Fault diagnosis prediction method and system based on operation and maintenance scene

Abstract

The invention relates to the technical field of operation and maintenance scenes, in particular to a fault diagnosis prediction method and system based on an operation and maintenance scene, and the method comprises the steps: collecting equipment data, and constructing a scene fault diagnosis model; performing nonlinear fusion on the collected operation and maintenance data and historical maintenance records to obtain a real-time data fusion result; constructing an adaptive prediction algorithm based on interaction of a scenarized fault diagnosis model and real-time data fusion; and establishing a multi-level warning system according to a prediction result. According to the method, through multi-dimensional data acquisition, nonlinear data fusion, adaptive algorithm prediction and a multi-level warning system, accurate prediction and effective prevention of equipment faults are realized. Experimental results show that compared with a traditional method and a general machine learning method, the method has remarkable advantages in the aspects of early warning accuracy, false alarm rate, early warning time in advance and the like, meanwhile, the maintenance cost is effectively reduced, the downtime of equipment is shortened, and the method has good application prospects.

Description

Technical Field

[0001] The present invention relates to the technical field of operation and maintenance scene technology, and in particular to a fault diagnosis and prediction method and system based on operation and maintenance scene. Background Art

[0002] With the rapid development of industrial automation and intelligent manufacturing, the reliability and stability of equipment play a vital role in production efficiency and economic benefits. Traditional fault diagnosis and prediction methods mainly rely on fixed thresholds and simple statistical models, which cannot effectively cope with complex and changeable operation and maintenance scenarios. In recent years, with the advancement of big data analysis and machine learning technology, data-driven fault diagnosis and prediction methods have gradually become a research hotspot. These methods collect and analyze massive amounts of equipment operation data and construct complex mathematical models in order to achieve more accurate fault prediction and diagnosis. However, current research still has some limitations, such as the lack of adaptability of the model to environmental factors and load changes, the difficulty in effectively integrating multi-source heterogeneous data, and the lack of interpretability of prediction results.

[0003] Existing fault diagnosis and prediction methods face many challenges in practical applications. First, most methods fail to fully consider the particularity of operation and maintenance scenarios, and it is difficult to capture the dynamic characteristics of equipment under different working conditions. Secondly, traditional data fusion technology often uses simple linear combinations or weighted averages, which cannot effectively handle complex nonlinear relationships and time series dependencies. Furthermore, most existing prediction models use static parameter settings, which are difficult to adapt to the dynamic changes of equipment status and environment. In addition, many methods lack fine-grained risk level classification and corresponding operation suggestions when outputting warning information, making it difficult for maintenance personnel to respond quickly and appropriately. Finally, existing technologies generally lack consideration of the system as a whole, making it difficult to assess the potential impact of a single equipment failure on the entire production system. Therefore, it is urgent to develop a fault diagnosis and prediction method that can comprehensively consider the characteristics of operation and maintenance scenarios, realize dynamic fusion of multi-source data, have adaptive prediction capabilities, and provide refined warnings and decision support. Summary of the invention

[0004] In view of the problems existing in the prior art, the present invention is proposed.

[0005] Therefore, the problem to be solved by the present invention is that most methods fail to fully consider the particularity of operation and maintenance scenarios, and it is difficult to capture the dynamic characteristics of equipment under different working conditions. Secondly, traditional data fusion technology often adopts simple linear combinations or weighted averages, which cannot effectively handle complex nonlinear relationships and time series dependencies. Furthermore, most of the existing prediction models use static parameter settings, which are difficult to adapt to the dynamic changes of equipment status and environment. In addition, many methods lack fine-grained risk level division and corresponding operation suggestions when outputting early warning information, which makes it difficult for maintenance personnel to respond quickly and appropriately. Finally, the existing technology generally lacks consideration of the system as a whole, making it difficult to assess the potential impact of a single equipment failure on the entire production system.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, an embodiment of the present invention provides a fault diagnosis and prediction method based on an operation and maintenance scenario, which includes collecting equipment data, building a scenario-based fault diagnosis model, and monitoring the operation of the equipment in real time through the scenario-based fault diagnosis model;

[0008] The collected operation and maintenance data and the historical maintenance record data of the equipment are integrated to obtain real-time data fusion. The output of the scenario-based fault diagnosis model is used as the input of the real-time data fusion. The health index and the current equipment status and potential failure risk of the current equipment are obtained based on the real-time data fusion.

