Maintenance method and device for cutting equipment, equipment and storage medium

Abstract

The invention relates to the technical field of intelligent equipment maintenance, and discloses a cutting equipment maintenance method and device, equipment and a storage medium, and the method comprises the steps: receiving a problem request of target equipment; acquiring operation data corresponding to the equipment identification information according to the equipment identification information; determining at least one candidate fault reason based on the fault description information and a preset knowledge graph; in combination with the real-time operation data and the historical operation data, performing confidence scoring on each candidate fault reason to obtain a confidence score corresponding to each candidate fault reason; and sorting the candidate fault causes based on the confidence score, and generating a maintenance scheme based on a sorting result. According to the method, the problem request is received, analysis is performed based on the preset knowledge graph, accurate fault diagnosis is realized, a user can reduce dependence on professional staff, the fault can be solved by himself / herself, and the fault positioning accuracy is improved, so that the maintenance efficiency of the cutting equipment is improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent equipment maintenance technology, specifically to maintenance methods, devices, equipment, and storage media for cutting equipment. Background Technology

[0002] Cutting equipment is specialized equipment used for material segmentation and cutting, especially flexible material cutting equipment, which is widely used in industries such as leather, textiles, and composite materials. It is a key processing unit in flexible manufacturing systems. Cutting equipment typically integrates high-precision servo drives, vision positioning, intelligent feeding and cutting execution mechanisms, and is equipped with various sensors to achieve closed-loop control and process monitoring. As the manufacturing industry develops towards intelligence and flexibility, cutting equipment is transforming from a single mechanical execution unit into a complex electromechanical system with multiple intelligent agents working together.

[0003] Traditional after-sales support for cutting equipment relies mainly on manual experience, telephone communication, or on-site service, resulting in slow response times, high costs, and difficulty in systematically accumulating fault solutions. With the intelligent development of flexible manufacturing systems, equipment integrates multiple types of sensors and actuators, generating a large amount of operational data. However, existing after-sales systems have failed to fully integrate real-time data, historical cases, and knowledge bases, leading to inaccurate problem localization, unintuitive solutions, and low maintenance efficiency. Summary of the Invention

[0004] In view of this, the present invention provides a method, apparatus, device and storage medium for maintaining cutting equipment, so as to solve the problem of low maintenance efficiency of cutting equipment.

[0005] In a first aspect, the present invention provides a method for maintaining a cutting device, the method comprising: Receive a problem request from the target device, the problem request including at least fault description information and device identification information, the device identification information corresponding to the target device; Based on the device identification information, obtain the operating data corresponding to the device identification information, the operating data including real-time operating data and historical operating data; Based on the fault description information and the preset knowledge graph, at least one candidate fault cause is determined; wherein, the preset knowledge graph at least associates and stores fault phenomena, fault causes, component information, component inspection steps, and solutions corresponding to fault causes. By combining the real-time operation data and historical operation data, a confidence score is assigned to each candidate fault cause to obtain a confidence score corresponding to each candidate fault cause; wherein, the confidence score incorporates at least: the correlation strength between the candidate fault cause and the fault description information in the preset knowledge graph, the matching degree between the real-time operation data and the preset fault mode corresponding to the candidate fault cause, and the success rate of the candidate fault cause in historical maintenance cases; The candidate fault causes are ranked based on the confidence score, and a repair plan is generated based on the ranking result. The repair plan includes at least one or more components to be investigated, inspection steps for the components to be investigated, and a recommended inspection order.

[0006] The maintenance method for cutting equipment provided in this embodiment includes: receiving a problem request from the target equipment; obtaining operating data corresponding to the equipment identification information based on the equipment identification information; determining at least one candidate fault cause based on fault description information and a preset knowledge graph; combining real-time operating data and historical operating data to perform confidence scoring on each candidate fault cause to obtain a confidence score corresponding to each candidate fault cause; sorting the candidate fault causes based on the confidence scores, and generating a maintenance plan based on the sorting results.

[0007] This method receives problem requests and analyzes them based on a pre-set knowledge graph to achieve accurate fault diagnosis. Users can reduce their reliance on professional staff, resolve faults themselves, improve the accuracy of fault location, and thus improve the maintenance efficiency of cutting equipment.

[0008] In one optional implementation, the step of combining the real-time operating data and historical operating data to perform confidence scoring on each candidate fault cause, thereby obtaining a confidence score corresponding to each candidate fault cause, includes: For any candidate cause of failure, perform the following steps: The correlation strength between the candidate fault cause and the fault description information in the preset knowledge graph is used as the graph correlation score. Obtain a preset fault mode corresponding to the candidate fault cause, calculate the matching degree between the real-time running data and the preset fault mode, and obtain a real-time matching score; The success rate of resolving the candidate fault causes in historical maintenance cases is statistically analyzed from the historical operation data and used as the historical resolution score. The historical operation data includes historical maintenance cases. By combining the graph association score, real-time matching score, and historical resolution score, a confidence score is obtained for the candidate fault cause.

[0009] In one optional implementation, the confidence score is calculated according to the following formula: Confidence score = w1 × GraphWeight(F) i , S) +w2×DataMatch(F i D t +w3×HistorySuccessRate(F i ) Among them, GraphWeight(F i S) represents the candidate fault cause F i The graph association score between the fault description information S and the fault description information S, where w1 represents the first weighting coefficient, and D t Represents real-time running data, DataMatch(F) i D t ) represents the real-time matching score, w2 represents the second weight coefficient, and HistorySuccessRate(F) represents the second weight coefficient. i ) represents the historical solution score, w3 represents the third weight coefficient, and w1 + w2 + w3 = 1.

[0010] In one optional implementation, after performing fault analysis based on the fault description information, operational data, and a preset knowledge graph to generate a maintenance plan, the method further includes: Obtain feedback on the execution results of the aforementioned repair plan; The preset knowledge graph is updated based on the execution results.

[0011] In one optional implementation, updating the preset knowledge graph based on the execution result feedback includes: If the execution result feedback indicates that the repair was successful, the association weight between the fault description information, the adopted solution, and the actual fault cause in the preset knowledge graph is enhanced. If the execution result feedback indicates that the repair has failed, the actual cause of the failure and the actual solution will be integrated into the preset knowledge graph.

