A medical device life prediction and maintenance cycle planning method

Abstract

The application discloses a kind of medical equipment service life prediction and maintenance cycle planning method, comprising: collecting the equipment running data of the medical equipment to be predicted, environmental state data and historical maintenance data and carries out standardized pretreatment, constructs graph convolution-gate recurrent spatiotemporal fusion network model, realize the extraction and fusion of equipment spatiotemporal characteristics, generate the characteristic vector reflecting the comprehensive health state of equipment and calculate health state index;Adopt goodness-of-fit screening mechanism to determine life distribution model, and obtain life distribution parameter and service life prediction value based on its characteristic vector input;Combining life distribution parameter and health state index, and using analytic hierarchy process to construct maintenance decision model, dynamically calculate and output recommended maintenance cycle and maintenance level;Combining historical failure mode and component inventory state, generate customized preventive maintenance scheme and push to operation and maintenance management terminal, so that preventive maintenance scheme is more targeted, executable.

Description

Technical Field

[0001] This invention relates to the field of medical device health management technology, and in particular to a method for predicting the service life and planning the maintenance cycle of medical devices. Background Technology

[0002] As medical equipment becomes increasingly intelligent and complex, its operational stability directly impacts treatment safety and hospital operational efficiency. Accurate prediction of remaining service life and scientific maintenance cycle planning are crucial for ensuring reliable equipment operation, reducing the risk of sudden failures, and optimizing maintenance costs. Current preventative maintenance strategies for medical equipment often rely on fixed time periods or simple runtime thresholds, lacking consideration for real-time equipment health status and individual differences. This can easily lead to over-maintenance, increasing costs, or under-maintenance, causing malfunctions. Summary of the Invention

[0003] To overcome the aforementioned shortcomings of the prior art, this invention provides a method for predicting the service life of medical devices and planning maintenance cycles. It aims to address the problem that existing preventive maintenance strategies for medical devices often rely on fixed time periods or simple runtime thresholds, lacking consideration of the real-time health status of the device and individual differences, which can easily lead to over-maintenance increasing costs or under-maintenance causing malfunctions.

[0004] The technical solution adopted by this invention to solve its technical problem is: a method for predicting the service life and planning the maintenance cycle of medical equipment, comprising the following steps: S1. Collect equipment operation data, environmental status data, and historical maintenance data of the medical equipment to be predicted, and perform standardized preprocessing on the collected data; S2. Construct a graph convolution-gated cyclic spatiotemporal fusion network model. Based on the graph convolution-gated cyclic spatiotemporal fusion network model, extract and fuse spatiotemporal features of the standardized preprocessed data to generate a feature vector reflecting the comprehensive health status of the equipment. Calculate the health status index based on the feature vector. S3. Input the feature vector into the pre-trained lifetime distribution model to obtain the corresponding lifetime distribution parameters, and output the lifetime prediction value based on the lifetime distribution parameters. The lifetime distribution model is determined from multiple candidate distribution models through a goodness-of-fit screening mechanism. S4. Based on the lifespan distribution parameters and the health status index, a maintenance decision model is constructed using the analytic hierarchy process (AHP) to dynamically calculate and output recommended maintenance cycles and maintenance levels. S5. Based on the recommended maintenance cycle and maintenance level, and combined with the historical failure modes and component inventory status obtained from the analysis of historical maintenance data, generate a customized preventive maintenance plan and push it to the operation and maintenance management terminal.

[0005] As a further improvement of the present invention: in step S2, the construction and training process of the graph convolution-gated recurrent spatiotemporal fusion network model includes: A spatiotemporal fusion network model of graph convolution-gated recurrent unit is constructed, which is composed of cascaded graph convolutional layers and gated recurrent unit layers. Based on the device topology, the standardized preprocessed data is aggregated into component-level state vectors, and graph convolutional layers are used to propagate and aggregate information between components, generating subsystem-level and whole-machine-level spatial features. Spatial features are input into the gated recurrent unit layer to learn the evolution of equipment status over time and output a feature vector that reflects the overall health status of the equipment.