[0009] An adaptive prediction algorithm is obtained based on the interaction between the scenario-based fault diagnosis model and real-time data fusion. The warning information of the equipment is judged according to the value output by the adaptive prediction algorithm. The scope of measures that the operator should take corresponds to the degree of the warning information.

[0010] As a preferred solution of the fault diagnosis and prediction method based on the operation and maintenance scene of the present invention, wherein: the collected equipment data includes basic equipment information, operation status data, environmental factors, maintenance history records, maintenance operation data, load conditions, related equipment status, and early warning information;

[0011] Basic equipment information: equipment model, manufacturer, service life; equipment specifications and parameters; normal operating parameter range;

[0012] Operation status data: temperature, pressure, vibration, current, voltage, power; operation time, start and stop times;

[0013] Environmental factors: ambient temperature and humidity; dust concentration and corrosive gas concentration; geographical factors such as altitude and air pressure;

[0014] Maintenance history records: fault type and frequency, maintenance measures and effects, and replacement parts records;

[0015] Maintenance operation data: maintenance personnel's operation steps and sequence, maintenance tool usage, maintenance duration;

[0016] Load conditions: current load level of the equipment, load change trend, and peak load record;

[0017] Related equipment status: operating status of upstream and downstream equipment, and overall system operating parameters;

[0018] Early warning information: warning information automatically generated by the system and abnormal phenomena observed by operators.

[0019] As a preferred solution of the fault diagnosis and prediction method based on operation and maintenance scenarios described in the present invention, the steps of collecting equipment data, building a scenario-based fault diagnosis model, and monitoring the equipment operation status in real time through the scenario-based fault diagnosis model are as follows:

[0020] The data is preprocessed, the actual operation and maintenance data is normalized and feature synthesized, and multiple sensor data are synthesized using logarithmic changes. The UC Huang Anqi data includes temperature, pressure, and vibration. The multidimensional feature characterization emphasizes extreme values ​​and nonlinear relationships. The multidimensional feature synthesis can be expressed as follows:

[0021]

[0022] Among them, α i is the feature weight, ∈ is a small amount to prevent the logarithm from being negative; calculate the environmental adjustment factor, which emphasizes the nonlinear effect of environmental factors on device performance through an exponential function. The environmental adjustment factor can be expressed as follows:

[0023]

[0024] According to data preprocessing, a dynamic health status model can be constructed, which includes equipment health score and load sensitivity analysis. The dynamic health status model dynamically reflects the equipment status based on real-time data; the environmental adjustment factor and maintenance history are brought into the equipment health score to obtain the performance of the equipment itself after evaluation. The equipment health score can be expressed as follows:

[0025]

[0026] Among them, the logistic function model is used to calculate the equipment health score, Maintenance score Among them, the gamma coefficient is calculated based on the maintenance history and replacement parts record to control the influence of various factors;

[0027] The load sensitivity analysis further adjusts the health score. The load sensitivity analysis adjusts the health score through a hyperbolic tangent function. The load sensitivity analysis expresses the relationship between the load level and the optimal load of the device, which uses the following expression:

[0028] DHS adjusted =DHS·(1-δ·tanh(Load level -Load optimal ))

[0029] Among them, δ is the adjustment coefficient, DHS adjusted For load sensitivity analysis;