[0012] In an optional implementation, the method further includes: In response to a remote assistance request, establish a connection with the remote expert's terminal; The real-time operating data and on-site video of the target device are pushed to the remote expert terminal; The system receives and displays guidance information from the remote expert terminal. The guidance information includes at least one of the following: virtual markers on the live video, voice guidance instructions, and guidance animations.

[0013] In an optional implementation, the method further includes: Based on the operating data of the target device and the preset life prediction model, calculate the remaining service life of at least one component; If the remaining service life is lower than a preset threshold, a maintenance reminder is generated; wherein the maintenance reminder includes at least information about the component to be maintained and a suggested maintenance time.

[0014] In a second aspect, the present invention provides a maintenance device for a cutting device, the device comprising: A request receiving module is used to receive a problem request from a target device. The problem request includes at least fault description information and device identification information, and the device identification information corresponds to the target device. The operation data acquisition module is used to acquire operation data corresponding to the device identification information based on the device identification information; The fault analysis module is used to determine at least one candidate fault cause based on the fault description information and a preset knowledge graph; wherein the preset knowledge graph at least associates and stores fault phenomena, fault causes, component information, component inspection steps, and solutions corresponding to the fault causes. The scoring calculation module is used to combine the real-time operation data and historical operation data to score the confidence of each candidate fault cause, thereby obtaining a confidence score corresponding to each candidate fault cause; wherein, the confidence score incorporates at least: the correlation strength between the candidate fault cause and the fault description information in the preset knowledge graph, the matching degree between the real-time operation data and the preset fault mode corresponding to the candidate fault cause, and the success rate of the candidate fault cause in historical maintenance cases; The solution determination module is used to rank the candidate fault causes based on the confidence score, and generate a repair solution based on the ranking result. The repair solution includes at least one or more components to be investigated, inspection steps for the components to be investigated, and a recommended inspection order.

[0015] Thirdly, the present invention provides a computer device, comprising: a memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, and the processor executing the computer instructions to perform the maintenance method of the cutting device described in the first aspect or any corresponding embodiment thereof.

[0016] Fourthly, the present invention provides a computer-readable storage medium storing computer instructions for causing a computer to perform the maintenance method of the cutting device according to the first aspect or any corresponding embodiment thereof. Attached Figure Description

[0017] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating a maintenance method for a cutting device according to an embodiment of the present invention; Figure 2 This is a structural block diagram of a maintenance device for a cutting device according to an embodiment of the present invention; Figure 3 This is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Cutting equipment is a specialized cutting device suitable for processing flexible materials such as leather, textiles, and composite materials. This type of equipment integrates high-precision servo drives, machine vision positioning, intelligent feeding, and cutting execution mechanisms, and is equipped with various sensors for closed-loop control and real-time process monitoring. As the manufacturing industry continues to evolve towards intelligence and flexibility, cutting equipment has gradually evolved from traditional independent mechanical execution units into complex mechatronic systems based on multi-agent collaboration, generating massive amounts of timing and status data during its operation.

[0021] Currently, the after-sales technical support system for cutting equipment still largely follows the traditional model, mainly relying on the personal experience of maintenance personnel, telephone communication, or on-site troubleshooting by engineers. This after-sales maintenance model suffers from problems such as long response delays, high service costs, and low fault handling efficiency, and it is difficult to systematically accumulate and reuse effective maintenance knowledge and solutions. Although modern cutting equipment has rich data acquisition capabilities and can generate a large amount of operating status and process parameters in real time, fault diagnosis still relies on subjective judgment, solutions lack operability, and the maintenance process involves repeated trial and error, thus limiting overall operation and maintenance efficiency.

[0022] Based on this, the present invention provides a maintenance method for cutting equipment, applied to an intelligent after-sales support system for cutting equipment. This intelligent after-sales support system is deployed in the cloud, which serves as a remote intelligent service platform. It interacts with the local control system deployed at the cutting equipment site via a secure network connection for data and command exchange. This method primarily uses the cloud system as the execution entity and can be used with intelligent terminal devices.

[0023] According to an embodiment of the present invention, a method for maintaining a cutting device is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0024] This embodiment provides a maintenance method for a cutting device, used in the aforementioned cloud-based intelligent after-sales support system. Figure 1 This is a flowchart of a maintenance method for a cutting device according to an embodiment of the present invention, such as... Figure 1 As shown, the process includes the following steps: Step S101: Receive a problem request from the target device.

[0025] The problem request should include at least a fault description and a device identification information, with the device identification information corresponding to the target device.

[0026] The target device refers to the cutting equipment that is currently experiencing an anomaly or requires maintenance. The problem request is a structured request data packet initiated by the user that describes the problem with the equipment. The user can be a mechanic, operator, or technician.

[0027] Specifically, users can input problem requests via voice, text, or other means through the human-computer interaction interface of a smart terminal. The problem request includes a fault description and equipment identification information. The fault description describes the fault phenomenon of the target equipment, and the equipment identification information identifies the target equipment. Each cutting device has a unique corresponding equipment identification information. For example, the fault description is "material shift during cutting".

[0028] Step S102: Obtain the operating data corresponding to the device identification information based on the device identification information.

[0029] Operational data includes real-time operational data and historical operational data. Operational data comprises various parameters and records generated during the operation of the target equipment, reflecting its status. Real-time operational data represents the operating parameters of each component of the target equipment at the requested time or within the most recent time window (e.g., the last 5 minutes). Historical operational data represents the target equipment's operational records, process parameters, maintenance records, etc., over a past period (e.g., the past 30 days).

[0030] Based on the equipment identification information, operational data related to the target equipment is collected. Real-time operational data may include real-time parameters such as the current of the cutter head servo motor, vibration accelerometer data, and spindle speed. Recent historical operational data may include daily average load, cumulative working time of key components, and historical records of similar vibration alarms. The cloud system, acting as the requesting party, requests operational data from the cutting equipment.

[0031] Step S103: Based on the fault description information and the preset knowledge graph, determine at least one candidate fault cause.

[0032] The preset knowledge graph at least associates and stores fault phenomena, fault causes, component information, component inspection steps, and solutions corresponding to fault causes.

