Operation and maintenance alarm pushing method and device based on role recognition and medium

Abstract

The invention discloses an operation and maintenance alarm pushing method and device based on role recognition and a medium, and relates to the technical field of operation and maintenance alarm. The method comprises the following steps: collecting user task state data in real time, wherein the state data comprises a task state, a user position, task associated equipment and a priority; adjusting the role authority of the user based on the state data, and predicting the processing adaptation degree of the fault equipment to obtain a user candidate pool; in response to the received operation and maintenance alarm, constructing a multi-dimensional matching model based on the user candidate pool, determining a multi-dimensional weight through an analytic hierarchy process, and generating a priority score through weighted calculation; sorting the users in the candidate user pool by using a machine learning model according to the priority scores so as to determine an alarm pushing sequence; and performing multiple permission verification on the sorted users, and pushing alarm information to the users passing the permission verification. According to the method, accurate, efficient and intelligent pushing of the operation and maintenance alarm is realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of operation and maintenance alarm, in particular to an operation and maintenance alarm pushing method based on role recognition, equipment and medium. BACKGROUND

[0002] With the rapid development of smart parks, industrial 4.0 and other fields, operation and maintenance alarm data is growing explosively. The existing alarm pushing system has been difficult to meet the precise and intelligent operation and maintenance requirements due to the limitations of the technical architecture.

[0003] At present, the mainstream scheme in the industry has significant defects: the two-dimensional pushing scheme based on static roles and alarm levels cannot perceive the changes in the real-time task state of users, resulting in a serious disconnection between alarm pushing and on-site working conditions, and the adjustment of permissions relies on manual work, with a response cycle of several hours. The nearby pushing scheme based on single-dimensional matching of geographic location, although considering the distance factor, completely ignores the skill matching degree and historical response ability of users, resulting in a large number of alarms being pushed to nearby personnel who do not have the processing capacity, with a high misprocessing rate. In addition, the existing system generally uses a fixed rule engine, lacks self-learning optimization capability, and the strategy is fixed and has a long scene adaptation cycle. At the same time, the alarm information is disconnected from the visual operation and maintenance tools such as AR, and the operation and maintenance personnel need to query the drawings and manuals additionally, resulting in low fault positioning and disposal efficiency.

[0004] Through the above analysis, the problems and defects of the prior art are: The operation and maintenance alarm pushing system in the prior art is not accurate in pushing, has long response delay, low disposal efficiency and high false alarm rate due to the static role adaptation, single matching dimension, lack of self-learning mechanism and disconnection between alarm and operation and maintenance execution. SUMMARY

[0005] The embodiments of the present application provide an operation and maintenance alarm pushing method based on role recognition, equipment and medium, which can solve the problems of inaccurate pushing, long response delay, low disposal efficiency and high false alarm rate of the operation and maintenance alarm pushing system in the prior art due to static role adaptation, single matching dimension, lack of self-learning mechanism and disconnection between alarm and operation and maintenance execution.

[0006] In a first aspect, the embodiments of the present application provide a role recognition-based operation and maintenance alarm pushing method, characterized in that the method comprises: collecting user task state data in real time, the state data comprising task state, user position, task associated equipment and priority; adjusting the role permission of the user based on the state data, and predicting the processing adaptation degree of the fault equipment to obtain a user candidate pool; in response to a received operation and maintenance alarm, constructing a multi-dimensional matching model based on the user candidate pool, determining the weight of the multi-dimensions by the analytic hierarchy process, and generating a priority score by weighted calculation; sorting the users in the candidate user pool by using a machine learning model according to the priority score to determine the alarm pushing sequence; performing multiple permission checks on the sorted users, and pushing alarm information to the users passing the permission check.

[0007] In an implementation manner of the present application, the user task state data is collected in real time, specifically comprising: obtaining the positioning and three-dimensional model data of the fault equipment in the digital twin system; pulling the task state of the user in real time in the work order management system, the state comprising processing, waiting for cooperation, idle or offline; performing real-time verification on the collected task associated equipment, and if the region and type of the task associated equipment are inconsistent with the historical record, triggering a data correction reminder and synchronously updating.