[0006] As a further improvement of the present invention: in step S3, the process of constructing and determining the parameters of the lifetime distribution model includes: Acquire failure duration data for multiple similar devices under accelerated stress; The cumulative failure function of the candidate distribution model with the highest goodness of fit determined by the goodness-of-fit screening mechanism. Modeling is performed, in which, η For scale parameters, β For shape parameters; Based on the cumulative failure function, the failure duration data is fitted to calculate the value under accelerated stress. η a and β a ,in, η a To accelerate the dimensional parameters under stress, β a Shape parameters under accelerated stress; Construct an accelerated degradation model and calculate the acceleration factor. A F The dimensional parameters under accelerated stress are converted to those under normal stress. The shape parameters remain unchanged. Thus, a lifetime distribution model under normal stress is obtained.

[0007] As a further improvement of the present invention: the goodness-of-fit screening mechanism includes: Calculate the goodness of fit of failure duration data under candidate distribution models such as Weibull distribution and log-normal distribution. R 2 The calculation formula is: in, n The number of failure duration data. ti For the first iEach failure duration data point Fo(ti) The cumulative failure rate for model fitting. Fn(ti) The standard cumulative failure rate, For all Fn(ti) The average value; The candidate distribution model with the highest goodness of fit was selected as the lifetime distribution model.

[0008] As a further improvement of the present invention: in step S4, the process of constructing the maintenance decision model using the analytic hierarchy process includes: Determine the set of criteria-level factors, including remaining life risk factors, health status index, and maintenance economics factors; Based on the set of criteria layer factors, a pairwise comparison judgment matrix is ​​constructed using the analytic hierarchy process (AHP), and after passing the consistency test, the weight vector of the set of criteria layer factors is calculated. The original calculated values ​​of the remaining life risk factor, health status index and maintenance economy factor are normalized and mapped to the interval [0, 1]. Equipment maintenance urgency index MI The calculation formula is: ; in, R r , H , C m These are the normalized values ​​of the remaining life risk factor, health status index, and maintenance economy factor, respectively. W R , W H and W C These are the components corresponding to the remaining life risk factor, health status index, and maintenance economy factor in the weight vector, respectively. W R , W H and W C All are weighting coefficients.

[0009] As a further improvement of the present invention: the original calculated value of the remaining life risk factor R r原始 The calculation formula is: ; in, T RUL_预测 This is a predicted service life value. T 设计寿命 This refers to the rated design life of the equipment.

[0010] As a further improvement of the present invention: the original calculated value of the current health index H 原始 The calculation formula is: ; in, T 0 represents the set device health baseline value. S k The first one is calculated directly based on the equipment operation data. k The standardized degradation value of each performance indicator W k For the first k Preset weights for each performance indicator.

[0011] As a further improvement of the present invention: In step S4, dynamically calculating and outputting the recommended maintenance cycle and maintenance level specifically includes: Establish maintenance urgency index in advance MI The first mapping relationship with the recommended maintenance cycle, and the maintenance urgency index. MI The second mapping relationship with maintenance level; Based on the calculated maintenance urgency index MI The recommended maintenance cycle is determined based on the first mapping relationship, and the maintenance level is determined based on the second mapping relationship.

[0012] As a further improvement of the present invention: in step S5, the process of generating a customized preventive maintenance plan includes: Based on the maintenance level, retrieve the basic maintenance items corresponding to that maintenance level from the preset maintenance knowledge base; Based on the historical failure modes, the basic maintenance items are supplemented and prioritized to form a maintenance item list and its priority ranking result; Based on the equipment identification, query the inventory status of the components involved in the maintenance project list, mark the warning for components whose inventory is lower than the safety threshold, generate component inventory warning information and trigger procurement suggestions; By integrating the maintenance item list, priority ranking results, and component inventory warning information, a customized preventive maintenance plan is generated.

