A method and system for predicting equipment health status based on large model agent capabilities

Abstract

The present invention discloses a method and system for predicting the health status of equipment based on the ability of a large model AGENT. The method collects equipment vibration, temperature, pressure, current and other data in real time through multimodal sensors, and combines the virtual data generated by digital twin simulation to achieve data fusion and self-supervised learning, thereby extracting equipment operation characteristics and calculating the equipment health index. The present invention comprehensively considers the health index and its rate of change to achieve a dynamic assessment of equipment failure risks. At the same time, it introduces a dynamic weight adaptive update mechanism to optimize the model parameters online to ensure that the model maintains high-precision predictions during long-term operation. The system can be widely applied to various types of heterogeneous equipment, realizing closed-loop management of the entire process from data collection, intelligent analysis to risk warning, and providing a scientific basis for equipment maintenance and preventive maintenance, thereby reducing equipment failure rates and extending equipment service life.

Description

Technical Field

[0001] The present invention relates to the technical field of industrial equipment health monitoring and predictive maintenance, and in particular to a method and system for predicting equipment health status based on large model agent capabilities. Background Art

[0002] With the increasing intelligence of industrial equipment, monitoring the operating status of key equipment and predicting faults have become essential for ensuring production safety and efficiency. Traditional methods typically rely on expert experience and fixed algorithms to analyze equipment data. These methods struggle to process large amounts of multi-source data from diverse equipment types, and the models lack adaptability.

[0003] For example, Chinese patent CN119668195A discloses a multi-area monitoring system and method based on an AI visual model, which performs multi-area status monitoring of equipment by collecting images through cameras. However, this existing technology mainly relies on visual data, and is limited in obtaining internal performance parameters of the equipment and in-depth health status assessment. In addition, due to the use of specific AI models, it is difficult to adapt to heterogeneous equipment, does not have the ability to self-evolve and learn, and cannot dynamically adjust model parameters according to equipment operation data. Therefore, in actual applications, there are still problems such as insufficient prediction accuracy, poor generalization ability, and inability to quantify risks in advance, making it difficult to detect potential equipment failures in a timely manner and take preventive measures.

[0004] In summary, existing technologies have the defects of insufficient adaptability and accuracy in predicting equipment health status. It is necessary to propose a new technical solution to integrate multi-source data and utilize more advanced large-model agent capabilities to achieve accurate prediction and risk quantification of the health status of different types of equipment. Summary of the Invention

[0005] Technical purpose: In response to the problems in the existing technology of equipment health status prediction methods, such as poor generalization ability, difficulty in adapting to heterogeneous equipment, insufficient self-learning ability, and lack of quantitative assessment of health degradation risks, the present invention discloses a method and system for predicting equipment health status based on large model agent capabilities. It uses pre-trained large model intelligent agents to effectively integrate multi-source data of equipment and digital twin simulation to achieve accurate assessment of the operating health status of different types of equipment, and provides quantitative indicators of the risks brought about by dynamic changes in health status, thereby providing early warning of potential failures and optimizing maintenance decisions.

[0006] Technical solution: To achieve the above technical objectives, the present invention adopts the following technical solution:

[0007] A method for predicting the health status of equipment based on the large model agent capability includes the following steps:

[0008] Obtaining the operating data of the equipment to be monitored through a data acquisition device;

[0009] Preprocess the running data to generate standard time series data vectors;

[0010] Use digital twin simulation models to generate virtual operating data to expand abnormal operating condition samples, and fuse virtual data with actual data to form comprehensive input data;

[0011] The comprehensive input data is fed into the pre-trained large model AGENT for analysis to obtain the health index H(t) representing the health status of the equipment;

[0012] Calculate the risk index R(t) of the equipment based on the health index H(t) and its rate of change over time;

[0013] The prediction results of the equipment health status are output based on the health index H(t) and risk index R(t).