[0008] In an implementation manner of the present application, the role permission of the user is adjusted based on the state data, and the processing adaptation degree of the fault equipment is predicted to obtain a user candidate pool, specifically comprising: adjusting the role permission in real time when the task priority is changed, triggering the alarm receiving range, and judging whether to shield or enable the alarm pushing of the preset level; predicting the processing adaptation degree of the user to the fault equipment by using a logistic regression model, and generating an adaptation score based on the historical behavior data; obtaining the user candidate pool according to the user position and the current task state, the adjusted role permission, the updated alarm receiving range and the adaptation score.

[0009] In an implementation manner of the present application, the multi-dimensional matching model is constructed, specifically comprising: defining the quantification score rule for the multi-dimensions, and determining the weight distribution of the multi-dimensions based on the analytic hierarchy process; obtaining the priority score by weighted calculation according to the weight distribution of the multi-dimensions; screening the user candidate pool according to the priority score, and if the candidate pool is empty, the score threshold is reduced in a stepwise manner.

[0010] In an implementation manner of the present application, the users in the candidate user pool are sorted by using a machine learning model according to the priority score to determine the alarm pushing sequence, specifically comprising: obtaining historical alarm data, extracting multi-dimensional feature vectors in the historical alarm data to construct a model training set; training a user sorting model by using a gradient boosting decision tree algorithm, taking the response quality of the user to the historical alarm as a supervision signal; inputting the real-time features of the users in the candidate user pool into the sorting model to output a descendingly arranged user sequence.

[0011] In an implementation form of the present application, the alarm information is pushed to the user passing the permission check, specifically comprising: deriving a 3D model of the fault equipment and fault point annotation data in the digital twin system; packaging the standard operation procedure animation and the alarm information, and pushing the packaged data packet to the user terminal.

[0012] In an implementation form of the present application, the sorted users are subjected to multiple permission checks, specifically comprising: checking whether the user has the role permission of the fault equipment; checking whether the current task state of the user allows receiving a new alarm; checking whether the user is within a preset working time period; if any of the checks fails, the alarm information is transferred to the candidate pool of the second high priority user.

[0013] In an implementation form of the present application, the method further comprises: collecting behavior data of the user on the historical alarm in real time, the behavior data including clicking, ignoring, delayed processing and processing completion; triggering incremental training of the model based on the behavior data, and dynamically adjusting the push weight of the alarm information.

[0014] In a second aspect, the embodiments of the present application also provide a device for operation and maintenance alarm pushing based on role recognition, the device comprising at least one processor; and a memory in communication connection with the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to: collect task state data of the user in real time, the state data including task state, user location, task associated equipment and priority; adjust the role permission of the user based on the state data, and predict the processing adaptation degree of the fault equipment to obtain a user candidate pool; in response to the received operation and maintenance alarm, constructing a multi-dimensional matching model based on the user candidate pool, determining the weight of the multi-dimension by the analytic hierarchy process, and generating a priority score by weighted calculation; sorting the users in the candidate user pool by using a machine learning model according to the priority score to determine the alarm pushing sequence; performing multiple permission checks on the sorted users, and pushing the alarm information to the user passing the permission check.

[0015] In a third aspect, the embodiments of the present application further provide a non-volatile computer storage medium based on role recognition for operation and maintenance alarm pushing, which stores computer executable instructions, and the computer executable instructions are configured to: collect user task state data in real time, the state data including task state, user location, task associated equipment and priority; adjust the role permission of the user based on the state data, and predict the processing adaptation degree of the fault equipment to obtain a user candidate pool; in response to the received operation and maintenance alarm, based on the user candidate pool, a multi-dimensional matching model is constructed, the weights of the multi-dimensions are determined by the analytic hierarchy process, and the priority score is generated by weighted calculation; according to the priority score, the users in the candidate user pool are sorted by using a machine learning model to determine the alarm pushing sequence; the sorted users are subjected to multiple permission checks, and the alarm information is pushed to the users passing the permission check.