[0033] The pre-built knowledge graph, stored in the cloud, is a pre-constructed knowledge base for the cutting equipment domain, organized in a graph structure. Entities include fault phenomena, fault causes, component information, component inspection steps, and corresponding solutions for the fault causes. These entities are interconnected through relationships. The fault description information is semantically matched with the fault phenomenon entities in the pre-built knowledge graph, and can be associated with corresponding component nodes. Path reasoning is performed within the pre-built knowledge graph using real-time and historical operational data to calculate one or more candidate fault causes. Each candidate fault cause corresponds to a component to be investigated and a corresponding inspection step. The solution is the standard operating procedure corresponding to the fault cause, used to guide users on how to repair the fault after confirmation.

[0034] Specifically, when a user describes a problem via voice or text through the interaction layer agent, the fault description information is input into the decision layer agent. The decision layer agent then queries a pre-defined knowledge graph stored in the data layer agent. This pre-defined knowledge graph stores a large amount of related knowledge in the form of triples. This related knowledge can be stored in an entity-relationship-attribute structure, as shown in the following example: (Equipment entity: Cutting machine #A001) -- [Subsidiary model] --> (Model entity: RZCUT-2516); (Model name: RZCUT-2516) -- [Includes components] --> (Component: Feed roller); (Component entity: feed roller) -- [with failure mode] --> (failure entity: wear); (Fault entity: Wear) -- [Causes symptoms] --> (Symptom entity: Material offset); (Fault entity: Wear) -- [Corresponding component] --> (Component entity: Roller); (Component entity: Roller) -- [Inspection steps] --> (Step entity: Measure roller diameter).

[0035] (Fault Entity: Wear) -- [Solution] --> (Solution Entity: SOP - Replace Feed Roller).

[0036] The fault description information is parsed using natural language processing to extract key symptom entities (e.g., material offset), which are then queried in a pre-defined knowledge graph. For example, a Cypher-like query language can be used: `MATCH (s:Symptom {name:'Material Offset'})<-[:causing symptom]-(f:Fault) RETURN f.name`. This process retrieves all candidate fault causes related to "material offset," such as "feed roller wear," "insufficient air pressure," and "uneven material tension," forming a preliminary set of candidate fault causes.

[0037] Step S104: Combining real-time operation data and historical operation data, a confidence score is given for each candidate fault cause to obtain the confidence score corresponding to each candidate fault cause.

[0038] The confidence score incorporates at least the following: the strength of the association between candidate fault causes and fault description information in the pre-defined knowledge graph; the matching degree between real-time operational data and the pre-defined fault modes corresponding to candidate fault causes; and the success rate of resolving candidate fault causes in historical maintenance cases. The generated inspection steps for the components to be investigated are displayed on the interactive interface for user reference. If multiple components are to be investigated, a confidence score can be calculated for each component; a higher confidence score indicates a higher probability that the component is the cause of the current fault. The confidence score is a comprehensive evaluation result integrating multiple pieces of evidence. The strength of the association between candidate fault causes and fault description information in the pre-defined knowledge graph reflects the static logical reasoning results based on the domain knowledge base. The pre-defined knowledge graph pre-constructs the relationships between entities such as fault phenomena, fault causes, and equipment components, and characterizes the tightness of the relationships between entities through relation weights. When the fault description information input by the user is mapped to a specific fault phenomenon entity in the knowledge graph, the path relationship and weights between that phenomenon entity and each candidate fault cause entity constitute the evidence for the first dimension. Therefore, the strength of association reflects the structured accumulation of domain expert knowledge and historical experience.

[0039] The matching degree between real-time operational data and preset fault modes corresponding to candidate fault causes reflects the degree of agreement between the current actual operating state of the equipment and typical fault characteristics. For each known fault cause, there is a corresponding preset fault mode, which describes the typical characteristics that relevant sensor data or equipment status should exhibit when the fault occurs (e.g., abnormal vibration spectrum, current waveform distortion, temperature exceeding limits, etc.). By comparing the real-time acquired equipment operational data with these preset fault modes, the degree to which the current equipment status conforms to the typical characteristics of the fault can be quantified. Therefore, the matching degree reflects dynamic evidence based on real-time monitoring data.

[0040] The success rate of resolving candidate fault causes in historical maintenance cases reflects reliability evidence based on statistical experience. The system's accumulated historical maintenance case database records various past faults and their ultimately confirmed causes. By statistically analyzing the percentage of cases where a particular candidate fault cause was confirmed and successfully resolved under similar fault phenomena, the historical reliability of that cause can be determined. Reliability evidence can reflect statistical patterns learned from past maintenance practices.

[0041] Specifically, for any candidate cause of failure, perform the following steps: Step S1041: The correlation strength between the candidate fault cause and the fault description information in the preset knowledge graph is used as the graph correlation score.

[0042] The system locates the current fault symptom entity and candidate fault cause entities in the knowledge graph maintained by the data layer intelligent agent. For example, the fault symptom (i.e., fault description information) is material offset, and the candidate fault cause is feed roller wear. The system analyzes the relationship path between the candidate fault cause and the fault description information in the preset knowledge graph and calculates the association strength between them. The association strength is a comprehensive function (e.g., a weighted sum) of the weights of all relationships on the connection path in the preset knowledge graph. The candidate fault cause and the fault description information may not be directly connected; they may involve multi-hop relationships. For example, material offset may be connected to feeding anomaly via a "caused by..." relationship, and then to feed roller wear.

[0043] The correlation strength between the calculated candidate fault causes and fault description information is the graph correlation score.

[0044] Step S1042: Obtain the preset fault mode corresponding to the candidate fault cause, calculate the matching degree between the real-time running data and the preset fault mode, and obtain the real-time matching score.

[0045] Based on the candidate fault causes, the corresponding preset fault modes are retrieved from the fault mode library. For example, the preset fault mode is a multi-dimensional feature vector, which may include the expected feature range or threshold of multiple sensors or monitoring indicators related to the fault cause.

