control interface sends an operation confirmation request to the remote operation interface before performing the emergency operation.

[0013] In one possible implementation, the system further includes the remote operation interface, providing a web or mobile access interface that allows authorized users to remotely view device status and perform restart or reset operations.

[0014] In one possible implementation, the remote operation interface supports role-based hierarchical permission management, where users at different levels are granted different operation permissions.

[0015] In one possible implementation, the diagnostic unit is configured with an expert knowledge base for rapid logical judgment of short circuits and overloads in electrical faults.

[0016] Secondly, embodiments of this application provide a device operation and maintenance management method based on intelligent diagnostics, including: The system collects operational data from the target device and uploads it to the edge computing node or cloud server via wired or wireless communication. The system invokes one or more diagnostic models, including diagnostic units, machine learning classifiers, and deep learning models, to perform parallel analysis on the runtime data in order to generate preliminary diagnostic results that include at least the fault type and severity level. The preliminary diagnostic results are weighted and fused with confidence levels to generate a target diagnostic conclusion. Based on the severity level of the target diagnostic conclusion, a corresponding early warning operation is triggered. The early warning operation includes prompt information, notification information, or automatic triggering of emergency control instructions. Upon receiving the emergency control command, the system automatically executes preset emergency operations, including switching to backup equipment or adjusting system operating parameters.

[0017] Thirdly, embodiments of this application provide a computing device, including: Memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, the steps of the above-described equipment operation and maintenance management method based on intelligent diagnosis are implemented.

[0018] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer-executable instructions that, when executed by a processor, implement the steps of the above-described intelligent diagnostic-based equipment operation and maintenance management method.

[0019] Fifthly, embodiments of this application provide a computer program, wherein when the computer program is executed in a computer, it causes the computer to perform the steps of the above-described intelligent diagnostic-based equipment operation and maintenance management method.

[0020] The technical solution provided in this application first collects real-time operational data from industrial equipment through a data acquisition and transmission module, and then streams the data to edge or cloud computing nodes via wired or wireless networks. Subsequently, this data is fed into a multi-model diagnostic engine. This engine does not employ a single algorithm but instead utilizes multiple heterogeneous algorithms in parallel, such as rule-based diagnostic units, machine learning classifiers, and deep learning models, to collaboratively analyze the same data. The rule base is responsible for quickly matching known and explicit fault modes, the machine learning model identifies anomalies at the statistical feature level, and the deep learning model mines complex nonlinear and temporal features in the data, thereby jointly outputting a set of preliminary diagnostic results covering fault type and severity level. Next, the decision fusion and early warning center integrates and arbitrates the above multiple preliminary results, deriving a unified and more reliable final diagnostic conclusion through a specific fusion strategy (such as weighted voting or confidence comparison). Based on the severity level of this conclusion, different levels of early warning actions are precisely triggered, ranging from simple status prompts and sending alarm notifications via communication networks to generating emergency control commands. Finally, upon receiving an emergency control command representing the highest level of urgency, the linkage control interface automatically executes predefined emergency operations without waiting for manual confirmation. These operations include quickly switching to backup equipment or forcibly adjusting process parameters to suppress fault progression. By tightly coupling multiple functional modules and enabling parallel collaboration of multi-model diagnostic engines, the system effectively overcomes the shortcomings of poor adaptability and weak generalization ability of single models. It comprehensively utilizes the advantages of different algorithms, thereby significantly improving the accuracy and comprehensiveness of fault diagnosis and significantly reducing false alarms and missed alarms. Secondly, the seamless integration from data acquisition to linkage control constructs a highly automated closed-loop control system. This allows the system to respond to severe sudden faults at millisecond levels, transforming the slow mode from perception to alarm to manual intervention to an instant mode from perception to decision-making to automatic execution. This greatly shortens fault handling time and avoids production interruptions or equipment damage that may result from delays caused by manual intervention. Ultimately, this system forms a complete technical closed loop from condition monitoring and intelligent diagnosis to proactive protection, significantly improving the intelligence level of equipment operation and maintenance and the overall system's safety and reliability. Attached Figure Description

[0021] Figure 1 This is a schematic diagram illustrating a scenario of an intelligent diagnostic-based equipment operation and maintenance management method provided in one embodiment of this application; Figure 2 This is a schematic diagram of the structure of an equipment operation and maintenance management system based on intelligent diagnosis, provided in one embodiment of this application; Figure 3 This is a flowchart of a device operation and maintenance management method based on intelligent diagnosis provided in one embodiment of this application; Figure 4This is a structural block diagram of a computing device provided in one embodiment of this application. Detailed Implementation

[0022] Many specific details are set forth in the following description to provide a full understanding of this application. However, this application can be implemented in many other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of this application; therefore, this application is not limited to the specific embodiments disclosed below.

[0023] The terminology used in one or more embodiments of this application is for the purpose of describing particular embodiments only and is not intended to limit the scope of one or more embodiments of this application. The singular forms “a” and “the” as used in one or more embodiments of this application and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used in one or more embodiments of this application refers to and includes any or all possible combinations of one or more associated listed items.