[0014] Preferably, the risk indicator R(t) is calculated using the following formula:

[0015] ,

[0016] Among them, H(t) is the health index of the device at time t, is the rate of change of health index over time, and is the weight coefficient, which adjusts the ratio of the impact of health index and change rate on risk, satisfying , is the sensitivity parameter of the health index influencing factor, It is the sensitivity parameter of the health index to the time-varying rate factor.

[0017] Preferably, after outputting the prediction results, the internal parameter weights of the large model AGENT are adaptively updated to improve the accuracy of subsequent predictions. The weight update is based on the self-supervised learning rule. When the prediction error of multiple features is detected, When deviating from the average level, adjust the corresponding weight according to the following formula :

[0018] ,

[0019] in, is the average value of all feature errors, is the standard deviation of the characteristic error, is the learning rate, is the preset minimum constant.

[0020] Preferably, the large model AGENT is composed of a deep neural network, which obtains general feature extraction capabilities through pre-training on massive multi-source equipment data, and uses the equipment's own operating data for continuous fine-tuning training through adaptive update steps, so as to adapt to the health status prediction of different equipment and working conditions.

[0021] Preferably, it also includes using a digital twin simulation model to generate virtual operating data of the equipment, and inputting the virtual data together with the actually collected data into the large model AGENT for analysis and training, so as to make up for the lack of real data under abnormal working conditions and enhance the model's ability to identify potential failure modes of the equipment.

[0022] Preferably, the equipment operation data includes one or more sensor data of vibration, temperature, pressure, current, and voltage, and the health index H(t) is a dimensionless value from 0 to 1, where 1 indicates that the equipment is completely healthy and 0 indicates that the equipment is in a failed state.

[0023] A system for predicting equipment health status based on large-scale model agent capabilities, including:

[0024] Data acquisition module, used to obtain equipment operation data;

[0025] Data processing module, used for synchronizing and preprocessing the collected operation data;

[0026] A large model AGENT analysis module, connected to the data acquisition module, is used to receive the operating data and calculate the equipment health index H(t);

[0027] A risk assessment module, connected to the AGENT analysis module, for calculating the risk index R(t) based on the health index H(t) and its change rate;

[0028] An output module, used to output the prediction results of the health index and risk index;

[0029] The large model AGENT analysis module includes a pre-trained deep learning model and an adaptive update unit, and is configured to execute the method described above for predicting the health status of equipment based on the large model AGENT capability.

[0030] Preferably, it also includes a digital twin simulation module, which interacts with the large model AGENT analysis module to generate simulation data of the equipment under different working conditions and provide it to the large model AGENT analysis module to enrich the model training data source and improve the prediction ability of abnormal conditions.

[0031] Preferably, the large model AGENT analysis module includes:

[0032] A feature extraction unit, configured to extract and fuse multimodal features of the operation data;

[0033] a health assessment unit connected to the feature extraction unit, for calculating the device health index H(t) based on the extracted features;

[0034] The model training unit is used to update the model parameters of the agent online using a self-supervised learning algorithm.

[0035] Beneficial effects: The method and system for predicting equipment health status based on large-model agent capabilities provided by the present invention have the following beneficial effects:

[0036] 1. The large-scale AGENT model of the present invention is trained on massive multi-domain data and possesses extensive knowledge representation capabilities. It is continuously fine-tuned through an adaptive weight update mechanism and can be applied to equipment of different types and working conditions, thus overcoming the shortcomings of traditional models, which are highly targeted but lack versatility. Whether it is a CNC machine tool, an engine, or power equipment, the present invention can effectively predict its health status. At the same time, a self-supervised training strategy is introduced to automatically adjust model parameters, enabling the AGENT to learn from unlabeled data. The model continues to evolve in actual operation, becoming "smarter" with use, effectively responding to slow drifts and sudden changes in equipment status, and achieving long-term accurate predictions.

[0037] 2. This invention uses digital twin technology to generate virtual fault and operating condition data, enriching the training sample space. Agent can learn device behavior under extreme conditions in a safe and controllable virtual environment, improving the ability to predict rare faults. This simulation fusion approach overcomes the limitation of insufficient actual fault data and enhances the robustness of the model.