[0046] For example, for the cause of failure, wear of the feed roller, the preset failure mode can be represented as: "Vibration characteristics": {"Frequency band": "800-1200Hz", "Threshold": 5.0mm / s}, "Current characteristics": {"Trend": "Slowly rising", "Slope threshold": 0.1A / h}, "Visual characteristics": {"Wear level": 2, "Texture similarity": 0.7}.

[0047] Extract actual measurement values ​​corresponding to each dimension of the preset fault mode from real-time operation data. For example, extract the energy value of the vibration sensor in the 800-1200Hz frequency band, the slope of the recent change in the current of the drive motor, and the wear level score and texture similarity obtained by analyzing the roller image through computer vision model.

[0048] Calculate the matching degree for each data dimension separately: For threshold-type features (e.g., vibration energy), the matching degree is calculated as follows: Matching degree = min(1.0, real-time measurement value / preset threshold). When the real-time measurement value exceeds the preset threshold, the matching degree is 1.0. For graded features (e.g., wear level), the matching degree is calculated as follows: Matching degree = 1.0 - |Preset grade - Real-time grade| / Preset grade; For similarity-based features (e.g., texture similarity), the real-time similarity value can be used directly as the matching degree.

[0049] Finally, the matching degrees of each data dimension are weighted and fused to obtain a comprehensive real-time matching score. The weights can be pre-set based on the contribution of each dimension to fault diagnosis; for example, vibration characteristics have a weight of 0.4, current characteristics have a weight of 0.3, and visual characteristics have a weight of 0.3. .

[0050] The higher the real-time matching score, the more the real-time operating data of the target device matches the preset fault mode.

[0051] Step S1043: Calculate the success rate of candidate fault causes in historical maintenance cases from historical operation data, and use it as the historical resolution score.

[0052] Historical operation data includes historical maintenance cases. The historical operation data is used to retrieve all historical maintenance cases related to similar or consistent fault phenomena that occurred during the target equipment's historical operation. The number of cases where candidate fault causes were recorded as final fault causes and repairs were marked as successful is calculated, and the proportion of such successful cases to the total number of related cases is used to obtain the historical resolution score. For example, assuming there are 100 historical maintenance cases related to material misalignment, of which 65 were ultimately confirmed as feed roller wear and successfully repaired, and 15 were for insufficient air pressure and successfully repaired, then the historical resolution score for feed roller wear is 0.65, and for insufficient air pressure it is 0.15.

[0053] Step S1044: Combine the graph association score, real-time matching score, and historical resolution score to obtain the confidence score corresponding to the candidate fault cause.

[0054] The graph association score, real-time matching score, and historical resolution score each have corresponding preset weight coefficients, namely the first weight coefficient, the second weight coefficient, and the third weight coefficient.

[0055] The confidence score is calculated using the following formula: Confidence score = w1 × GraphWeight(F) i , S) +w2×DataMatch(F i D t +w3×HistorySuccessRate(F i ), where GraphWeight(F i S) represents the candidate fault cause F i The graph association score between the fault description information S and the fault description information S, where w1 represents the first weighting coefficient, and D tRepresents real-time running data, DataMatch(F) i D t ) represents the real-time matching score, w2 represents the second weight coefficient, and HistorySuccessRate(F) represents the second weight coefficient. i ) represents the historical resolution score, w3 represents the third weight coefficient, w1+ w2+ w3=1, for example, w1=0.4, w2=0.3, w3=0.3.

[0056] Step S105: Sort the candidate fault causes according to the confidence score, and generate a maintenance plan based on the sorting results.

[0057] The repair plan includes at least one or more components to be investigated, inspection steps for those components, and a recommended inspection sequence. A multi-level association between fault causes, component information, and inspection steps is pre-built in a preset knowledge graph. Each candidate fault cause corresponds to a component to be investigated, and each component to be investigated corresponds to a specific set of inspection steps.

[0058] A repair plan is a collection of all components to be troubleshooted and their corresponding inspection steps. The pushed repair plan lists each component and its corresponding inspection steps in descending order of confidence level, allowing users to inspect the components sequentially. All candidate fault causes are sorted in descending order of confidence score (i.e., a recommended troubleshooting order), and a repair plan is generated based on this recommended order. The repair plan prioritizes the candidate fault cause with the highest confidence level, including the corresponding component and its inspection steps. The repair plan also provides a solution corresponding to the fault cause to guide the user in completing the final repair.

[0059] As an example, for the fault description information of material misalignment, candidate causes include feed roller wear, insufficient air pressure, and uneven material tension. The repair solution is as follows: For the candidate cause of failure, namely feed roller wear, the corresponding component to be investigated is the roller, the corresponding inspection steps are to measure the roller diameter, and the solution is to replace the feed roller. For insufficient air pressure as a possible cause of failure, the corresponding component to be checked is the air pressure gauge. The corresponding inspection steps include checking the air pressure gauge reading and checking for leaks in the pipeline. The solution is to replace the seals or adjust the air pressure. For the candidate cause of uneven material tension, the corresponding component to be investigated is the tension sensor. The corresponding inspection steps include rechecking the tension sensor value and calibrating the zero point of the tension sensor. The solution is to calibrate or replace the sensor.

[0060] The maintenance method for cutting equipment provided in this embodiment includes: receiving a problem request from the target equipment; obtaining operational data corresponding to the equipment identification information based on the equipment identification information; determining at least one candidate fault cause based on fault description information and a preset knowledge graph; combining real-time operational data and historical operational data to assign a confidence score to each candidate fault cause, thereby obtaining a confidence score for each candidate fault cause; ranking the candidate fault causes based on the confidence scores; and generating a maintenance plan based on the ranking results. This method achieves accurate fault diagnosis by receiving problem requests and analyzing them using a preset knowledge graph. Users can reduce their reliance on professional personnel, resolve faults themselves, improve the accuracy of fault location, and thus enhance the maintenance efficiency of the cutting equipment.