[0016] The embodiments of the present application provide a role recognition based operation and maintenance alarm pushing method, device and medium, which realizes accurate, efficient and intelligent pushing of operation and maintenance alarms by constructing a full-link technology closed loop of dynamic role recognition, multi-dimensional weight matching, machine learning optimization and operation and maintenance linkage. Specifically, by introducing a five-dimensional matching model and the analytic hierarchy process, the role, location, historical ability, equipment association and emergency degree are comprehensively considered, so that the accuracy of the pushing object screening is improved; the XGBoost model is used and a user behavior feedback closed loop is established, so that the system has self-learning ability and realizes adaptive optimization of the strategy; finally, through the deep linkage of alarm and AR visualization guidance, the fault positioning time is shortened, and the risk of unauthorized operation is completely eliminated by means of the threefold permission check mechanism, thereby greatly improving the timeliness of operation and maintenance response, the accuracy of disposal and the security of the system as a whole. BRIEF DESCRIPTION OF DRAWINGS

[0017] The accompanying drawings, which are included to provide a further understanding of the present application, form a part of the present application and illustrate the illustrative embodiments of the present application and the description thereof, and do not constitute improper limitations to the present application. In the drawings: Figure 1 A flowchart of a role recognition based operation and maintenance alarm pushing method provided by the embodiments of the present application; Figure 2 An internal structure schematic diagram of a role recognition based operation and maintenance alarm pushing device provided by the embodiments of the present application. DETAILED DESCRIPTION

[0018] In order to make the purposes, technical solutions and advantages of the present application clearer, the technical solutions of the present application will be described clearly and completely below in combination with specific embodiments of the present application and corresponding drawings. Obviously, the described embodiments are only some of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor fall within the scope of protection of the present application.

[0019] The embodiments of the present application provide a role recognition-based operation and maintenance alarm pushing method, device and medium, and solve the problems of inaccurate pushing, long response delay, low disposal efficiency and high false alarm rate caused by the static role adaptation, single matching dimension, lack of self-learning mechanism and disconnection between alarms and operation and maintenance execution of the operation and maintenance alarm pushing system in the prior art.

[0020] The technical solutions of the embodiments of the present application will be described in detail below with reference to the drawings.

[0021] Figure 1 A role recognition-based operation and maintenance alarm pushing method flowchart is provided in the embodiments of the present application. As shown in the figure, Figure 1 The role recognition-based operation and maintenance alarm pushing method provided by the embodiments of the present application specifically includes the following steps: Step 10: Real-time collection of user task state data, including task state, user location, task associated equipment and priority; As an optional embodiment, real-time collection of user task state data can specifically include: step 101: acquisition of positioning and three-dimensional model data of a fault equipment in a digital twin system; step 102: real-time pulling of a task state of a user in a work order management system, including processing, waiting for cooperation, idle or offline; step 103: real-time verification of the collected task associated equipment, if the region, type and historical record of the task associated equipment are inconsistent, triggering a data correction reminder and synchronous update.

[0022] In this step, the digital twin system and the work order management system are connected through RESTful API, and user task data is collected in real time, including: task state: processing, waiting for cooperation, idle, offline; task associated equipment: equipment ID, region, equipment type; task priority: P0 for network interruption, P1 for regional interruption, P2 for single equipment fault, P3 for early warning; the data is encapsulated in JSON format and transmitted in real time through WebSocket protocol.

[0023] Step 20: Adjustment of the role permission of the user based on the state data, and prediction of the processing adaptation degree to the fault equipment, to obtain a user candidate pool; In this step, the dynamic permission rule is extended based on the RBAC (Role-Based Access Control) model, combined with the real-time location of the user, through LBS / UWB positioning, and the role range is adjusted according to the task state: the on-site engineer U1 originally responsible for the equipment in area A temporarily assists in the maintenance of area B, and the system automatically gives U1 the alarm receiving permission of the equipment in area B, and suspends the P2 / P3 level non-emergency alarm push in area A; Scene 2: When the supervisor U4 is in a cross-department meeting state, the system only pushes P0 / P1 level emergency alarms to him, and shields P2 / P3 level alarms; The permission adjustment is automatically executed by the rule engine without manual intervention.