[0038] 3. This invention simultaneously outputs an easy-to-understand health index H(t) and a quantitative risk index R(t). The health index intuitively reflects the current state of the equipment, while the risk index quantifies the likelihood and urgency of failures in the future. By integrating status and trends through risk indicators, it provides richer information than a single health score, assisting operations and maintenance personnel in making more informed decisions about maintenance timing and avoiding premature or delayed maintenance.

[0039] By comprehensively applying the above technical solutions, the present invention can maximize the normal operation time of equipment while ensuring equipment safety. Compared with traditional experience judgment or periodic maintenance strategies, the prediction results of the present invention can transform maintenance from passive to active, and intervene before a failure occurs. This will reduce the number of unexpected shutdowns, reduce maintenance costs, and improve the continuity and reliability of industrial production. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for describing the embodiments or the prior art.

[0041] Figure 1 This is a flow chart of the method for predicting the health status of equipment based on the large model AGENT capability of the present invention;

[0042] Figure 2 This is the overall architecture diagram of the system for predicting the health status of equipment based on the large model AGENT capability of the present invention;

[0043] Figure 3 This is a relationship diagram between the health index and risk indicators of the present invention. DETAILED DESCRIPTION

[0044] The present invention will be described more clearly and completely below by way of a preferred embodiment in conjunction with the accompanying drawings, but the present invention is not limited to the scope of the embodiment.

[0045] like Figure 1 As shown, a method for predicting the health status of equipment based on the large model agent capability includes the following steps:

[0046] S1. Obtain the operating data of the equipment to be monitored through a data acquisition device;

[0047] Collect multimodal data of the target equipment during operation, including parameters such as vibration, temperature, pressure, current measured by sensors, as well as optional visual images, control logs, etc.

[0048] S2. Preprocess the running data to generate a standard time series data vector;

[0049] The collected data is preprocessed by cleaning, filtering, and normalization to form a unified time series data input format.

[0050] S3. Generate virtual operating data using the digital twin simulation model to expand the abnormal operating condition samples, and fuse the virtual data with the actual data to form comprehensive input data;

[0051] A digital twin model is constructed for the target device to simulate its behavior under different operating conditions and failure modes. The simulation data generated by the digital twin is used to enhance the training set of the large agent model to compensate for the lack of actual failure data, thereby improving the model's ability to identify abnormal operating conditions. Simulation data can be periodically fed into the large agent model for self-supervised training, continuously updating the agent parameters to closely match the device's actual state.

[0052] S4. Send the comprehensive input data to the pre-trained large model AGENT for analysis to obtain the health index H(t) representing the health status of the equipment;

[0053] The preprocessed device data is fed into a pre-trained large-scale agent model. This agent, comprised of a large-scale deep neural network, possesses general feature extraction and inference capabilities acquired through massive data training, and adapts to specific devices through self-supervised learning. The agent performs multi-level feature analysis on the input data, extracting key health indicators of the device's operating status and comprehensively outputting the device's current health index, H(t). The health index, H(t), is a dimensionless value ranging from 0 to 1, where 1 indicates a completely healthy device and 0 indicates a failed or worst-case health state.

[0054] S5. Calculate the risk index R(t) of the equipment based on the health index H(t) and its rate of change over time;

[0055] Based on the health index H(t) and its rate of change, the risk index R(t) of the equipment health status is calculated. The present invention proposes the following dynamic risk quantification formula:

[0056] ,

[0057] In the formula, and is the weight coefficient, which adjusts the contribution weight of the current health level and health change rate of the equipment to the risk respectively; and is a positive sensitivity coefficient that controls the rate at which risk increases with declining health and increasing health changes; H(t) is the health index at time t; is the rate of change of the health index over time. This risk indicator, R(t), increases with the degree of equipment health deterioration (1-H(t)) and the rate of health decline, and is used to quantify the relative risk of future failure. Through the above formula, the present invention can incorporate both the current state of the equipment and its dynamic degradation trend into the risk value, enabling a quantitative assessment of potential failure risk.