[0061] The confidence score provided in this embodiment achieves a multi-faceted comprehensive evaluation of candidate fault causes by fusing three independent evidence dimensions: graph association score, real-time matching score, and historical resolution score. These three evidence dimensions are cross-validated from the perspectives of domain knowledge, real-time equipment status, and historical statistical experience, improving the accuracy and reliability of fault diagnosis. Specifically, the real-time matching score compares each item based on the physical characteristics of preset fault modes, making the diagnostic basis clearly interpretable, allowing maintenance personnel to clearly understand the reasons for high or low confidence scores. The introduction of historical resolution scores enables the system to continuously learn from past maintenance practices, and the accuracy of identifying common faults continuously improves with the accumulation of cases. The recommended troubleshooting sequence generated based on the confidence score transforms the diagnostic results into clear action guidelines, effectively avoiding blind trial and error and significantly improving on-site maintenance efficiency.

[0062] In some optional implementations, after step S103 above, the method further includes: Step S301: Obtain feedback on the execution results of the maintenance plan.

[0063] After users troubleshoot and repair the target device according to the pushed repair plan, they can enter the execution result feedback on the corresponding feedback interface. The execution result feedback can include the repair result (including success or failure) and the corresponding actual repair operations.

[0064] Step S302: Update the preset knowledge graph based on the execution result feedback.

[0065] Specifically, step S302 includes: Step S3021: If the execution result feedback indicates successful repair, then enhance the correlation weight between the fault description information, the adopted solution, and the actual fault cause in the preset knowledge graph.

[0066] In the knowledge graph maintained by the data layer intelligent agent, the system locates the entities involved in this case: the entity of the fault symptom (e.g., material offset), the entity of the actual cause of the fault (e.g., feed roller wear), and the entity of the solution adopted (e.g., SOP - replace feed roller).

[0067] Positively enhance the weights of the relational edges connecting these entities (e.g., (symptom) -- [cause of symptom] --> (cause of failure), (cause of failure) -- [adopted solution] --> (solution)).

[0068] The weights are updated according to the following formula:

[0069] in, Set a forgetting factor (e.g., 0.95, to retain most of the historical information). The strength of evidence, f, quantifies the contribution of the repair result to the weight update. It is calculated by combining case matching degree, solution effectiveness score, and sensor data consistency. It should be noted that the update can only positively enhance the weights when the strength of evidence f is greater than the old weight.

[0070] Specifically, ,in, , , These are the weighting coefficients. The weighting coefficients can be set according to the actual application scenario, and can be equal in weight (1 / 3 each).

[0071] Case matching degree M measures the similarity between the current case and historical successful cases, reflecting the credibility of new knowledge at the experiential level. It is calculated by comparing the feature vector of the current case (e.g., material type, thickness, pattern complexity, anomaly type, etc.) with the feature vectors of cases in the historical case library.

[0072] in, Indicates similarity. This represents the feature vector of the current case. The feature vector representing historical cases, This represents the total number of dimensions of the feature vector. This represents the value of the j-th feature in the current case. This represents the value of the j-th feature in historical cases. The case matching degree M is calculated according to the following formula:

[0073] The Solution Effectiveness Score (E) is used to evaluate the actual improvement effect of the adopted compensation strategy or process parameter adjustment on the cutting quality. Key quality indicators before and after compensation are collected, such as the flatness of the cut edges, dimensional deviation rate, and material utilization rate.

[0074] If there is only one key quality indicator, calculate the improvement rate and use it as the solution effectiveness score E. For negative indicators (the smaller the value, the better the quality, such as edge smoothness and dimensional deviation rate), the improvement rate is calculated using the following formula:

[0075] in, The index value before compensation. The value is the compensated index value.

[0076] For positive indicators (higher values ​​indicate better quality, such as material utilization rate), the improvement rate is calculated using the following formula:

[0077] in, The index value before compensation. The value is the compensated index value.

[0078] If multiple indicators improve simultaneously, the weighted average of the improvement rates of the multiple indicators can be calculated as the solution effectiveness score E.

[0079] Sensor data fit S represents the degree of matching between real-time acquired sensor data and expected normal data or model prediction data, reflecting the authenticity of the physical process and the consistency with knowledge. Feature extraction can be performed on sensor data (force, vibration, acoustic emission) to obtain a feature vector x. real Extract the expected features x from the preset knowledge graph that are associated with the sensor data. exp The expected features can be predicted through the model. The sensor data fit S is calculated according to the following formula:

[0080] in, This represents the total number of dimensions of the sensor feature vector. The feature vector representing the feature of the j-th sensor. This represents the expected feature of the j-th sensor feature.

[0081] Feature extraction is performed on sensor data (force, vibration, acoustic emission) to obtain feature vector x. real The specific method used is as follows: Extracting time-domain features: Calculating the mean, variance, root mean square, peak value, and waveform factor of sensor data; Extracting frequency domain features: Performing a fast Fourier transform on the sensor data to extract the centroid frequency, mean square frequency, and energy percentage of a specific frequency band; The above time-domain and frequency-domain features are concatenated to form the original feature vector. To eliminate redundancy and dimensional differences among features, principal component analysis is used to reduce the dimensionality of the original feature vector. The top d principal components with a cumulative variance contribution rate ≥ X (e.g., X = 95%) are selected to form the final feature vector representing the current device state. ( Represents the set of real numbers. This represents a d-dimensional real vector space, which is the set of all ordered arrays of d real numbers.

[0082] Predicted feature x exp It is obtained through a pre-trained autoencoder model. An autoencoder is an unsupervised neural network whose structure includes an encoder and a decoder. The training objective is to minimize the reconstruction error between the input data and the reconstructed data.

[0083] Model input: The model consists of a sequence of sensor data within a fixed-length time window preceding the current moment. As input. Among them, " All of these are feature vectors obtained through the aforementioned feature extraction methods.

[0084] Model Training and Label Determination: Using a large amount of sensor data from historical normal cutting states (i.e., sensor data collected when the equipment was in a healthy, fault-free state and performing standard cutting processes during past operations), a set of feature vectors is first obtained according to the feature extraction method described above, forming the training set. The autoencoder model uses these feature vectors as input, and the training objective is to minimize the reconstruction error between the input and output, that is, to use the input data itself as the supervision signal (label), without any manual annotation.

[0085] Specifically, the model consists of an encoder and a decoder. The encoder compresses the input features into a low-dimensional latent representation, and the decoder then reconstructs the latent representation into an output with the same dimension as the input. By optimizing the mean squared error loss function, the model learns to capture the inherent structure and distribution patterns of normal data, thereby accurately reconstructing feature vectors under normal patterns.