[0024] It should be understood that although the terms first, second, etc., may be used to describe various information in one or more embodiments of this application, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, first may also be referred to as second without departing from the scope of one or more embodiments of this application, and similarly, second may also be referred to as first. Depending on the context, the word "if" as used herein may be interpreted as "when," "when," or "in response to a determination."

[0025] This application provides an equipment operation and maintenance management system and method based on intelligent diagnosis. This application also relates to a computing device and a computer-readable storage medium, which will be described in detail in the following embodiments.

[0026] See Figure 1 , Figure 1 The illustration shows a scenario diagram of an equipment operation and maintenance management method based on intelligent diagnostics according to an embodiment of this application.

[0027] exist Figure 1 In one application scenario, the data acquisition and transmission module 101 can be deployed on each wind turbine to continuously collect key operational data such as the vibration spectrum of the gearbox, the temperature of the generator windings, and the current and power output of the yaw motor. This data is transmitted in real time to the edge computing node located in the wind farm control center via a 4G / 5G wireless communication module deployed on-site.

[0028] On the edge computing node, the multi-model diagnostic engine 102 immediately initiates parallel analysis of the data: its built-in rule-based diagnostic unit quickly filters data based on expert rules such as "if the vibration amplitude exceeds X and the current fluctuation exceeds Y, it may be gear wear"; simultaneously, machine learning classifiers (such as SVM) classify vibration feature vectors to identify potential bearing loosening; and deep learning models (such as CNN) focus on analyzing the time-frequency spectrum of the vibration signal to detect minute cracks in the gear. These three processes work in parallel, each outputting preliminary diagnostic results with a severity rating.

[0029] These preliminary results were compiled and sent to the Decision Fusion and Early Warning Center 103. The center fused the three results according to preset weights and finally determined that the equipment had "secondary gear wear". Since it belonged to the "general fault" level, it automatically triggered an email / SMS notification to send the alarm information to the maintenance engineer's mobile phone.

[0030] The decision fusion and early warning center determined that the fault was likely to worsen. To protect the equipment, it issued an emergency control command to the linkage control interface 104 to "reduce load operation". The linkage control interface 104 responded immediately and automatically lowered the power setpoint of the wind turbine to a safe range through the control system network, thereby effectively preventing the fault from escalating and buying valuable time for on-site maintenance.

[0031] It should be noted that in other application scenarios, the equipment operation and maintenance management system and method based on intelligent diagnosis provided in this application embodiment can also achieve the same or similar technical effects.

[0032] Figure 2 This is a schematic diagram of the structure of an equipment operation and maintenance management system based on intelligent diagnosis, provided as an embodiment of this application.

[0033] Reference Figure 2 As shown, the system may include: The data acquisition and transmission module 101 is configured to acquire operational data from the target device and upload it to the edge computing node or cloud server via wired or wireless communication. The multi-model diagnostic engine 102, deployed on an edge computing node or cloud server, is configured to receive runtime data and call one or more diagnostic models among the diagnostic unit 102a, machine learning classifier 102b, and deep learning model 102c to perform parallel analysis on the runtime data in order to generate preliminary diagnostic results that include at least the fault type and severity level. The decision fusion and early warning center 103 is configured to perform confidence-weighted fusion of the preliminary diagnostic results, generate the target diagnostic conclusion, and trigger the corresponding early warning operation according to the severity level of the target diagnostic conclusion. The early warning operation includes prompt information, notification information or automatic triggering of emergency control instructions. The linkage control interface 104 is configured to automatically execute preset emergency operations when an emergency control command is received. The emergency operations include switching to backup equipment or adjusting system operating parameters.

[0034] In some embodiments, the data acquisition and transmission module 101 collects the physical state parameters and operational logic data of the target device in real time through various sensors and controllers integrated on the target device. Subsequently, based on the on-site infrastructure conditions and communication requirements, the module flexibly utilizes its built-in wired or wireless communication units to establish a transmission link. After standardizing and encapsulating the collected data, it continuously uploads it to a designated computing node. For real-time control scenarios requiring low-latency processing, the data is sent to an edge computing node near the device; while for analysis scenarios requiring large-scale storage and deep computing, the data is transmitted to a cloud server for centralized processing. This acquisition and transmission process ensures the comprehensiveness of data acquisition and the reliability of the transmission path, providing a stable and timely data supply to the backend system.

[0035] In some embodiments, during wireless communication, the data acquisition and transmission module 101 dynamically selects one of the following communication protocols: 4G, 5G, LoRa, or NB-IoT, based on the communication link quality and data priority. In one embodiment, the data acquisition and transmission module 101 integrates multiple communication modules (4G, 5G, LoRa, and NB-IoT) and runs a link decision algorithm. This module continuously monitors the link quality parameters of each wireless channel (such as signal strength, bit error rate, and network latency) and identifies the priority attributes of the data to be transmitted (e.g., real-time control commands have high priority, while periodic status data has normal priority). Based on these real-time parameters, the module dynamically performs protocol switching: when transmitting large volumes of real-time vibration data, it automatically enables 5G connectivity to ensure bandwidth and low latency; when transmitting small batches of temperature status information in signal edge areas, it switches to the more widely covered NB-IoT or LoRa network. This adaptively selects the optimal communication path in complex industrial environments, ensuring the reliability and efficiency of the data transmission link.