[0058] S6. Output the prediction result of the equipment health status based on the health index H(t) and the risk index R(t).

[0059] The health index H(t) and risk index R(t) are output as the equipment health status prediction results. If the risk index R(t) exceeds the preset threshold, an alarm or maintenance plan recommendation can be triggered in advance, and the results can be provided to operation and maintenance personnel for decision support.

[0060] During the continuous operation of the equipment, the present invention uses a self-supervised learning mechanism to dynamically update the parameter weights of the large model AGENT. The following weight adaptive update formula is proposed:

[0061] ,

[0062] In the formula, Represents the weight value of the i-th parameter in the Agent at time t; represents the prediction error or evaluation index calculated based on the i-th data feature of the device at time t; is all characteristic errors at that moment The average value of is the standard deviation of the error at that moment; is the learning rate; is a preset minimum constant to avoid division by zero. According to the above update rules, when the error of a feature is higher than the average level, its corresponding weight This will result in positive adjustments (increases) and negative adjustments (decreases), allowing the large agent model to pay more attention to abnormal characteristic signals and correct its own model. This adaptive adjustment mechanism, which requires no manual labeling and falls under the category of self-supervised training, enables the model to continuously optimize as data accumulates, enhancing its adaptability to device degradation patterns and achieving adaptive generalization across heterogeneous devices.

[0063] In a specific embodiment, the method of the present invention is specifically operated according to the following process:

[0064] Initialization: After the system is powered on or started, the pre-trained model parameters of the large model AGENT are loaded, basic device information and historical data records are read, risk thresholds and other parameters are set, and the monitoring cycle is entered after confirming that each module is operating normally.

[0065] Data acquisition and preprocessing: The data acquisition module collects data from various sensors of the device at fixed time intervals (such as every second or every minute). The data processing module synchronizes and filters the acquired multi-source data and packages the results into a feature vector at the current time t , which contains normalized values such as temperature T(t), vibration a(t), and pressure P(t).

[0066] AGNET health assessment: The large model AGENT analysis module receives the current feature vector , using its internal deep model to calculate and output the current health index H(t). For example, AGNET may calculate H(t)=0.85, indicating that the device is at an 85% health level.

[0067] Risk assessment: The risk assessment module reads the current H(t) and the previous moment H(t-△t) and calculates the health change rate Substituting this into the risk indicator calculation formula yields the R(t) value. For example, if H(t) = 0.85 and has been slowly decreasing over time, R(t) is likely around 0.2, indicating a low risk level. Based on this, the assessment module concludes the risk level is "low," and no alarm is triggered, merely recording the status.

[0068] Result Output: The current H(t) and R(t), along with the risk level, are displayed via the output module. If R(t) exceeds the threshold—for example, if R(t) = 0.78, which exceeds the warning threshold of 0.7—the output module issues a "medium risk" alarm and recommends scheduling maintenance. This allows maintenance personnel to plan maintenance in advance and avoid failures.

[0069] Model update: The model update module summarizes the prediction error of the current cycle. For unsupervised cases, the AGNET model can be used to reconstruct the error or prediction residual of the collected data as Assuming that some sensor data patterns change in this period, resulting in an increase in the prediction residual, the corresponding feature Above average , according to the weight adaptive update formula, the weights of these features will be increased Conversely, for features with errors below the average, their weights are appropriately reduced. The updated model parameters are used for predictions in the next cycle. Through continuous iteration, the AGENT model is gradually optimized for the specific device.

[0070] Loop execution: The system returns to the data collection step and enters the next monitoring cycle, repeating this cycle to achieve continuous monitoring and prediction during equipment operation. The process ends when the equipment stops operating or monitoring is manually stopped.