[0013] This invention also provides a medical device lifespan prediction and maintenance cycle planning system, comprising: The multi-source data acquisition and preprocessing module is used to synchronously acquire equipment operation data, environmental status data, and historical maintenance data of the medical equipment to be predicted through a distributed data acquisition interface, and to perform standardized preprocessing on the acquired data. The cross-domain data fusion and feature extraction module is used to construct a graph convolution-gated cyclic spatiotemporal fusion network model, extract and fuse spatiotemporal features of standardized preprocessed data based on the graph convolution-gated cyclic spatiotemporal fusion network model, generate feature vectors reflecting the comprehensive health status of equipment, and calculate the health status index based on the feature vectors. The lifetime distribution prediction module is used to input the feature vector into a pre-trained lifetime distribution model to obtain the corresponding lifetime distribution parameters, and output the lifetime prediction value based on the lifetime distribution parameters. The lifetime distribution model is determined from multiple candidate distribution models through a goodness-of-fit screening mechanism. The dynamic maintenance cycle planning module is used to construct a maintenance decision model based on the life distribution parameters and the health status index using the analytic hierarchy process, and dynamically calculate and output recommended maintenance cycles and maintenance levels. The maintenance plan generation and output module is used to generate customized preventive maintenance plans based on the recommended maintenance cycle and maintenance level, combined with historical failure modes and component inventory status obtained from historical maintenance data, and push them to the operation and maintenance management terminal.

[0014] Compared with the prior art, the beneficial effects of the present invention are: This invention establishes a unified and reliable model input foundation by integrating multi-source heterogeneous data, including equipment operation data, environmental status data, and historical maintenance data of the medical equipment to be predicted, and performing standardized preprocessing. By constructing a graph convolutional-gated recurrent spatiotemporal fusion network model, it extracts and fuses the spatiotemporal features of the equipment, generating feature vectors reflecting the comprehensive health status of the equipment and calculating a health status index, thereby providing a more comprehensive and dynamic assessment of the actual operating conditions of the equipment. A goodness-of-fit screening mechanism is used to determine the lifespan distribution model, and lifespan distribution parameters and predicted lifespan values ​​are obtained based on its feature vector input, making lifespan prediction more statistically sound and reliable. By combining lifespan distribution parameters and the health status index, and using the analytic hierarchy process (AHP) to construct a maintenance decision model, it dynamically calculates and outputs recommended maintenance cycles and maintenance levels, achieving recommended maintenance cycle and maintenance level planning based on the real-time status and long-term reliability of the equipment. By combining historical failure modes and component inventory status, customized preventative maintenance plans are generated and pushed to the operation and maintenance management terminal, making preventative maintenance plans more targeted and executable, while improving the efficiency of operation and maintenance resource utilization. Attached Figure Description

[0015] Figure 1 This is a flowchart of a method for predicting the service life and maintenance cycle planning of medical equipment according to the present invention. Detailed Implementation

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the present invention will be briefly introduced below in conjunction with the accompanying drawings and descriptions of the embodiments or the prior art. Obviously, the following description of the structure of the accompanying drawings is only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. It should be noted that the description of these embodiments is for the purpose of helping to understand the present invention, but does not constitute a limitation of the present invention.

[0017] It should be understood that the terms “comprising” and “having”, and any variations thereof, are intended to cover non-exclusive inclusion, for example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or units that are not explicitly listed or that are inherent to such process, method, product, or device.

[0018] It should be understood that specific details are provided in the following description to facilitate a complete understanding of the exemplary embodiments. However, those skilled in the art will understand that the exemplary embodiments can be implemented without these specific details. For example, the system may be shown in block diagrams to avoid obscuring the example with unnecessary details. In other instances, well-known processes, structures, and techniques may be shown without unnecessary details to avoid obscuring the exemplary embodiments.