[0036] In some embodiments, after receiving the running data uploaded by the data acquisition and transmission module 101, the multi-model diagnostic engine 102 synchronously launches multiple heterogeneous diagnostic models integrated within it, including a rule-based diagnostic unit, a machine learning classifier, and a deep learning model. These models perform parallel analysis on the same batch of running data: the rule-based diagnostic unit performs rapid logical matching and state judgment based on a preset expert knowledge base; the machine learning classifier performs statistical classification and recognition on the extracted feature vectors; and the deep learning model performs end-to-end deep pattern mining on the original data or high-dimensional features. Each model independently outputs its analysis conclusions, and the engine encapsulates these conclusions into a set of preliminary diagnostic results containing specific fault types (such as "bearing wear" and "insulation aging") and corresponding severity levels (such as "minor," "moderate," and "severe"). The multi-model diagnostic engine 102 provides a parallel and collaborative analysis architecture. Specifically, by operating and mutually verifying multiple algorithm models simultaneously, it effectively overcomes the deficiency of insufficient generalization ability of a single algorithm under complex working conditions, significantly improves the accuracy and comprehensiveness of fault identification, especially the discovery of early hidden faults, and significantly reduces the risk of missed and false alarms. Secondly, the engine provides rich and multi-dimensional diagnostic information for backend decision-making.

[0037] In some embodiments, the diagnostic unit 102a is configured with an expert knowledge base for rapid logical judgment of short circuits and overloads in electrical faults. In one embodiment, the expert knowledge base pre-sets a series of judgment rules for electrical faults. When operating data is input, the unit synchronously initiates real-time pattern matching of electrical parameters. For example, when a sudden and sharp rise in current exceeding a safety threshold is detected, accompanied by a simultaneous drop in voltage, a "short circuit" fault rule is immediately triggered; when a current continuously exceeds the rated load and is accompanied by an abnormal temperature rise curve, an "overload" fault rule is triggered. Through this rapid logical matching of input signals with the pre-set rule base, the unit can directly output the corresponding fault type judgment result within milliseconds, providing a crucial speed advantage for subsequent decision fusion.

[0038] In some embodiments, the deep learning model 102c in the multi-model diagnostic engine 102 can be a model based on a recurrent neural network (RNN) to model the time-series of equipment operation data to identify progressive mechanical faults. In one implementation, the RNN receives time-series data (such as vibration signals, temperature changes, and noise spectra) generated during equipment operation and uses the recurrent feedback structure of its internal hidden layers to dynamically model the time dependencies in the data. The network learns the gradual change pattern of equipment status from normal to abnormal by processing the input at each time step sequentially and retaining its memory of historical states. This ultimately enables the identification and early warning of progressive mechanical faults with long latency periods and slow development, such as bearing wear and gear pitting. In this process, by capturing the temporal patterns and trend characteristics exhibited during the development of faults, the system significantly improves the early identification capability and predictive maintenance level of progressive mechanical faults, realizing a shift from passive response to proactive prevention. Secondly, this deep time series model can extract weak features related to equipment health degradation from seemingly chaotic operating noise, effectively reducing the false negative rate of progressive failures and enhancing the reliability of the diagnostic system in predicting the health status of complex mechanical systems.

[0039] In some embodiments, the system may further include a device health assessment module configured to: The real-time health index of the target device is calculated periodically and sent to the multi-model diagnostic engine 102. The multi-model diagnostic engine 102 then combines the target device's operating data and the real-time health index to analyze and output a more accurate preliminary diagnostic result.

[0040] In one implementation, the device health assessment module periodically acquires multi-dimensional operating parameters of the target device and uses a built-in health assessment algorithm to comprehensively calculate these parameters, generating a quantified real-time health index. This index is transmitted in real-time to a multi-model diagnostic engine. When performing fault diagnosis, the engine integrates and analyzes the traditional operating data sequence with this health index. By examining the correlation between abnormal operating data and the decline in the health index, it outputs a more accurate preliminary diagnostic result. By introducing a comprehensive health index as an important reference dimension for diagnosis, the comprehensive analytical capability of fault diagnosis is effectively improved. This allows the system to not only identify sudden faults but also capture the gradual degradation trend of equipment performance, thereby significantly enhancing the comprehensiveness of diagnostic conclusions and the reliability of early warnings.

[0041] In one implementation, the calculation process of the real-time health index includes: Obtain parameters of vibration, temperature, and load of the target equipment; Calculate the normalized deviations of vibration, temperature, and load respectively. The vibration deviation is the square of the ratio of the real-time vibration amplitude to the maximum permissible vibration amplitude. The temperature deviation is the ratio of the absolute deviation of the real-time temperature value relative to the rated temperature value to the maximum permissible temperature deviation. The load deviation is the ratio of the real-time load value to the maximum permissible load value. These normalized deviations are multiplied by their respective preset weighting coefficients and summed to obtain a comprehensive degradation index. The real-time health index is obtained by subtracting the baseline health value from the comprehensive deterioration index.