[0071] Formula parameter selection and adjustment: In specific implementation, the parameters of the risk indicator calculation formula and weight adaptive update formula can be set and adjusted according to the equipment type and historical data. For example, for key equipment with serious consequences of failure, a larger To increase the sensitivity to rapid changes in health; for wear-related slow-changing faults, the weight can be increased Increase the contribution of reduced health to risk. and Through simulation test optimization, the response of risk R(t) to the health index and its rate of change can be made consistent with engineering experience. For example, by fitting historical failure data, it can be determined that when the health index is lower than 0.4, the risk rises sharply. Value. Learning rate Generally, a small value (such as 0.01) is taken to ensure smooth convergence of weight updates; Typically, the order of 10-6 is used to prevent division by zero. By properly configuring the above parameters, the method of the present invention can be customized for different application scenarios to achieve optimal prediction performance.

[0072] like Figure 2 As shown, a system for predicting the health status of equipment based on the large model agent capability includes:

[0073] Data acquisition module, used to obtain equipment operation data;

[0074] Data processing module, used for synchronizing and preprocessing the collected operation data;

[0075] A large model AGENT analysis module, connected to the data acquisition module, is used to receive the operating data and calculate the equipment health index H(t);

[0076] The large model AGENT analysis module includes:

[0077] A feature extraction unit, configured to extract and fuse multimodal features of the operation data;

[0078] a health assessment unit connected to the feature extraction unit, for calculating the device health index H(t) based on the extracted features;

[0079] Model training unit, used to update the model parameters of AGENT online using self-supervised learning algorithm;

[0080] A risk assessment module, connected to the AGENT analysis module, for calculating the risk index R(t) based on the health index H(t) and its change rate;

[0081] An output module, used to output the prediction results of the health index and risk index;

[0082] The large model AGENT analysis module includes a pre-trained deep learning model and an adaptive update unit, and is configured to execute the method described above for predicting the health status of equipment based on the large model AGENT capability.

[0083] The system of the present invention also includes a digital twin simulation module, which interacts with the large model AGENT analysis module to generate simulation data of the equipment under different working conditions and provide it to the large model AGENT analysis module to enrich the model training data source and improve the prediction ability of abnormal conditions.

[0084] In a specific embodiment, the system of the present invention includes the following components:

[0085] Data Acquisition Module: This module is used to acquire operational data from the monitored equipment. This data includes multiple sensor signals, such as vibration acceleration, temperature, pressure, current, voltage, and equipment operating control parameters. For scenarios requiring image monitoring, the data acquisition module also includes industrial cameras or visual sensors to capture images / video of the equipment's appearance or operating conditions. After analog-to-digital conversion and initial buffering, this raw data is sent to the data processing module.

[0086] Data processing module: Synchronizes and preprocesses collected multi-source data, including denoising and filtering, outlier removal, time alignment, and normalization, generating time series data in a unified format. This processed data serves as a feature sequence for subsequent analysis. This module ensures the quality and comparability of data from different sensors and reduces interference.

[0087] Large-model AGNET analysis module: This is the core intelligent analysis unit of the present invention, including a pre-trained large-scale deep learning model and an inference engine. Preferably, the AGNET is based on advanced architectures such as Transformer and Graph Neural Network, integrating prior knowledge of device physics and big data training experience. The feature sequence output by the data processing module is fed into the large-model AGNET as input, and after multiple layers of nonlinear transformation and feature extraction, the health index H(t) of the device is calculated. Specifically, the AGENT internally establishes a representation vector of the device health status, uses the attention mechanism to fuse the features of each sensor, and outputs a health index H(t) in the range of 0 to 1. The higher the value, the healthier the device. In order to enhance generalization, the large-model AGENT analysis module has been pre-trained using public large-scale industrial data sets and historical operating data during initial deployment, and the self-supervision mechanism of the present invention can be used to continuously update parameters on-site to adapt to the specific feature patterns of specific devices.