[0019] Please see Figure 1 A method for predicting the lifespan and planning the maintenance cycle of medical devices includes the following steps: S1. Multi-source data acquisition: Through a distributed data acquisition interface, the system synchronously collects equipment operation data, environmental status data, and historical maintenance data of the medical equipment to be predicted, and performs standardized preprocessing on the collected data. S2. Cross-domain data fusion and feature extraction: Construct a graph convolution-gated cyclic spatiotemporal fusion network model, extract and fuse spatiotemporal features of standardized preprocessed data based on the graph convolution-gated cyclic spatiotemporal fusion network model, generate feature vectors reflecting the comprehensive health status of equipment, and calculate the health status index based on the feature vectors; S3. Lifetime distribution prediction: The feature vector is input into the pre-trained lifetime distribution model to obtain the corresponding lifetime distribution parameters, and the lifetime prediction value is output based on the lifetime distribution parameters. The lifetime distribution model is determined from multiple candidate distribution models through a goodness-of-fit screening mechanism. S4. Dynamic maintenance cycle planning: Based on the life distribution parameters and the health status index, a maintenance decision model is constructed using the analytic hierarchy process (AHP) to dynamically calculate and output recommended maintenance cycles and maintenance levels. S5. Maintenance plan generation and output: Based on the recommended maintenance cycle and maintenance level, combined with the historical failure modes and component inventory status obtained from the analysis of historical maintenance data, a customized preventive maintenance plan is generated and pushed to the operation and maintenance management terminal.

[0020] By integrating multi-source heterogeneous data such as equipment operation data, environmental status data, and historical maintenance data of the medical equipment to be predicted, and performing standardized preprocessing, a unified and reliable model input foundation was established.

[0021] By constructing a graph convolution-gated recurrent spatiotemporal fusion network model, the spatiotemporal features of the equipment are extracted and fused to generate feature vectors reflecting the overall health status of the equipment and calculate the health status index, thereby more comprehensively and dynamically evaluating the actual working condition of the equipment.

[0022] By employing a goodness-of-fit screening mechanism to determine the lifetime distribution model, and based on its feature vector input, the lifetime distribution parameters and lifetime prediction values ​​are obtained, thus making lifetime prediction more statistically sound and reliable.

[0023] By combining lifespan distribution parameters and health status indices, and using the analytic hierarchy process (AHP) to construct a maintenance decision model, the recommended maintenance cycle and maintenance level are dynamically calculated and output, enabling the planning of recommended maintenance cycles and maintenance levels based on the real-time status and long-term reliability of the equipment. By combining historical failure modes with component inventory status, customized preventive maintenance plans are generated and pushed to the operation and maintenance management terminal, making the preventive maintenance plans more targeted and executable, while improving the efficiency of operation and maintenance resource utilization.

[0024] In some implementations, step S2, the construction and training process of the graph convolution-gated recurrent spatiotemporal fusion network model includes: A spatiotemporal fusion network model of graph convolution-gated recurrent unit is constructed, which is composed of cascaded graph convolutional layers and gated recurrent unit layers. Based on the device topology, the standardized preprocessed data is aggregated into component-level state vectors, and graph convolutional layers are used to propagate and aggregate information between components, generating subsystem-level and whole-machine-level spatial features. Spatial features are input into the gated recurrent unit layer to learn the evolution of equipment status over time and output a feature vector that reflects the overall health status of the equipment.

[0025] By aggregating standardized preprocessed data into component-level state vectors based on the device topology, and using graph convolutional layers to propagate and aggregate information between components, spatial features at the subsystem and whole-machine levels are generated. This effectively captures the complex spatial relationships and hierarchical structures within medical devices, thereby generating more comprehensive and accurate spatial features.

[0026] By inputting the spatial features extracted by the graph convolutional layer into the gated recurrent unit layer, the evolution law of equipment status in the time dimension is learned, and the deep fusion of spatial and temporal features is achieved. This generates a feature vector that can comprehensively reflect the overall health status of the equipment, providing a highly representative input for subsequent life distribution prediction and maintenance scheme generation.