[0024] As an optional embodiment, the role permission of the user is adjusted based on the state data, and the processing adaptation degree of the user to the fault equipment is predicted to obtain a user candidate pool, which can specifically include: step 201: when the task priority is changed, the role permission is adjusted in real time, the alarm receiving range is triggered, and it is judged whether to shield or enable the alarm push of a preset level; step 202: using a logistic regression model to predict the processing adaptation degree of the user to the fault equipment, and generating an adaptation score based on historical behavior data; step 203: obtaining a user candidate pool according to the user location and the current task state, and the adjusted role permission, the updated alarm receiving range and the adaptation score.

[0025] In this step, when the task priority changes from P2 level to P1 level or from P1 level to P2 level, the user role permission is adjusted in real time based on the preset role permission mapping rule; when the user's current processing task is upgraded to P0 level network interruption fault, temporary cross-region equipment scheduling permission is given to the user; the recalculation of the alarm receiving range is triggered synchronously, the geographical boundary and the equipment range of the alarm that the user can receive are updated according to the matching relationship between the task associated equipment, the user real-time location and the task priority; At the same time, it is judged whether to shield or enable the alarm push of a preset level, for example, when the user is processing P0 level fault, P3 level early warning type alarm is automatically shielded, and only P0 / P1 level emergency alarm receiving permission is reserved.

[0026] Further, a logistic regression model is constructed to predict the processing adaptation degree of the user to the fault equipment, which takes the historical behavior data of the user as input features, including 12-dimensional features such as the number of times of processing the same fault in the past 3 months, the average solution time, the success rate, the skill label matching degree, and the training examination results; The model is trained through 5-fold cross-validation; Based on the model output result, an adaptation score of 0-100 points is generated, wherein the score ≥70 points indicates that the user has a higher adaptation to process the fault, 40-69 points indicates that the adaptation is general, and <40 points indicates that the adaptation is lower.

[0027] Optionally, the geographical position is within the updated alarm receiving range, such as indoor UWB positioning ≤100 meters or outdoor LBS positioning ≤500 meters; the current task state is idle, waiting for cooperation or processing but the task priority is lower than the new alarm priority; the adjusted role permission contains the disposal permission of the fault equipment, and the core network equipment only allows senior engineers and above roles to receive; the adaptation score is ≥60 points, and if the number of candidate pool users after initial screening is <3, it is reduced to ≥50 points; finally, the users who meet the above conditions at the same time are included in the candidate pool, and key parameters such as permission level, position distance, and adaptation score of each user are recorded, providing basic data for subsequent priority sorting.

[0028] Step 30: In response to the received operation and maintenance alarm, a multi-dimensional matching model is constructed based on the user candidate pool, the weights of the multi-dimensions are determined by the analytic hierarchy process, and the priority score is generated by weighted calculation. As an optional embodiment, the multi-dimensional matching model is constructed, which can specifically include: step 301: defining the quantitative scoring rules for multi-dimensions, and determining the weight distribution of multi-dimensions based on the analytic hierarchy process; step 302: according to the weight distribution of multi-dimensions, the priority score is obtained by weighted calculation; step 303: according to the priority score, the user candidate pool is screened, and if the candidate pool is empty, the score threshold is reduced step by step.

[0029] In this step, for the five core dimensions of role adaptation, geographical position, historical response ability, device association degree, and emergency degree, quantitative standards of 0-100 points are respectively formulated: role adaptation: score according to the matching degree of the current role of the user and the fault processing requirement; geographical position: score based on the straight-line distance between the real-time position of the user and the fault equipment; historical response ability: weighted calculation combined with the average response time and solution success rate of similar faults in the past three months; device association degree: score according to the frequency of the user's historical processing of the device; emergency degree: directly mapped according to the alarm priority. By constructing a judgment matrix, the weight of each dimension is calculated: for example, emergency degree 30%, role adaptation 25%, historical response ability 20%, geographical position 15%, and device association degree 10%, and the consistency test is passed to ensure the rationality of the weight distribution.