[0088] Digital twin simulation module: This module is used to build a digital twin model of the monitored equipment and generate simulation data. This module can be selected independently according to actual usage. The digital twin model can use multi-body dynamics simulation, finite element analysis or a simulation model trained based on historical data to realistically reproduce the behavior of the equipment under various working conditions. When the equipment operates within a safe range, the digital twin simulation module can simulate the response of the equipment under abnormal working conditions (such as overload, component wear and failure, etc.), and generate corresponding virtual sensor data and health status evolution trajectory. These simulation data are sent to the large model AGENT analysis module to expand the training set to cover extreme or rare situations, helping AGENT learn to identify and predict health changes in such situations. Through the combination of real and virtual data, the present invention significantly improves the model's predictive reliability when facing unknown situations.

[0089] Risk assessment module: used to calculate the risk index R(t) of the equipment based on the health index H(t) and its changing trend output by the large model AGNET analysis module, such as Figure 3 The figure shows the relationship between the equipment health index H(t) and the risk R(t). As shown in the risk indicator formula above, the risk assessment module first calculates the equipment health deterioration degree (1-H(t)) based on the current health index H(t), and combines the health index difference at adjacent moments to obtain the health change rate. Then, the pre-set weight coefficient is applied. 、 and sensitivity parameters 、 Substituting this into the formula, the risk index R(t) is calculated in real time. The risk index R(t) ranges from 0 to approximately 2, with larger values indicating a closer proximity to failure or a higher probability of failure. The risk assessment module generates a warning signal based on the comparison of the calculated R(t) with a threshold. For example, when R(t) exceeds a certain threshold, Rth, the equipment is deemed to be in a high-risk state and requires maintenance.

[0090] Model update module: This module implements the adaptive update mechanism of model parameters. In specific implementation, after each prediction cycle, the system will compare the health index and risk index obtained by the risk assessment module with the actual equipment status (such as whether a fault occurs in subsequent operation) and calculate the prediction error of each feature. The error can be obtained by comparing the actual health index of the equipment at a later time (if it can be obtained by manual evaluation or shutdown maintenance) with the predicted value, or by using unsupervised indicators (such as reconstruction error, residual error, etc.). The model update module averages all feature errors. and standard deviation , and then adjust the weights of the corresponding features within the large model AGENT analysis module This adjustment is reflected in the software by modifying the weight parameters of some connections in the neural network, so that when predicting the next cycle, AGNET pays more attention to or appropriately corrects features with large errors. This self-learning process requires no human intervention. When the operating environment or state statistics of the device gradually change, the model can automatically adapt to maintain stable prediction performance.

[0091] Output Module: This module outputs device health status predictions, including health index H(t), risk indicator R(t), risk level, and recommended actions. This output can be displayed in real-time on a human-machine interface dashboard as numerical values and trend charts, or sent to a host computer or cloud-based monitoring system. If a risk indicator exceeds a threshold or meets a warning condition, the alarm module triggers an audible or visual alarm or sends an alarm notification to maintenance personnel. Event logs are also recorded for future analysis. This module can also archive prediction data as needed, providing a basis for further offline analysis and model improvement.

[0092] Taking the CNC machine tool spindle system as an example, after the system of the present invention is deployed, the sensor continuously collects signals such as spindle vibration acceleration, temperature and current. The large model AGENT combines the historical data of the machine tool and the wear and failure data simulated by the digital twin to analyze and output the spindle health index H(t). In the early stage of normal operation, H(t) remains above 0.9, which is highly healthy. As the usage time accumulates, H(t) slowly decreases to about 0.6, and the vibration signal increases abnormally, resulting in The absolute value of H(t) increases. According to the calculation formula of the risk index, the risk index R(t) increases from the initial 0.1 to 0.5, approaching the warning threshold. The system alarm reminds that the spindle bearing needs to be repaired. After the maintenance personnel replaced the bearings, the vibration level dropped, H(t) returned to 0.85, R(t) dropped below 0.2, and the system lifted the alarm. This process shows that the present invention can effectively track changes in equipment health and give a quantitative risk assessment, providing a scientific basis for maintenance decisions. Compared with manual experience, the method of the present invention is more objective and accurate, and timely avoids the occurrence of major spindle failures.