[0042] In the calculation of the real-time health index, the system first synchronously acquires the core operating parameters of the equipment, including vibration amplitude, real-time temperature, and operating load. Then, it performs normalized deviation calculations on these three key parameters. Vibration deviation uses a square relationship to amplify the impact of abnormal vibration, temperature deviation reflects the degree of thermal anomaly based on the absolute deviation from the rated value, and load deviation directly reflects the current stress state of the equipment. Next, these three normalized deviation values ​​are multiplied by their preset weighting coefficients and summed to generate a comprehensive index that fully reflects the overall deterioration status of the equipment. Finally, by subtracting this comprehensive deterioration index from the baseline health value, a standardized health index ranging from 0 to 1 is obtained. This calculation process, through multi-parameter fusion and normalization, transforms heterogeneous physical quantities into quantifiable health indicators. It considers the differences in the impact of different parameters on equipment health and enhances the sensitivity to vibration anomalies through square calculations, thereby significantly improving the accuracy and engineering practicality of health assessment and providing an intuitive and reliable quantitative basis for fault diagnosis.

[0043] In one implementation, the real-time health index is calculated using the following formula: in, The real-time health index, For real-time vibration amplitude, For the maximum permissible vibration amplitude, This is the real-time temperature value. This is the rated temperature value. For the maximum allowable temperature deviation, This is the real-time load value. The maximum allowable load value, , These are the weighting coefficients for vibration, temperature, and load, respectively, and they satisfy... .

[0044] In some embodiments, after receiving multiple preliminary diagnostic results from the multi-model diagnostic engine 102, the decision fusion and early warning center 103 processes these results using a weighted fusion strategy. This involves assigning different confidence weights to the results output by different diagnostic models and generating a unified, more reliable target diagnostic conclusion through weighted averaging. Subsequently, the center automatically analyzes the severity level determined in this conclusion and precisely triggers tiered early warning operations accordingly: for minor levels, only an interface prompt is generated; for general levels, notifications such as emails or SMS messages are automatically sent; and for severe levels, emergency control commands are immediately generated and issued automatically. By fusing diagnostic opinions from multiple models, misjudgments that may occur with a single model are effectively avoided, significantly improving the reliability and accuracy of the final decision. Secondly, precise tiered responses are achieved based on the severity level of the fused conclusion, ensuring the matching and adaptability of system early warnings to the severity of faults. Finally, the automated command triggering mechanism for severe faults realizes a rapid closed loop from diagnosis to control, improving the system's ability to handle critical faults in real time and its overall safety.

[0045] In some embodiments, the confidence-weighted fusion process of the decision fusion and early warning center 103 for the preliminary diagnostic results may include the following steps: S101: Calculate the overall confidence score of the target diagnostic conclusion; S102: When the overall confidence score is greater than or equal to the confidence threshold, the warning operation is triggered.

[0046] In one implementation, after obtaining multiple preliminary diagnostic results and their corresponding model confidence levels, the decision fusion and early warning center 103 first generates a comprehensive confidence score characterizing the reliability of the final conclusion through weighted calculation. Subsequently, the system compares this score in real time with a preset confidence threshold representing the minimum reliability requirement for the conclusion to be adopted. Only when the calculated comprehensive confidence score is greater than or equal to this threshold is the diagnostic conclusion deemed sufficiently reliable, and the system is allowed to execute subsequent corresponding early warning operations. By introducing a quantitative confidence score and threshold arbitration mechanism, low-reliability diagnostic results caused by misjudgment of a single model or conflicting conclusions between models are effectively filtered out, thereby significantly improving the accuracy and rigor of the system's early warning actions. This fundamentally avoids invalid alarms or erroneous linkage control caused by misdiagnosis, enhancing the overall reliability of the system output.

[0047] In one implementation, the overall confidence score is determined using the following formula: in, To calculate the overall confidence score, Let be the preset weight factor for the i-th diagnostic model in the multi-model diagnostic engine, and satisfy . , is the confidence value output by the i-th diagnostic model for the current diagnostic result.

[0048] In the process of calculating the comprehensive confidence score, the Decision Fusion and Early Warning Center first obtains the preliminary diagnostic conclusions and corresponding confidence scores of each diagnostic model in the multi-model diagnostic engine for the current operating condition, and simultaneously reads the pre-set weight coefficients reflecting the relative importance of each model. Then, the system multiplies the weight coefficient of each model by its output confidence score to obtain the weighted confidence score of that model. Finally, the weighted confidence scores of all models are summed to calculate the comprehensive confidence score representing the reliability of the final diagnostic conclusion. By introducing weight coefficients, the contribution of different diagnostic models is quantified and differentiated, allowing models with higher reliability to have a greater weight in the final decision. At the same time, the weighted summation condenses multi-dimensional diagnostic information into a single quantitative indicator, providing a clear and reliable decision-making basis for subsequent threshold judgments, thereby significantly improving the system's anti-interference capability and effectively avoiding system malfunctions caused by accidental misjudgments by a single model.