[0027] In some implementations, step S3, the process of constructing and determining the parameters of the lifetime distribution model, includes: Acquire failure duration data for multiple similar devices under accelerated stress; The cumulative failure function of the candidate distribution model with the highest goodness of fit determined by the goodness-of-fit screening mechanism. Modeling is performed, in which, η For scale parameters, β For shape parameters; Based on the cumulative failure function, the failure duration data is fitted to calculate the value under accelerated stress. η a and β a ,in, η a To accelerate the dimensional parameters under stress, β a Shape parameters under accelerated stress; Construct an accelerated degradation model and calculate the acceleration factor. A F The dimensional parameters under accelerated stress are converted to those under normal stress. The shape parameters remain unchanged. Thus, a lifetime distribution model under normal stress is obtained.

[0028] By acquiring failure duration data of multiple similar devices under accelerated stress, rather than relying on long-term failure data under normal stress which is difficult to collect, the data acquisition cycle is significantly shortened.

[0029] By fitting the cumulative failure function of the candidate distribution model with the highest goodness of fit, determined through the goodness-of-fit screening mechanism, to the failure duration data under accelerated stress, the following calculations are performed: ηa and βa This enabled the preliminary and effective modeling of the statistical patterns of equipment failure.

[0030] Acceleration factor was calculated by constructing an accelerated degradation model. A FFurthermore, the scalar parameters under accelerated stress are converted to those under normal stress, while the shape parameters remain unchanged, thus obtaining a life distribution model under normal stress. This makes the final life prediction value closer to the actual operating conditions of the equipment, significantly improving the engineering practicality and reliability of the prediction results.

[0031] In some implementations, the goodness-of-fit screening mechanism includes: Calculate the goodness of fit of failure duration data under candidate distribution models such as Weibull distribution and log-normal distribution. R 2 The calculation formula is: in, n The number of failure duration data. ti For the first i Each failure duration data point Fo(ti) The cumulative failure rate for model fitting. Fn(ti) The standard cumulative failure rate, For all Fn(ti) The average value; The candidate distribution model with the highest goodness of fit was selected as the lifetime distribution model.

[0032] By setting Weibull and log-normal distributions as candidate distribution models and comparing their goodness of fit, the potential assumption bias of a single distribution model is avoided, making the model selection more comprehensive and objective.

[0033] The goodness of fit of each candidate distribution model was calculated and compared. R² This approach enables the objective selection of the lifetime distribution model that best fits the current failure duration data based on quantitative indicators. This allows the final lifetime distribution model to adaptively match the actual statistical characteristics of equipment failure data, thereby improving the accuracy and reliability of subsequent lifetime predictions.

[0034] In some implementations, step S4, the process of constructing a maintenance decision model using the analytic hierarchy process (AHP), includes: Determine the set of criteria-level factors, including remaining life risk factors, health status index, and maintenance economic factors, wherein the health status index is calculated in step S2; Based on the set of criteria layer factors, a pairwise comparison judgment matrix is ​​constructed using the analytic hierarchy process (AHP), and after passing the consistency test, the weight vector of the set of criteria layer factors is calculated. The original calculated values ​​of the remaining life risk factor, health status index and maintenance economy factor are normalized and mapped to the interval [0, 1]. Equipment maintenance urgency index MIThe calculation formula is: ; in, R r , H , C m These are the normalized values ​​of the remaining life risk factor, health status index, and maintenance economy factor, respectively. W R , W H and W C These are the components corresponding to the remaining life risk factor, health status index, and maintenance economy factor in the weight vector, respectively. W R , W H and W C All are weighting coefficients; A maintenance decision model is constructed based on the criterion-level factor set, judgment matrix, and maintenance urgency index.

[0035] By identifying remaining life risk factors, health status index, and maintenance economic factors as the criterion layer factor set, maintenance decisions can comprehensively consider key indicators from multiple dimensions such as equipment reliability, real-time health status, and economic costs, thus ensuring the comprehensiveness of the decision-making basis.

[0036] By constructing a pairwise comparison judgment matrix using the analytic hierarchy process and calculating the weight vector after a consistency test, the importance of the three criteria factors—remaining life risk factor, health status index, and maintenance economic factor—is scientifically and quantitatively assigned, reducing subjective arbitrariness and making the maintenance decision-making model more objective and reasonable.