[0030] Further, for each user in the user candidate pool, extract its quantitative score in 5 dimensions, such as a user role adaptation degree of 80 points, a geographic location of 100 points, a historical response capability of 90 points, a device association degree of 80 points, and an emergency degree of 100 points; calculate the score according to the formula priority score = Σ (dimension score x dimension weight), for example: 80x25% + 100x15% + 90x20% + 80x10% + 100x30% = 90 points. Set the initial screening threshold to 60 points, and retain users with a priority score ≥ 60 points in the candidate pool; if the number of users in the candidate pool after screening is < 2, which is not enough to support alarm dispatching, then reduce the threshold by 5 points each time, to 55 points, 50 points, …, until the number of users in the candidate pool ≥ 2; if the threshold is reduced to 30 points and there is still no user meeting the conditions, then trigger the range expansion mechanism, temporarily include users with basic permissions in adjacent areas, and mark the cross-area support label in the score.

[0031] Step 40: According to the priority score, use a machine learning model to sort the users in the candidate user pool to determine the alarm push order; As an optional embodiment, according to the priority score, use a machine learning model to sort the users in the candidate user pool to determine the alarm push order, which can specifically include: step 401: acquire historical alarm data, extract multi-dimensional feature vectors in the historical alarm data to construct a model training set; In this step, 23 features are extracted, covering three categories to solve the problem of insufficient feature dimensions in traditional systems: alarm features: fault type: PLC / frequency converter / server, level P0-P3, location area ID / coordinate, impact range, device quantity; user features: role type, skill label (10+ categories), historical response time, success rate of similar faults, current task status; environmental features: current period, system load rate, weather conditions; numerical features use Min-Max standardization, classification features use One-Hot encoding, and missing values are filled by the mean value of similar users.

[0032] Step 402: Use a gradient boosting decision tree algorithm to train a user sorting model with the response quality of users to historical alarms as the supervision signal; In this step, an XGBoost gradient boosting tree model is used to sort users according to the response quality of users to alarms, with the target variable being the success rate x (1-response time / baseline time), to achieve user priority sorting; compared with traditional rule engines, the model can capture non-linear relationships between features; the model updates use a rolling update mechanism without interrupting service.

[0033] Step 403: Input the real-time features of users in the candidate user pool into the sorting model to output a descendingly arranged user sequence.

[0034] Step 50: Multi-privilege check is performed on the sorted users, and alarm information is pushed to the users passing the privilege check.

[0035] As an optional embodiment, the multi-privilege check on the sorted users can specifically include: step 501: checking whether the user has the role privilege of the faulty device; In this step, it is checked whether the user has the disposal privilege of the faulty device, such as the park power distribution room PLC allowing only senior engineers and above to dispose; the ordinary sensor such as the temperature and humidity sensor allowing outsourcing personnel to dispose; the privilege information is stored in the privilege management database, supporting dynamic update, such as automatically upgrading the privilege after the user is promoted, and eliminating the risk of outsourcing personnel operating core equipment.

[0036] Step 502: checking whether the current task state of the user allows receiving a new alarm; In this step, if the user is in the state of handling P0-level emergency fault, only higher-level alarms are pushed, and no higher-level alarms are pushed; if the user is in the state of idle or handling low-priority task, normal alarms meeting the conditions are pushed; if the user is offline or on leave, the user is automatically skipped, and the alarm is pushed to the next high-priority user in the candidate pool; for example, when an engineer is handling a P1-level fault, the system does not push P2-level alarms to him, avoiding task conflicts.

[0037] Step 503: checking whether the user is in a preset working period; In this step, it is checked whether the user is in a preset working period, and the normal working period is 8:00-20:00 on weekdays, allowing receiving all alarms; only P0 / P1-level emergency alarms are pushed during non-working hours, and double reminders through SMS and APP are required.

[0038] Step 504: if any check fails, the alarm information is transferred to the next high-priority user in the candidate pool.

[0039] In this step, the alarm is pushed to the user only after the three checks are passed; when any check fails, the alarm is automatically transferred to the next high-priority user in the candidate pool, ensuring that the alarm disposal is not interrupted.