[0049] In some embodiments, the weighting factors in the comprehensive confidence score calculation process are dynamically adjusted based on the historical diagnostic accuracy of each diagnostic model invoked. Under this dynamic weighting adjustment mechanism, the system continuously monitors and records the historical diagnostic performance of each diagnostic model. By comparing the model output with the actual equipment status or maintenance confirmation results, the system periodically calculates the historical diagnostic accuracy of each model under different fault types. Based on these updated accuracy data, the decision fusion and early warning center automatically adjusts the weighting factors of the corresponding models according to preset mapping rules. Models with consistently stable performance and high accuracy have their weights appropriately increased, while models with recent accuracy declines have their weights decreased accordingly, thereby achieving dynamic matching between weight configuration and actual model performance. This mechanism, by enabling weighting factors to adaptively optimize with model performance, effectively improves the long-term adaptability and diagnostic reliability of the system when facing complex factors such as changes in equipment operating conditions and component aging, ensuring that the scientific nature of fusion decision-making continuously improves over time.

[0050] In some embodiments, when the linkage control interface 104 receives an emergency control command from the decision fusion and early warning center 103, it immediately activates its internally preset emergency response program and automatically executes the preset specific emergency operations. These operations may include rapidly disconnecting the main equipment from the operating line and simultaneously switching to standby equipment to maintain continuous system operation, or automatically adjusting key operating parameters (such as reducing output power or limiting load current) to put the equipment into a safe operating mode. By establishing an automated link from diagnostic conclusions to control execution, immediate handling of serious faults is achieved, transforming the traditional delayed response relying on manual judgment and operation into a millisecond-level response autonomously completed by the system. This effectively curbs the further development of faults and significantly improves the system's autonomous protection capability and overall operational safety in the face of critical conditions.

[0051] In some embodiments, the system also includes a remote operation interface, providing a web or mobile access interface to support authorized users to remotely view device status and perform restart or reset operations.

[0052] In one implementation, the remote operation interface can be implemented by building a web application based on a B / S (browser / server) architecture and an application adapted to mainstream mobile operating systems. When a user initiates an access request, the interface first forces a redirect to a unified identity authentication page, requiring the user to enter pre-registered account credentials. After successful verification, the backend service dynamically loads the corresponding operation interface and function list based on the role permissions bound to the account. Authenticated users can view real-time equipment status information (including operating parameters, health indices, fault records, etc.) uploaded and analyzed by the data acquisition module on the interface. When equipment malfunctions, users with the corresponding operating permissions can remotely trigger preset control commands (such as soft restart, hard reset, etc.). These commands are transmitted to the on-site control system via an encrypted channel for final execution, thereby realizing cross-regional equipment operation and maintenance management. By combining strict permission control with convenient remote access, the response efficiency and management convenience of equipment operation and maintenance are significantly improved while ensuring system operation security.

[0053] In some embodiments, the remote operation interface supports role-based hierarchical permission management, where users at different levels are granted different operation permissions.

[0054] In one implementation, the remote operation interface achieves hierarchical permission management by implementing a Role-Based Access Control (RBAC) model: the system predefines multiple role categories (such as inspector, maintenance engineer, and system administrator), and each role is granted a strictly differentiated set of operation permissions; when a user is authenticated, the system dynamically renders the operation interface and activates the corresponding functional modules according to their role. For example, an inspector can only view equipment status charts, a maintenance engineer has additional operation permissions such as equipment reset and parameter adjustment, while a system administrator has user management, permission allocation, and all advanced operation permissions. By precisely binding system functions with user roles, the professionalism and controllability of critical operations are ensured, while the security risks caused by unauthorized operations are effectively prevented, achieving an organic balance between convenience and security.

[0055] In some embodiments, the linkage control interface 104 sends an operation confirmation request to the remote operation interface before performing the emergency operation.

[0056] In one implementation, the process by which the linkage control interface 104 sends an operation confirmation request to the remote operation interface before executing an emergency operation is as follows: When the linkage control interface receives an emergency control command from the decision fusion and early warning center, it does not immediately execute the corresponding emergency operation. Instead, it first generates a confirmation request containing fault details, suggested operations, and execution time limits, and pushes this request to the remote operation interface in real time via the system's internal communication bus. Upon receiving the request, the remote operation interface immediately sends a prominent prompt to the authorized administrator on the web interface or mobile terminal and starts a countdown mechanism. The administrator can choose to approve or cancel the operation based on the actual situation on site, and the decision result will be immediately fed back to the linkage control interface. Only when a clear approval instruction is obtained will the linkage control interface finally trigger the execution of the preset emergency operation. By introducing a manual final confirmation step for key operations, effective safety redundancy is established while maintaining the system's rapid response capability. This ensures timely handling of critical faults in unattended scenarios and empowers operators with the final decision-making authority for major system operations, effectively preventing the system from executing unnecessary emergency actions due to misjudgment or interference, and significantly improving the reliability and safety of the entire control system.