[0086] After training, for the real-time sensor data, the same feature vector x is extracted to obtain the real-time feature vector. real When input into an autoencoder, the model's output is the expected feature x reconstructed based on the normal pattern. exp When the equipment is in normal condition, the reconstruction error is ||x real -x exp|| Relatively small; when anomalies exist, the reconstruction error will increase significantly because the model has not seen the anomaly pattern, thus realizing the anomaly detection and consistency measurement of the equipment status.

[0087] In step S3022, if the execution result feedback indicates that the repair has failed, the actual cause of the fault and the actual solution will be integrated into the preset knowledge graph.

[0088] If the repair fails following the repair plan, the system enters a learning mode, guiding the user to record detailed information such as the actual repair plan, the actual cause of the failure, the replaced parts, and the adjusted parameters. The user can also upload photos or videos of the process. For new causes of failure, natural language processing technology is used to extract new entities from the user's description (e.g., "The polyurethane layer of the new model feed roller has an abnormal micron-level clearance with the drive shaft"). The relationship between the entity causing the failure, the cause, and the actual symptom entity (symptom: material misalignment) -- [leading to the symptom] --> (new cause of failure) is created as a new triple and added to the preset knowledge graph.

[0089] For the new solution, the actual operation steps recorded by the user are structured into a new solution entity and associated with the corresponding fault cause entity.

[0090] When correcting existing knowledge, if an existing knowledge entry (e.g., the association between a cause of a fault and its symptoms) is found to be incorrect, the system may reduce its weight or add a negative relationship to mark it, and then clean it up or overwrite it after more cases are verified.

[0091] For new knowledge with low to medium confidence or that is critical, it can be pushed to an expert review platform. After review and confirmation by the manufacturer's technicians, it can be formally merged into the main knowledge graph to ensure the accuracy of the knowledge.

[0092] In some alternative implementations, the method further includes: Step S401: In response to the remote assistance request, establish a connection with the remote expert terminal.

[0093] When users (on-site maintenance personnel) encounter difficulties in implementing maintenance plans, or believe that the fault is beyond their ability to handle, they can proactively request remote assistance through the intelligent after-sales system.

[0094] Specifically, users can initiate a remote assistance request by triggering the remote expert assistance function through the interface of the interactive layer intelligent agent on their user terminals (e.g., smart tablets, AR glasses). The cloud responds to the remote assistance request by matching and calling a suitable remote expert terminal based on the type of problem carried in the request, as well as information such as the available expert skills and online status. The remote expert terminal can be the expert's smart mobile terminal or VR device, etc. A connection is established with the remote expert terminal, which can be based on protocols such as WebRTC or WebSocket, negotiating a point-to-point communication link through a signaling server or via a cloud relay.

[0095] Step S402: Push the real-time operating data and on-site video of the target device to the remote expert terminal.

[0096] The camera on the terminal used by the on-site user is turned on to capture video of the target device. The real-time operating data of the target device and the on-site video are then pushed to the remote expert terminal. The on-site video and the real-time operating data of the target device are displayed on the interactive interface of the remote expert terminal.

[0097] Step S403: Receive guidance information from the remote expert terminal and display the guidance information.

[0098] The guidance information includes at least one of the following: virtual markers on the live video, voice guidance instructions, and guidance animations.

[0099] Optionally, remote experts can directly annotate the on-site video feed from their terminals using a mouse or stylus, such as drawing circles, arrows, or text annotations (e.g., "Inspect this bolt," "Wear point here"), creating virtual markers. The annotation information is then overlaid in real-time on the same video feed from the on-site personnel's terminals using virtual annotation synchronization technology (e.g., WebSocket-based coordinate and graphical command synchronization).

[0100] Optionally, experts can conduct real-time voice calls directly through the established audio channels to provide explanations, step-by-step instructions, or answer questions instantly, with key instructions being converted into text prompts in real time.

[0101] Optionally, for complex disassembly or calibration procedures, experts can retrieve pre-stored 3D disassembly or synchronous teaching animations from a cloud-based knowledge base and send them to on-site personnel's terminals for playback. On-site personnel can follow the animation step by step. Animation resources can be accessed based on equipment model and components.

[0102] On-site personnel can operate based on virtual markers, voice instructions, or animated guidance, and provide feedback to experts on the results at any time via video and voice.

[0103] In some alternative implementations, the method further includes: Step S501: Based on the operating data of the target equipment and the preset life prediction model, calculate the remaining service life of at least one component.

[0104] The system continuously monitors key components (such as feed rollers, spindle bearings, and cutting blades) through intelligent agents (various sensors) deployed on-site at the cutting equipment. The collected data is uploaded to a time-series database in the cloud via an IoT gateway. Based on the collected data, a preset lifespan prediction model is trained. The model training process utilizes historical data, taking the complete operation process of the component as input features and the time of its final failure as a label (i.e., the moment when the remaining lifespan = 0). Machine learning algorithms (such as gradient descent to optimize model parameters) are used to learn the complex mapping relationship between operating features and remaining lifespan until the model's prediction accuracy for known data reaches a preset standard.

[0105] During the operation of the target equipment, the lifespan of each component can be predicted periodically (e.g., daily). The relevant operating data of any component can be used as input data for a preset lifespan prediction model to calculate the remaining service life of the component.

[0106] Step S502: If the remaining service life is lower than a preset threshold, a maintenance reminder is generated.

[0107] The maintenance reminder should include at least the information of the component to be maintained and the suggested maintenance time. Preset thresholds can be set for each component, or a uniform preset threshold can be set. If the remaining service life of a component is less than or equal to the preset threshold, a maintenance reminder will be generated for user reference. The preset threshold can be set by considering various factors such as spare parts procurement cycle, production planning, and maintenance operations.

[0108] Maintenance reminders can be pushed to user terminals in the following formats: Warning ID: ALERT_20240520001; Component: Feed roller (Serial number: SR-2023-0456); Remaining service life: 6.2 days; Recommended action: Please arrange for a replacement within 7 days; Recommended spare part: SP-1107; Inventory check: In stock (Current inventory: 3 items); The information on components to be maintained includes the warning ID and the component itself, while the recommended maintenance time includes the remaining service life and suggested measures.