[0093] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for predicting equipment health status based on large model agent capabilities, characterized in that: The specific steps include: Obtaining the operating data of the equipment to be monitored through a data acquisition device; Preprocess the running data to generate standard time series data vectors; Use digital twin simulation models to generate virtual operating data to expand abnormal operating condition samples, and fuse virtual data with actual data to form comprehensive input data; The comprehensive input data is fed into the pre-trained large model AGENT for analysis to obtain the health index H(t) representing the health status of the equipment; The risk index R(t) of the equipment is calculated based on the health index H(t) and its rate of change over time. The risk index R(t) is calculated using the following formula: , Among them, H(t) is the health index of the device at time t, is the rate of change of health index over time, and is the weight coefficient, which adjusts the ratio of the impact of health index and change rate on risk, satisfying , is the sensitivity parameter of the health index influencing factor, is the sensitivity parameter of the health index to the time-varying rate factor; The prediction results of the equipment health status are output based on the health index H(t) and risk index R(t).

2. The method for predicting equipment health status based on large model agent capability according to claim 1, characterized in that: After outputting the prediction results, the internal parameter weights of the large model AGENT are adaptively updated to improve the accuracy of subsequent predictions. The weight update is based on the self-supervised learning rule. When the prediction error of multiple features is detected, When deviating from the average level, adjust the corresponding weight according to the following formula : , in, is the average value of all feature errors, is the standard deviation of the characteristic error, is the learning rate, is the preset minimum constant.

3. The method for predicting equipment health status based on large model agent capability according to claim 1, characterized in that: The large model AGENT is composed of a deep neural network. It obtains general feature extraction capabilities through pre-training on massive multi-source equipment data, and uses the equipment's own operating data through adaptive update steps for continuous fine-tuning training, so as to adapt to the health status prediction of different equipment and working conditions.

4. The method for predicting equipment health status based on large model agent capability according to claim 1, characterized in that: It also includes using a digital twin simulation model to generate virtual operating data of the equipment, and inputting the virtual data together with the actual collected data into the large model AGENT for analysis and training to make up for the lack of real data under abnormal working conditions and improve the model's ability to identify potential failure modes of the equipment.

5. The method for predicting equipment health status based on large model agent capability according to claim 1, characterized in that: The equipment operation data includes one or more sensor data of vibration, temperature, pressure, current, and voltage, and the health index H(t) is a dimensionless value between 0 and 1, where 1 indicates that the equipment is completely healthy and 0 indicates that the equipment is in a failed state.

6. A system for predicting equipment health status based on large model agent capabilities, characterized in that: include: Data acquisition module, used to obtain equipment operation data; Data processing module, used for synchronizing and preprocessing the collected operation data; A large model AGENT analysis module, connected to the data acquisition module, is used to receive the operating data and calculate the equipment health index H(t); A risk assessment module, connected to the AGENT analysis module, for calculating the risk index R(t) based on the health index H(t) and its change rate; An output module, used to output the prediction results of the health index and risk index; The large model AGENT analysis module includes a pre-trained deep learning model and an adaptive update unit, and is configured to execute a method for predicting the health status of a device based on the large model AGENT capability according to any one of claims 1 to 5.

7. The system for predicting equipment health status based on large model agent capability according to claim 6, characterized in that: It also includes a digital twin simulation module, which interacts with the large model AGENT analysis module to generate simulation data of the equipment under different working conditions and provide it to the large model AGENT analysis device to enrich the model training data source and improve the prediction ability of abnormal conditions.

8. The system for predicting equipment health status based on large model agent capabilities according to claim 6, characterized in that: The large model AGENT analysis module includes: A feature extraction unit, configured to extract and fuse multimodal features of the operation data; a health assessment unit connected to the feature extraction unit, for calculating the device health index H(t) based on the extracted features; The model training unit is used to update the model parameters of the agent online using a self-supervised learning algorithm.

Images