[0057] The following is a detailed description of the intelligent diagnostic-based equipment operation and maintenance management system provided in this application through a complete embodiment.

[0058] This embodiment uses the monitoring of a large water pump unit as an example to explain in detail the complete workflow of this system.

[0059] 1. Data Acquisition and Transmission Vibration sensors, temperature probes, and current transformers installed on the pump unit continuously collect operating data. This data is collected and packaged by the data acquisition and transmission module. Based on the on-site network conditions, this module prioritizes low-latency 5G wireless networks to transmit the data stream in real time to the edge computing node located in the pump station control room.

[0060] 2. Parallel Diagnosis of Multiple Models The multi-model diagnostic engine on the edge computing node then initiates parallel analysis. Rule-based diagnostic units quickly determine the fault based on an expert knowledge base; for example, if the current exceeds the rated value for 10 seconds, an "overload" warning is triggered. Simultaneously, a machine learning classifier analyzes vibration feature vectors to identify imbalance faults. Meanwhile, a recurrent neural network (RNN) model performs in-depth analysis of the time-series vibration signal to detect early signs of bearing wear. All three work in parallel, each outputting a preliminary diagnostic result containing the fault type (e.g., "overload," "bearing wear") and severity level (e.g., "moderate," "severe").

[0061] 3. Comprehensive assessment of equipment health At the same time, the equipment health assessment module periodically obtains vibration, temperature, and load parameters from the same data source and performs health index calculation, resulting in a real-time health index of 0.65 (out of 1.0) for the current equipment. This index is then sent to the multi-model diagnostic engine as an auxiliary feature for its comprehensive judgment.

[0062] 4. Decision-making integration and hierarchical early warning The decision fusion and early warning center receives all the preliminary diagnostic results mentioned above. It uses a confidence-weighted fusion formula for calculation. In this cycle, the confidence scores of the rule unit, machine learning classifier, and RNN model outputs are 0.7, 0.8, and 0.9, respectively, and the system dynamically assigns weights of 0.2, 0.3, and 0.5 based on their recent accuracy. The calculated comprehensive confidence score S = (0.2*0.7) + (0.3*0.8) + (0.5*0.9) = 0.83, which is higher than the preset threshold of 0.75. Therefore, the fusion conclusion "severe bearing wear" is adopted as the final diagnostic conclusion. Based on its "severe" level, the early warning center automatically triggers the highest level of early warning action: on the one hand, it sends an alarm SMS to the maintenance personnel's mobile phone, and on the other hand, it generates an emergency control command to "immediately switch to the backup pump".

[0063] 5. Linkage control and safety confirmation Upon receiving the emergency control command, the linkage control interface did not execute it immediately. To ensure safety, it first sent a "standby pump switchover confirmation request" with a 60-second countdown to the remote operation interface. An engineer with "system administrator" privileges received the request and a detailed fault report via a mobile app. Given the urgency of the situation, the engineer clicked "confirm execution" before the countdown ended. Upon receiving the confirmation, the linkage control interface immediately activated the electrical control cabinet, automatically disconnecting the faulty pump from the pipeline network and seamlessly starting the standby pump, thus ensuring the continuous and stable operation of the water supply system.

[0064] 6. Remote monitoring and management Throughout the process, authorized engineers can view all pump status data, health index change curves, diagnostic process records, and historical alarms in real time via a web browser or mobile app through the remote operation interface. The system strictly adheres to role-based access control, ensuring that only high-level administrators are authorized to perform critical operations such as emergency switchovers, effectively preventing accidental operations.

[0065] See Figure 3 , Figure 3 A flowchart of a device operation and maintenance management method based on intelligent diagnostics according to an embodiment of this application is shown, which specifically includes the following steps.

[0066] Step 301: Collect operational data from the target device and upload it to the edge computing node or cloud server via wired or wireless communication. Step 302: Call one or more diagnostic models from the diagnostic unit, machine learning classifier, and deep learning model to perform parallel analysis on the running data to generate preliminary diagnostic results that include at least the fault type and severity level; Step 303: Perform confidence-weighted fusion on the preliminary diagnosis results to generate the target diagnosis conclusion, and trigger the corresponding early warning operation according to the severity level of the target diagnosis conclusion. The early warning operation includes prompt information, notification information or automatic triggering of emergency control instructions. Step 304: Upon receiving an emergency control command, automatically execute preset emergency operations, including switching to backup equipment or adjusting system operating parameters.

[0067] In one possible implementation, during wireless communication, a communication protocol of 4G, 5G, LoRa, or NB-IoT is dynamically selected based on the communication link quality and data priority.

[0068] In one possible implementation, the deep learning model is a recurrent neural network used to model the time sequence of equipment operation data to identify progressive mechanical failures.