[0109] This embodiment also provides a maintenance device for a cutting device, which is used to implement the above embodiments and preferred embodiments; details already described will not be repeated. As used below, the term "module" can be a combination of software and / or hardware that implements a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0110] This embodiment provides a maintenance device for a cutting device, such as... Figure 2 As shown, it includes: A request receiving module is used to receive a problem request from a target device. The problem request includes at least fault description information and device identification information, and the device identification information corresponds to the target device. The operation data acquisition module is used to acquire operation data corresponding to the device identification information based on the device identification information. The operation data includes real-time operation data and historical operation data. Fault analysis module, used for Based on the fault description information and the preset knowledge graph, at least one candidate fault cause is determined; wherein, the preset knowledge graph at least associates and stores fault phenomena, fault causes, component information, component inspection steps, and solutions corresponding to fault causes. The scoring calculation module is used to combine the real-time operation data and historical operation data to score the confidence of each candidate fault cause, thereby obtaining a confidence score corresponding to each candidate fault cause; wherein, the confidence score incorporates at least: the correlation strength between the candidate fault cause and the fault description information in the preset knowledge graph, the matching degree between the real-time operation data and the preset fault mode corresponding to the candidate fault cause, and the success rate of the candidate fault cause in historical maintenance cases; The solution determination module is used to rank the candidate fault causes based on the confidence score, and generate a repair solution based on the ranking result. The repair solution includes at least one or more components to be investigated, inspection steps for the components to be investigated, and a recommended inspection order.

[0111] In some alternative implementations, In some optional implementations, the fault analysis module is used to perform the following steps for any candidate fault cause: The correlation strength between the candidate fault cause and the fault description information in the preset knowledge graph is used as the graph correlation score. Obtain a preset fault mode corresponding to the candidate fault cause, calculate the matching degree between the real-time running data and the preset fault mode, and obtain a real-time matching score; The success rate of resolving the candidate fault causes in historical maintenance cases is statistically analyzed from the historical operation data and used as the historical resolution score. The historical operation data includes historical maintenance cases. By combining the graph association score, real-time matching score, and historical resolution score, a confidence score is obtained for the candidate fault cause.

[0112] In some alternative embodiments, the apparatus further includes: The feedback acquisition module is used to acquire the execution result feedback for the maintenance plan; The graph update module is used to update the preset knowledge graph based on the execution result feedback.

[0113] In some optional implementations, the graph update module is used to enhance the association weights between the fault description information, the adopted solution, and the actual fault cause in the preset knowledge graph if the execution result feedback indicates that the repair was successful. If the execution result feedback indicates that the repair has failed, the actual cause of the failure and the actual solution will be integrated into the preset knowledge graph.

[0114] In some alternative embodiments, the apparatus further includes: The connection establishment module is used to establish a connection with a remote expert terminal in response to a remote assistance request; The data push module is used to push the real-time operating data and on-site video of the target device to the remote expert terminal; The guidance receiving module is used to receive guidance information from the remote expert terminal and display the guidance information, which includes at least one of the following: virtual markers on the live video, voice guidance instructions, and guidance animations.

[0115] In some alternative embodiments, the apparatus further includes: The life prediction module is used to calculate the remaining life of at least one component based on the operating data of the target device and a preset life prediction model. The lifespan reminder module is used to generate a maintenance reminder if the remaining lifespan is lower than a preset threshold; wherein the maintenance reminder includes at least information about the component to be maintained and a suggested maintenance time.

[0116] Further functional descriptions of the above modules and units are the same as those in the corresponding embodiments described above, and will not be repeated here.

[0117] In this embodiment, the maintenance device for the cutting equipment is presented in the form of a functional unit. Here, a unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that execute one or more software or fixed programs, and / or other devices that can provide the above functions.

[0118] This invention also provides a computer device having a maintenance device for the aforementioned cutting equipment.

[0119] Please see Figure 3 , Figure 3 This is a schematic diagram of the structure of a computer device provided in an optional embodiment of the present invention, such as... Figure 3 As shown, the computer device includes one or more processors 10, memory 20, and interfaces for connecting the components, including high-speed interfaces and low-speed interfaces. The components communicate with each other via different buses and can be mounted on a common motherboard or otherwise installed as needed. The processors can process instructions executed within the computer device, including instructions stored in or on memory to display graphical information of a GUI on external input / output devices (such as display devices coupled to the interfaces). In some alternative implementations, multiple processors and / or multiple buses can be used with multiple memories and multiple memory modules, if desired. Similarly, multiple computer devices can be connected, each providing some of the necessary operations (e.g., as a server array, a group of blade servers, or a multiprocessor system). Figure 3 Take a processor 10 as an example.

[0120] Processor 10 may be a central processing unit, a network processor, or a combination thereof. Processor 10 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The programmable logic device may be a complex programmable logic device (CAMP), a field-programmable gate array (FPGA), a general-purpose array logic (GDA), or any combination thereof.

[0121] The memory 20 stores instructions executable by at least one processor 10 to cause the at least one processor 10 to perform the method shown in the above embodiments.

[0122] The memory 20 may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created based on the use of the computer device. Furthermore, the memory 20 may include high-speed random access memory and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some alternative embodiments, the memory 20 may optionally include memory remotely located relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0123] The memory 20 may include volatile memory, such as random access memory; the memory may also include non-volatile memory, such as flash memory, hard disk or solid-state drive; the memory 20 may also include a combination of the above types of memory.

[0124] The computer device also includes a communication interface 30 for communicating with other devices or communication networks.

[0125] This invention also provides a computer-readable storage medium. The methods described above according to embodiments of the invention can be implemented in hardware or firmware, or implemented as computer code that can be recorded on a storage medium, or implemented as computer code downloaded via a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and then stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium can also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code, which, when accessed and executed by the computer, processor, or hardware, implements the methods shown in the above embodiments.