[0040] As an optional embodiment, the alarm information is pushed to the user passing the privilege check, which can specifically include: step 505: exporting the 3D model of the faulty device and the fault point labeling data in the digital twin system; step 506: packaging the standard operation procedure animation and the alarm information, and pushing the packaged data packet to the user terminal.

[0041] In this step, the device 3D model interface is called from the digital twin platform to export a lightweight GLB format model of the faulty device; based on the fault location data, the fault point is automatically labeled in the 3D model, and the fault description information is associated; the exported model and the labeled data are checked, if the model does not match the actual model of the device or the labeling position deviation is greater than 0.5 meters, an artificial calibration reminder is triggered and is synchronously updated to the digital twin system.

[0042] The SOP database is called to match the step-by-step animation corresponding to the fault type, and the data packet is packaged according to the alarm header information + 3D model + fault point labeling + SOP animation + voice prompt structure, wherein the alarm header information includes the fault device ID, occurrence time, emergency level, and current state, such as continuous alarm or has been relieved; the push protocol is adaptively selected according to the user terminal type, and the breakpoint resume transmission is supported, when the network is interrupted and recovered, the unfinished data packet is automatically transmitted.

[0043] As an optional embodiment, the method can further include: collecting user behavior data on historical alarms in real time to obtain response quality, the behavior data including clicking, ignoring, delayed processing, and processing completion; triggering model incremental training based on the behavior data to dynamically adjust the push weight of the alarm information.

[0044] In this step, user behavior data is recorded through the burying point technology, including: clicking to view, recording the clicking time and staying time, and marking the ignoring reasons, such as non-duty range, already handled by someone, delayed processing, and processing completion; a response quality scoring model is constructed based on the behavior data: response quality = (solution success rate x 60%) + (1-response time / standard time) x 30% + (step compliance x 10%), wherein the standard time is preset according to the fault type, the standard time of P0 level fault is 30 minutes, and the standard time of P1 level is 60 minutes, and the scoring range is 0-100 points.

[0045] When the user response data of a certain type of fault accumulates ≥100 or the response quality score of a single user is <60 for 3 consecutive times, incremental training is triggered; the incremental training uses a sliding window mechanism to input the features in the behavior data into a machine learning ranking model to retrain the model parameters.

[0046] The above is the method embodiment of the present application. Based on the same inventive concept, the present application also provides an operation and maintenance alarm push device based on role recognition, which has a structure as shown in Figure 2 .

[0047] Figure 2 An internal structure diagram of an operation and maintenance alarm push device based on role recognition provided by the present application is shown in Figure 2 . As shown in , the device includes: and a memory 202 connected with the at least one processor in communication; The memory 202 stores instructions executable by the at least one processor 201, and the instructions are executed by the at least one processor 201 to enable the at least one processor 201 to: collect user task state data in real time, the state data including a task state, a user position, a task associated device, and a priority; adjust a role permission of the user based on the state data, and predict a processing adaptation degree for a fault device to obtain a user candidate pool; in response to a received operation and maintenance alarm, construct a multi-dimensional matching model based on the user candidate pool, determine weights of the multi-dimensions by an analytic hierarchy process, and generate a priority score by weighted calculation; sort users in the candidate user pool by using a machine learning model according to the priority score, to determine an alarm pushing sequence; and perform multiple permission verifications on the sorted users, and push alarm information to the users passing the permission verification.

[0048] Some embodiments of the present application provide a non-volatile computer storage medium for operation and maintenance alarm pushing based on role recognition corresponding to Figure 1 The non-volatile computer storage medium stores computer executable instructions, and the computer executable instructions are configured to: collect user task state data in real time, the state data including a task state, a user position, a task associated device, and a priority; adjust a role permission of the user based on the state data, and predict a processing adaptation degree for a fault device to obtain a user candidate pool; in response to a received operation and maintenance alarm, construct a multi-dimensional matching model based on the user candidate pool, determine weights of the multi-dimensions by an analytic hierarchy process, and generate a priority score by weighted calculation; sort users in the candidate user pool by using a machine learning model according to the priority score, to determine an alarm pushing sequence; and perform multiple permission verifications on the sorted users, and push alarm information to the users passing the permission verification.