[0037] The maintenance urgency index is calculated by normalizing the raw values ​​of the remaining life risk factor, health status index, and maintenance economic factor, and by incorporating weighting coefficients. MI This enables a clear and dynamic prioritization of equipment maintenance urgency, providing a precise and actionable basis for differentiated maintenance levels and recommended maintenance cycles.

[0038] In some implementations, the original calculated value of the remaining life risk factor R r原始 The calculation formula is: ; in, T RUL_预测 The predicted service life value output in step S3. T 设计寿命This refers to the rated design life of the equipment.

[0039] In some implementations, the original calculated value of the current health index H 原始 The calculation formula is: ; in, T 0 represents the set device health baseline value. S k The first one is calculated directly based on the equipment operation data. k The standardized degradation value of each performance indicator W k For the first k Preset weights for each performance indicator.

[0040] In some implementations, step S4, which involves dynamically calculating and outputting the recommended maintenance cycle and maintenance level, specifically includes: A first mapping relationship between the maintenance urgency index (MI) and the recommended maintenance cycle, and a second mapping relationship between the maintenance urgency index (MI) and the maintenance level are established in advance. Based on the calculated maintenance urgency index MI, the recommended maintenance cycle is determined according to the first mapping relationship, and the maintenance level is determined according to the second mapping relationship.

[0041] In some implementations, step S5, the process of generating a customized preventative maintenance plan, includes: Based on the maintenance level, retrieve the basic maintenance items corresponding to that maintenance level from the preset maintenance knowledge base; Based on the historical failure modes, the basic maintenance items are supplemented and prioritized to form a maintenance item list and its priority ranking result; Based on the equipment identification, query the inventory status of the components involved in the maintenance project list, mark the warning for components whose inventory is lower than the safety threshold, generate component inventory warning information and trigger procurement suggestions; By integrating the maintenance item list, priority ranking results, and component inventory warning information, a customized preventive maintenance plan is generated.

[0042] By retrieving corresponding basic maintenance items from a pre-set maintenance knowledge base based on maintenance level, and by supplementing and prioritizing these basic maintenance items based on historical failure modes, preventive maintenance plans can be dynamically optimized for the specific risk characteristics of equipment, achieving precise allocation and efficient utilization of maintenance resources.

[0043] By querying the inventory status of components involved in the maintenance project and marking low-inventory components with warnings, preventive maintenance plans can identify and trigger procurement needs in advance, effectively preventing maintenance plan delays caused by spare parts shortages and ensuring the feasibility of maintenance work.

[0044] By integrating the maintenance project list, priority ranking results, and component inventory warning information, customized preventive maintenance plans are generated. The output plan includes specific tasks, execution order, and resource guarantee information, realizing closed-loop management from decision-making to executable work orders, and significantly improving the timeliness and reliability of operation and maintenance response.

[0045] Implementation Case 2: A medical device lifespan prediction and maintenance cycle planning system, comprising: The multi-source data acquisition and preprocessing module is used to synchronously acquire equipment operation data, environmental status data, and historical maintenance data of the medical equipment to be predicted through a distributed data acquisition interface, and to perform standardized preprocessing on the acquired data. The cross-domain data fusion and feature extraction module is used to construct a graph convolution-gated cyclic spatiotemporal fusion network model, extract and fuse spatiotemporal features of standardized preprocessed data based on the graph convolution-gated cyclic spatiotemporal fusion network model, generate feature vectors reflecting the comprehensive health status of equipment, and calculate the health status index based on the feature vectors. The lifetime distribution prediction module is used to input the feature vector into a pre-trained lifetime distribution model to obtain the corresponding lifetime distribution parameters, and output the lifetime prediction value based on the lifetime distribution parameters. The lifetime distribution model is determined from multiple candidate distribution models through a goodness-of-fit screening mechanism. The dynamic maintenance cycle planning module is used to construct a maintenance decision model based on the life distribution parameters and the health status index using the analytic hierarchy process, and dynamically calculate and output recommended maintenance cycles and maintenance levels. The maintenance plan generation and output module is used to generate customized preventive maintenance plans based on the recommended maintenance cycle and maintenance level, combined with historical failure modes and component inventory status obtained from historical maintenance data, and push them to the operation and maintenance management terminal.