[0069] In one possible implementation, the specific steps for performing the confidence-weighted fusion include: Calculate the overall confidence score of the target diagnostic conclusion; When the overall confidence score is greater than or equal to the confidence threshold, the early warning operation is triggered. The overall confidence score is determined using the following formula: in, The overall confidence score is given. The weight factor for the i-th diagnostic model is preset in the multi-model diagnostic engine, and satisfies the following conditions: , is the confidence value output by the i-th diagnostic model for the current diagnostic result.

[0070] In one possible implementation, the weighting factor is dynamically adjusted based on the historical diagnostic accuracy of each invoked diagnostic model.

[0071] In one possible implementation, the method also includes: The real-time health index of the target device is calculated periodically, and the preliminary diagnostic results are analyzed and output by combining the operating data of the target device with the real-time health index. The calculation process of the real-time health index includes: Obtain parameters of vibration, temperature, and load of the target equipment; Calculate the normalized deviations of vibration, temperature, and load respectively, where vibration deviation is the square of the ratio of real-time vibration amplitude to maximum permissible vibration amplitude, temperature deviation is the ratio of the absolute deviation of real-time temperature value relative to rated temperature value to maximum permissible temperature deviation, and load deviation is the ratio of real-time load value to maximum permissible load value. These normalized deviations are multiplied by their respective preset weighting coefficients and summed to obtain a comprehensive degradation index. The real-time health index is obtained by subtracting the baseline health value from the comprehensive deterioration index.

[0072] In one possible implementation, the real-time health index is calculated using the following formula: in, The real-time health index, For real-time vibration amplitude, For the maximum permissible vibration amplitude, This is the real-time temperature value. This is the rated temperature value. For the maximum allowable temperature deviation, This is the real-time load value. The maximum allowable load value, , These are the weighting coefficients for vibration, temperature, and load, respectively, and they satisfy... .

[0073] In one possible implementation, an operation confirmation request is sent to the remote operation interface before the emergency operation is performed.

[0074] In one possible implementation, a web or mobile access interface is provided, allowing authorized users to remotely view device status and perform restart or reset operations.

[0075] One possible implementation supports role-based hierarchical permission management, where users at different levels are granted different operation permissions.

[0076] In one possible implementation, the diagnostic unit is configured with an expert knowledge base for rapid logical judgment of short circuits and overloads in electrical faults.

[0077] Figure 4 A structural block diagram of a computing device 400 according to an embodiment of this application is shown. The components of the computing device 400 include, but are not limited to, a memory 410 and a processor 420. The processor 420 is connected to the memory 410 via a bus 430, and a database 450 is used to store data.

[0078] The computing device 400 also includes an access device 440, which enables the computing device 400 to communicate via one or more networks 460. Examples of these networks include Public Switched Telephone Network (PSTN), Local Area Network (LAN), Wide Area Network (WAN), Personal Area Network (PAN), or combinations of communication networks such as the Internet. The access device 440 may include one or more of any type of wired or wireless network interface (e.g., a network interface card (NIC)), such as an IEEE 802.11 Wireless Local Area Network (WLAN) wireless interface, a Wi-MAX (Worldwide Interoperability for Microwave Access) interface, an Ethernet interface, a Universal Serial Bus (USB) interface, a cellular network interface, a Bluetooth interface, or a Near Field Communication (NFC) interface.

[0079] In one embodiment of this application, the aforementioned components of the computing device 400 and Figure 4 Other components, not shown, can also be connected to each other, for example, via a bus. It should be understood that... Figure 4 The block diagram of the computing device shown is for illustrative purposes only and is not intended to limit the scope of this application. Those skilled in the art can add or replace other components as needed.

[0080] The computing device 400 can be any type of stationary or mobile computing device, including mobile computers or mobile computing devices (e.g., tablet computers, personal digital assistants, laptop computers, notebook computers, netbooks, etc.), mobile phones (e.g., smartphones), wearable computing devices (e.g., smartwatches, smart glasses, etc.) or other types of mobile devices, or stationary computing devices such as desktop computers or personal computers (PCs). The computing device 400 can also be a mobile or stationary server.

[0081] The processor 420 executes the following computer-executable instructions, which, when executed by the processor, implement the steps of the aforementioned intelligent diagnostic-based equipment operation and maintenance management method. The above is an illustrative scheme of a computing device according to this embodiment. It should be noted that the technical solution of this computing device and the technical solution of the aforementioned intelligent diagnostic-based equipment operation and maintenance management method belong to the same concept. Details not described in detail in the technical solution of the computing device can be found in the description of the technical solution of the aforementioned intelligent diagnostic-based equipment operation and maintenance management method.

[0082] An embodiment of this application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the steps of the above-described intelligent diagnostic-based equipment operation and maintenance management method.

[0083] The above is an illustrative scheme of a computer-readable storage medium according to this embodiment. It should be noted that the technical solution of this storage medium belongs to the same concept as the technical solution of the above-described equipment operation and maintenance management method based on intelligent diagnosis. Details not described in detail in the technical solution of the storage medium can be found in the description of the technical solution of the above-described equipment operation and maintenance management method based on intelligent diagnosis.