[0049] Each of the embodiments in the present application is described in a progressive manner, and the same or similar parts of each of the embodiments can be referred to each other. Each of the embodiments focuses on the difference from other embodiments. In particular, for the Internet of Things device and medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the part of the method embodiments.

[0050] The system and medium provided by the embodiments of the present application are one-to-one corresponding to the method, and therefore, the system and medium also have similar beneficial technical effects to the method. Since the beneficial technical effects of the method have been described in detail above, the beneficial technical effects of the system and medium will not be described here.

[0051] Those skilled in the art will appreciate that embodiments of the present application can be readily used as software, hardware, or a combination of software and hardware. In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0052] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 one or more functions specified in the flowchart block or blocks. Figure 1 means for functionally implementing one or more functions specified in the flowchart block or blocks.

[0053] These computer program instructions can also be stored in a computer readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer readable memory produce an article of manufacture including instructions which implement the function specified in the flowchart block or blocks. Figure 1 one or more functions specified in the flowchart block or blocks. Figure 1 means for functionally implementing one or more functions specified in the flowchart block or blocks.

[0054] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 one or more functions specified in the flowchart block or blocks. Figure 1 means for functionally implementing one or more functions specified in the flowchart block or blocks.

[0055] In one typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0056] The memory can include non-persistent memory and / or persistent memory, such as flash memory, read-only memory (ROM), and / or volatile or non-volatile random access memory (RAM), among others. The memory is an example of computer readable media.

[0057] Computer-readable media includes permanent and non-permanent, movable and non-movable media that can be implemented by any method or technology to store information. The information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassette, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information accessible to a computing device. According to the definition herein, computer-readable media does not include transitory media such as modulated data signals and carriers.

[0058] It should also be noted that the terms "comprising", "containing", or any other variant thereof are intended to cover non-exclusive inclusions, so that a process, method, article or apparatus that includes a list of elements does not only include those elements, but also includes other elements not explicitly listed, or further includes elements inherent in such a process, method, article or apparatus. Without more limitations, the element defined by the statement "comprising a" does not exclude the presence of additional identical elements in the process, method, article or apparatus that includes the element.

[0059] The above only describes the embodiments of the present application and is not intended to limit the present application. For those skilled in the art, the present application can have various modifications and changes. Any modification, equivalent replacement, improvement, etc. within the spirit and principles of the present application shall be included in the scope of claims of the present application.

Claims

1. A method for pushing operation and maintenance alarms based on role recognition, characterized in that, The method includes: Real-time collection of user task status data, including task status, user location, task-associated devices, and priority; Based on the status data, the user's role permissions are adjusted, and the adaptability to the handling of faulty devices is predicted to obtain a user candidate pool. In response to the received operation and maintenance alarm, a multi-dimensional matching model is constructed based on the user candidate pool, the weights of the multi-dimensional model are determined by the analytic hierarchy process, and a priority score is generated by weighted calculation. Based on the priority score, a machine learning model is used to sort the users in the candidate user pool to determine the order of alarm pushes; Perform multiple permission checks on the sorted users and push alarm information to users who pass the permission checks.

2. The operation and maintenance alarm push method based on role recognition according to claim 1, characterized in that, The real-time collection of user task status data specifically includes: The location and 3D model data of the faulty device are obtained through a digital twin system; The task status of the user is retrieved in real time from the work order management system. The status includes processing, pending collaboration, idle or offline. The collected task-associated devices are verified in real time. If the region and type of the task-associated device are inconsistent with the historical records, a data correction reminder is triggered and the data is updated synchronously.