[0046] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For instance, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection between units or modules may be electrical or other forms.

[0047] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered within the scope of the claims and specification of the present invention.

Claims

1. A method for predicting the service life and planning the maintenance cycle of medical equipment, characterized in that: Includes the following steps: S1. Collect equipment operation data, environmental status data, and historical maintenance data of the medical equipment to be predicted, and perform standardized preprocessing on the collected data; S2. Construct a graph convolution-gated cyclic spatiotemporal fusion network model. Based on the graph convolution-gated cyclic spatiotemporal fusion network model, extract and fuse spatiotemporal features of the standardized preprocessed data to generate a feature vector reflecting the comprehensive health status of the equipment. Calculate the health status index based on the feature vector. S3. Input the feature vector into the pre-trained lifetime distribution model to obtain the corresponding lifetime distribution parameters, and output the lifetime prediction value based on the lifetime distribution parameters. The lifetime distribution model is determined from multiple candidate distribution models through a goodness-of-fit screening mechanism. S4. Based on the lifespan distribution parameters and the health status index, a maintenance decision model is constructed using the analytic hierarchy process (AHP) to dynamically calculate and output recommended maintenance cycles and maintenance levels. S5. Based on the recommended maintenance cycle and maintenance level, and combined with the historical failure modes and component inventory status obtained from the analysis of historical maintenance data, generate a customized preventive maintenance plan and push it to the operation and maintenance management terminal.

2. The method for predicting the service life and planning the maintenance cycle of medical equipment according to claim 1, characterized in that: In step S2, the construction and training process of the graph convolution-gated recurrent spatiotemporal fusion network model includes: A spatiotemporal fusion network model of graph convolution-gated recurrent unit is constructed, which is composed of cascaded graph convolutional layers and gated recurrent unit layers. Based on the device topology, the standardized preprocessed data is aggregated into component-level state vectors, and graph convolutional layers are used to propagate and aggregate information between components, generating subsystem-level and whole-machine-level spatial features. Spatial features are input into the gated recurrent unit layer to learn the evolution of equipment status over time and output a feature vector that reflects the overall health status of the equipment.

3. The method for predicting the service life and planning the maintenance cycle of medical equipment according to claim 1, characterized in that: In step S3, the process of constructing and determining the parameters of the lifetime distribution model includes: Acquire failure duration data for multiple similar devices under accelerated stress; The cumulative failure function of the candidate distribution model with the highest goodness of fit determined by the goodness-of-fit screening mechanism. Modeling is performed, in which, η For scale parameters, β For shape parameters; Based on the cumulative failure function, the failure duration data is fitted to calculate the value under accelerated stress. η a and β a ,in, η a To accelerate the dimensional parameters under stress, β a Shape parameters under accelerated stress; Construct an accelerated degradation model and calculate the acceleration factor. A F The dimensional parameters under accelerated stress are converted to those under normal stress. The shape parameters remain unchanged. Thus, a lifetime distribution model under normal stress is obtained.

4. The method for predicting the service life and planning the maintenance cycle of medical equipment according to claim 3, characterized in that: The goodness-of-fit screening mechanism includes: Calculate the goodness of fit of failure duration data under candidate distribution models such as Weibull distribution and log-normal distribution. R 2 The calculation formula is: in, n The number of failure duration data. ti For the first i Each failure duration data point Fo(ti) The cumulative failure rate for model fitting. Fn(ti) The standard cumulative failure rate, For all Fn(ti) The average value; The candidate distribution model with the highest goodness of fit was selected as the lifetime distribution model.