[0084] An embodiment of this application also provides a computer program, wherein when the computer program is executed in a computer, it causes the computer to perform the steps of the above-described intelligent diagnostic-based equipment operation and maintenance management method.

[0085] The above is an illustrative scheme of a computer program according to this embodiment. It should be noted that the technical solution of this computer program and the technical solution of the above-described equipment operation and maintenance management method based on intelligent diagnosis belong to the same concept. For details not described in detail in the technical solution of the computer program, please refer to the description of the technical solution of the above-described equipment operation and maintenance management method based on intelligent diagnosis.

[0086] The foregoing has described specific embodiments of this application. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired results. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0087] The computer instructions include computer program code, which may be in the form of source code, object code, executable file, or certain intermediate forms. The computer-readable medium may include any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium may be appropriately added to or subtracted according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media may not include electrical carrier signals and telecommunication signals.

[0088] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that the embodiments of this application are not limited to the described order of actions, because according to the embodiments of this application, some steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also understand that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily essential to the embodiments of this application.

[0089] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0090] The preferred embodiments disclosed above are merely illustrative of this application. The optional embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the embodiments of this application. These embodiments are selected and specifically described in this application to better explain the principles and practical applications of the embodiments of this application, thereby enabling those skilled in the art to better understand and utilize this application. This application is limited only by the claims and their full scope and equivalents.

Claims

1. A device operation and maintenance management system based on intelligent diagnosis, characterized in that, include: The data acquisition and transmission module is configured to collect operational data from the target device and upload it to the edge computing node or cloud server via wired or wireless communication. A multi-model diagnostic engine, deployed on the edge computing node or cloud server, is configured to receive the runtime data and call one or more diagnostic models among the diagnostic unit, machine learning classifier, and deep learning model to perform parallel analysis on the runtime data to generate preliminary diagnostic results that include at least the fault type and severity level. The decision fusion and early warning center is configured to perform confidence-weighted fusion on the preliminary diagnostic results, generate a target diagnostic conclusion, and trigger a corresponding early warning operation based on the severity level of the target diagnostic conclusion. The early warning operation includes prompt information, notification information, or automatic triggering of emergency control instructions. The linkage control interface is configured to automatically execute preset emergency operations upon receiving the emergency control command. These emergency operations include switching to backup equipment or adjusting system operating parameters.

2. The system according to claim 1, characterized in that, During wireless communication, the data acquisition and transmission module dynamically selects one of the following communication protocols: 4G, 5G, LoRa, or NB-IoT, based on the communication link quality and data priority.

3. The system according to claim 1, characterized in that, The deep learning model in the multi-model diagnostic engine is a recurrent neural network, which is used to model the time sequence of equipment operation data to identify progressive mechanical faults.

4. The system according to claim 1, characterized in that, The specific steps for the decision fusion and early warning center to perform the confidence-weighted fusion include: Calculate the overall confidence score of the target diagnostic conclusion; When the overall confidence score is greater than or equal to the confidence threshold, the early warning operation is triggered. The overall confidence score is determined using the following formula: in, The overall confidence score is given. The weight factor for the i-th diagnostic model is preset in the multi-model diagnostic engine, and satisfies the following conditions: , is the confidence value output by the i-th diagnostic model for the current diagnostic result.

5. The system according to claim 4, characterized in that, The weighting factors are dynamically adjusted based on the historical diagnostic accuracy of each diagnostic model invoked.

6. The system according to claim 1, characterized in that, The system also includes a device health assessment module, which is configured as follows: The real-time health index of the target device is calculated periodically and sent to the multi-model diagnostic engine. The multi-model diagnostic engine combines the operating data of the target device and the real-time health index to analyze and output preliminary diagnostic results. The calculation process of the real-time health index includes: Obtain parameters of vibration, temperature, and load of the target equipment; Calculate the normalized deviations of vibration, temperature, and load respectively, where vibration deviation is the square of the ratio of real-time vibration amplitude to maximum permissible vibration amplitude, temperature deviation is the ratio of the absolute deviation of real-time temperature value relative to rated temperature value to maximum permissible temperature deviation, and load deviation is the ratio of real-time load value to maximum permissible load value. These normalized deviations are multiplied by their respective preset weighting coefficients and summed to obtain a comprehensive degradation index. The real-time health index is obtained by subtracting the baseline health value from the comprehensive deterioration index.

7. The system according to claim 1, characterized in that, The system also includes the remote operation interface, which provides a web or mobile access interface and supports authorized users to remotely view the device status and perform restart or reset operations.

8. The system according to claim 7, characterized in that, Before performing the emergency operation, the linkage control interface sends an operation confirmation request to the remote operation interface.

9. The system according to claim 7, characterized in that, The remote operation interface supports role-based hierarchical permission management, with different levels of users being granted different operation permissions.

10. The system according to claim 1, characterized in that, The diagnostic unit is equipped with an expert knowledge base for rapid logical judgment of short circuits and overloads in electrical faults.

Images