[0126] A portion of this invention can be applied as a computer program product, such as computer program instructions, which, when executed by a computer, can invoke or provide the methods and / or technical solutions according to the invention through the operation of the computer. Those skilled in the art will understand that the forms in which computer program instructions exist in a computer-readable medium include, but are not limited to, source files, executable files, installation package files, etc. Correspondingly, the ways in which computer program instructions are executed by a computer include, but are not limited to: the computer directly executing the instructions, or the computer compiling the instructions and then executing the corresponding compiled program, or the computer reading and executing the instructions, or the computer reading and installing the instructions and then executing the corresponding installed program. Here, the computer-readable medium can be any available computer-readable storage medium or communication medium accessible to a computer.

[0127] Although embodiments of the invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the invention, and all such modifications and variations fall within the scope defined by the invention.

Claims

1. A method for maintaining a cutting device, characterized in that, The method includes: Receive a problem request from the target device, the problem request including at least fault description information and device identification information, the device identification information corresponding to the target device; Based on the device identification information, obtain the operating data corresponding to the device identification information, the operating data including real-time operating data and historical operating data; Based on the fault description information and the preset knowledge graph, at least one candidate fault cause is determined; wherein, the preset knowledge graph at least associates and stores fault phenomena, fault causes, component information, component inspection steps, and solutions corresponding to fault causes. By combining the real-time operation data and historical operation data, a confidence score is assigned to each candidate fault cause to obtain a confidence score corresponding to each candidate fault cause; wherein, the confidence score incorporates at least: the correlation strength between the candidate fault cause and the fault description information in the preset knowledge graph, the matching degree between the real-time operation data and the preset fault mode corresponding to the candidate fault cause, and the success rate of the candidate fault cause in historical maintenance cases; The candidate fault causes are ranked based on the confidence score, and a repair plan is generated based on the ranking result. The repair plan includes at least one or more components to be investigated, inspection steps for the components to be investigated, and a recommended inspection order.

2. The maintenance method for the cutting equipment according to claim 1, characterized in that, The process involves combining real-time and historical operational data to assign a confidence score to each candidate fault cause, resulting in a confidence score for each candidate fault cause, including: For any candidate cause of failure, perform the following steps: The correlation strength between the candidate fault cause and the fault description information in the preset knowledge graph is used as the graph correlation score. Obtain a preset fault mode corresponding to the candidate fault cause, calculate the matching degree between the real-time running data and the preset fault mode, and obtain a real-time matching score; The success rate of resolving the candidate fault causes in historical maintenance cases is statistically analyzed from the historical operation data and used as the historical resolution score. The historical operation data includes historical maintenance cases. By combining the graph association score, real-time matching score, and historical resolution score, a confidence score is obtained for the candidate fault cause.

3. The maintenance method for the cutting equipment according to claim 2, characterized in that, The confidence score is calculated according to the following formula: Confidence score = w1 × GraphWeight(F) i , S) +w2×DataMatch(F i D t +w3×HistorySuccessRate(F i ) Among them, GraphWeight(F i S) represents the candidate fault cause F i The graph association score between the fault description information S and the fault description information S, where w1 represents the first weighting coefficient, and D t Represents real-time running data, DataMatch(F) i D t ) represents the real-time matching score, w2 represents the second weight coefficient, and HistorySuccessRate(F) represents the second weight coefficient. i ) represents the historical solution score, w3 represents the third weight coefficient, and w1 + w2 + w3 = 1.

4. The maintenance method for the cutting equipment according to claim 1, characterized in that, After performing fault analysis based on the fault description information, operational data, and a preset knowledge graph to generate a maintenance plan, the method further includes: Obtain feedback on the execution results of the aforementioned repair plan; The preset knowledge graph is updated based on the execution results.

5. The maintenance method for the cutting equipment according to claim 4, characterized in that, The step of updating the preset knowledge graph based on the execution result feedback includes: If the execution result feedback indicates that the repair was successful, the association weight between the fault description information, the adopted solution, and the actual fault cause in the preset knowledge graph is enhanced. If the execution result feedback indicates that the repair has failed, the actual cause of the failure and the actual solution will be integrated into the preset knowledge graph.

6. The maintenance method for the cutting equipment according to claim 1, characterized in that, The method further includes: In response to a remote assistance request, establish a connection with the remote expert's terminal; The real-time operating data and on-site video of the target device are pushed to the remote expert terminal; The system receives and displays guidance information from the remote expert terminal. The guidance information includes at least one of the following: virtual markers on the live video, voice guidance instructions, and guidance animations.

7. The maintenance method for the cutting equipment according to claim 1, characterized in that, The method further includes: Based on the operating data of the target device and the preset life prediction model, calculate the remaining service life of at least one component; If the remaining service life is lower than a preset threshold, a maintenance reminder is generated; wherein the maintenance reminder includes at least information about the component to be maintained and a suggested maintenance time.

8. A maintenance device for a cutting machine, characterized in that, The device includes: A request receiving module is used to receive a problem request from a target device. The problem request includes at least fault description information and device identification information, and the device identification information corresponds to the target device. The operation data acquisition module is used to acquire operation data corresponding to the device identification information based on the device identification information. The operation data includes real-time operation data and historical operation data. The fault analysis module is used to determine at least one candidate fault cause based on the fault description information and a preset knowledge graph; wherein the preset knowledge graph at least associates and stores fault phenomena, fault causes, component information, component inspection steps, and solutions corresponding to the fault causes. The scoring calculation module is used to combine the real-time operation data and historical operation data to calculate the confidence score for each candidate fault cause, thereby obtaining the confidence score corresponding to each candidate fault cause; wherein, the confidence score incorporates at least: the correlation strength between the candidate fault cause and the fault description information in the preset knowledge graph, the matching degree between the real-time operation data and the preset fault mode corresponding to the candidate fault cause, and the success rate of the candidate fault cause in historical maintenance cases; The solution determination module is used to rank the candidate fault causes based on the confidence score, and generate a repair solution based on the ranking result. The repair solution includes at least one or more components to be investigated, inspection steps for the components to be investigated, and a recommended inspection order.

9. A computer device, characterized in that, include: A memory and a processor are communicatively connected, the memory storing computer instructions, and the processor executing the computer instructions to perform the maintenance method of the cutting device according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing a computer to perform the maintenance method of the cutting device according to any one of claims 1 to 7.

Images