5. The method for predicting the service life and planning the maintenance cycle of medical equipment according to claim 1, characterized in that: In step S4, the process of constructing the maintenance decision model using the analytic hierarchy process includes: Determine the set of criteria-level factors, including remaining life risk factors, health status index, and maintenance economics factors; Based on the set of criteria layer factors, a pairwise comparison judgment matrix is ​​constructed using the analytic hierarchy process (AHP), and after passing the consistency test, the weight vector of the set of criteria layer factors is calculated. The original calculated values ​​of the remaining life risk factor, health status index and maintenance economy factor are normalized and mapped to the interval [0, 1]. Equipment maintenance urgency index MI The calculation formula is: ； in, R r , H , C m These are the normalized values ​​of the remaining life risk factor, health status index, and maintenance economics factor, respectively. W R , W H and W C These are the components corresponding to the remaining life risk factor, health status index, and maintenance economy factor in the weight vector, respectively. W R , W H and W C All are weighting coefficients.

6. The method for predicting the service life and planning the maintenance cycle of medical equipment according to claim 5, characterized in that: The original calculated value of the remaining life risk factor R r原始 The calculation formula is: ； in, T RUL_预测 This is a predicted service life value. T 设计寿命 This refers to the rated design life of the equipment.

7. The method for predicting the service life and planning the maintenance cycle of medical equipment according to claim 6, characterized in that: The original calculated value of the current health index H 原始 The calculation formula is: ； in, T 0 represents the set device health baseline value. S k The first one is calculated directly based on the equipment operation data. k The standardized degradation value of each performance indicator W k For the first k Preset weights for each performance indicator.

8. The method for predicting the service life and planning the maintenance cycle of medical equipment according to claim 7, characterized in that: In step S4, the recommended maintenance cycle and maintenance level are dynamically calculated and output, specifically including: Establish maintenance urgency index in advance MI The first mapping relationship with the recommended maintenance cycle, and the maintenance urgency index. MI The second mapping relationship with maintenance level; Based on the calculated maintenance urgency index MI The recommended maintenance cycle is determined based on the first mapping relationship, and the maintenance level is determined based on the second mapping relationship.

9. The method for predicting the service life and planning the maintenance cycle of medical equipment according to claim 8, characterized in that: In step S5, the process of generating a customized preventative maintenance plan includes: Based on the maintenance level, retrieve the basic maintenance items corresponding to that maintenance level from the preset maintenance knowledge base; Based on the historical failure modes, the basic maintenance items are supplemented and prioritized to form a maintenance item list and its priority ranking result; Based on the equipment identification, query the inventory status of the components involved in the maintenance project list, mark the warning for components whose inventory is lower than the safety threshold, generate component inventory warning information and trigger procurement suggestions; By integrating the maintenance item list, priority ranking results, and component inventory warning information, a customized preventive maintenance plan is generated.

10. A system for predicting the service life and maintenance cycle planning of medical equipment, characterized in that: include: The multi-source data acquisition and preprocessing module is used to synchronously acquire equipment operation data, environmental status data, and historical maintenance data of the medical equipment to be predicted through a distributed data acquisition interface, and to perform standardized preprocessing on the acquired data. The cross-domain data fusion and feature extraction module is used to construct a graph convolution-gated cyclic spatiotemporal fusion network model, extract and fuse spatiotemporal features of standardized preprocessed data based on the graph convolution-gated cyclic spatiotemporal fusion network model, generate feature vectors reflecting the comprehensive health status of equipment, and calculate the health status index based on the feature vectors. The lifetime distribution prediction module is used to input the feature vector into a pre-trained lifetime distribution model to obtain the corresponding lifetime distribution parameters, and output the lifetime prediction value based on the lifetime distribution parameters. The lifetime distribution model is determined from multiple candidate distribution models through a goodness-of-fit screening mechanism. The dynamic maintenance cycle planning module is used to construct a maintenance decision model based on the life distribution parameters and the health status index using the analytic hierarchy process, and dynamically calculate and output recommended maintenance cycles and maintenance levels. The maintenance plan generation and output module is used to generate customized preventive maintenance plans based on the recommended maintenance cycle and maintenance level, combined with historical failure modes and component inventory status obtained from historical maintenance data, and push them to the operation and maintenance management terminal.

Images