3. The operation and maintenance alarm push method based on role recognition according to claim 1, characterized in that, The process of adjusting user role permissions based on the status data and predicting the adaptability to handling faulty devices to obtain a user candidate pool specifically includes: When the task priority changes, the role permissions are adjusted in real time, the alarm receiving range is triggered, and it is determined whether to block or enable alarm push at a preset level. The logistic regression model is used to predict the user's adaptability to the faulty equipment, and an adaptability score is generated based on historical behavior data. The user candidate pool is obtained based on the user's location and current task status, as well as the adjusted role permissions, updated alarm reception range, and adaptability score.

4. The operation and maintenance alarm push method based on role recognition according to claim 1, characterized in that, The construction of the multi-dimensional matching model specifically includes: Quantitative scoring rules are defined for the multi-dimensional data, and the weight allocation of the multi-dimensional data is determined based on the analytic hierarchy process (AHP). The priority score is obtained by weighted calculation based on the weight allocation of the multi-dimensional dimensions. The user candidate pool is filtered according to the priority score, and if the candidate pool is empty, the score threshold is reduced in a stepwise manner.

5. The operation and maintenance alarm push method based on role recognition according to claim 1, characterized in that, The step of sorting users in the candidate user pool according to the priority score using a machine learning model to determine the alarm push order specifically includes: Acquire historical alarm data, and extract multi-dimensional feature vectors from the historical alarm data to construct a model training set; The gradient boosting decision tree algorithm is used to train a user ranking model with the quality of user responses to historical alarms as a supervision signal. The real-time features of users in the candidate user pool are input into the ranking model, and the user sequence is output in descending order.

6. The operation and maintenance alarm push method based on role recognition according to claim 2, characterized in that, The process of pushing alert information to users who have passed the permission verification specifically includes: Export the 3D model of the faulty device and the fault point annotation data from the digital twin system; The standard operating procedure animation and alarm information are encapsulated, and the encapsulated data packet is pushed to the user terminal.

7. The operation and maintenance alarm push method based on role recognition according to claim 1, characterized in that, The process of performing multiple permission checks on the sorted users specifically includes: Verify whether the user has the necessary role permissions for the faulty device; Verify whether the user's current task status allows receiving new alarms; Verify whether the user is within the preset working hours; If any verification fails, the alarm information will be forwarded to the next highest priority user in the candidate pool.

8. The operation and maintenance alarm push method based on role recognition according to claim 5, characterized in that, The method further includes: Real-time collection of user behavior data in response to the historical alarms has yielded the response quality. The behavior data includes clicks, ignores, delayed processing, and processing completion. Incremental training of the model is triggered based on the behavioral data, and the push weight of the alarm information is dynamically adjusted.

9. A maintenance alarm push device based on role recognition, characterized in that, The device includes: At least one processor; And, a memory communicatively connected to the at least one processor; The memory stores instructions executable by the at least one processor, which, when executed by the at least one processor, enable the at least one processor to: Real-time collection of user task status data, including task status, user location, task-associated devices, and priority; Based on the status data, the user's role permissions are adjusted, and the adaptability to the handling of faulty devices is predicted to obtain a user candidate pool. In response to the received operation and maintenance alarm, a multi-dimensional matching model is constructed based on the user candidate pool, the weights of the multi-dimensional model are determined by the analytic hierarchy process, and a priority score is generated by weighted calculation. Based on the priority score, a machine learning model is used to sort the users in the candidate user pool to determine the order of alarm pushes; Perform multiple permission checks on the sorted users and push alarm information to users who pass the permission checks.

10. A non-volatile computer storage medium for operation and maintenance alarm push based on role recognition, storing computer-executable instructions, characterized in that, The computer-executable instructions are set as follows: Real-time collection of user task status data, including task status, user location, task-associated devices, and priority; Based on the status data, the user's role permissions are adjusted, and the adaptability to the handling of faulty devices is predicted to obtain a user candidate pool. In response to the received operation and maintenance alarm, a multi-dimensional matching model is constructed based on the user candidate pool, the weights of the multi-dimensional model are determined by the analytic hierarchy process, and a priority score is generated by weighted calculation. Based on the priority score, a machine learning model is used to sort the users in the candidate user pool to determine the order of alarm pushes; Perform multiple permission checks on the sorted users and push alarm information to users who pass the permission checks.

Images