[0030] The inherent faults are predicted by the output value of the health status assessment, and the fault probability model is constructed by load sensitivity analysis. The overall system impact model is constructed by the status of related equipment. The fault probability model uses the square ratio formula to increase the sensitivity to the intermediate value of the health score, and improves the accuracy of the fault probability model in predicting subtle changes. The overall system impact model considers the interaction between upstream and downstream equipment. The fault probability model and the overall system impact model are respectively expressed as follows:

[0031]

[0032]

[0033] Where P(failure) is the failure probability model, P system (failure) is the overall impact model of the system, σ is the normalization function, κ ij Interaction is the intensity of interaction between devices. ij The interaction between the states of the devices.

[0034] As a preferred solution of the fault diagnosis and prediction method based on operation and maintenance scenarios described in the present invention, wherein: the operation and maintenance data collected by the data are fused with the historical maintenance record data of the equipment to obtain real-time data fusion, the output of the scenario-based fault diagnosis model is used as the input of the real-time data fusion, and the health index and the current equipment status and the potential failure risk of the current equipment are obtained according to the real-time data fusion;

[0035] The real-time collected operation and maintenance data is subjected to power transformation and weighted summation, feature fusion is constructed according to the equipment model and maintenance history record, and feature fusion is combined with exponentially weighted environmental factors to obtain the impact of nonlinear reaction environment on equipment performance, which is expressed as follows:

[0036]

[0037] Among them, the parameter α iand β i Dynamic adjustment based on equipment model and historical maintenance records; γ j is the weight of environmental factors;

[0038] The logistic regression model is used to calculate the health index based on the environmentally adjusted characteristics and construct the basic health index. Considering the deviation between the current load and the optimal load and the historical maintenance effect, the basic health index is further adjusted and the adjustment model is constructed. The health index and the adjustment model are expressed as follows:

[0039]

[0040] BHI final =BHI·(1-λ·tanh(LL opt )+ξ·M effect )

[0041] Among them, BHI is the health index, BHI final is the adjustment model, k and θ are model parameters used to adjust the shape and position of the curve; parameters λ and ξ adjust the influence of load deviation and maintenance history; L is the current load; L opt is the optimal load; M effect It is the historical maintenance effect;

[0042] The square function of BHI is used to enhance the sensitivity of the model to the median value of the health index, thereby improving the accuracy of predicting the probability of failure and building a failure probability prediction model. At the same time, historical and real-time data are combined to increase the responsiveness of the model to the current state of the equipment and build real-time data fusion. The failure probability prediction model and real-time data fusion are expressed as follows:

[0043]

[0044] Data fusion =ω·P(failure)+(1-ω)·R hist

[0045] Where P(failure) is the probability of failure; Data fusion is real-time data fusion; R hist is the historical maintenance record, and the weight ω is dynamically adjusted according to the recent operation and maintenance frequency.

[0046] As a preferred solution of the fault diagnosis and prediction method based on operation and maintenance scenarios of the present invention, wherein: the adaptive prediction algorithm is obtained according to the interaction between the scenario-based fault diagnosis model and the real-time data fusion, the warning information of the equipment is judged according to the value output by the adaptive prediction algorithm, and the scope of measures that the operator should take is respectively corresponded according to the degree of the warning information;

[0047] Among them, by combining the results of data fusion and data trends, the adaptive algorithm prediction model can capture the dynamic changes of equipment status over time. The adaptive adjustment mechanism of parameters ensures that the model can continue to learn and adapt to the new operation and maintenance environment. The sigmoid function is used to establish a basic prediction model; at the same time, the adaptive adjustment mechanism of parameters is adopted, which uses the following expressions:

[0048] P adaptive (t) = sigmoid(a·Data fusion (t)+b·trend(Data fusion ,t)+c)

[0049]

[0050] Where a, b, c are model parameters, Data fusion (t) is the output of real-time data fusion; in addition, trend (Data fusion ,t) Trend function This function analyzes the trend of data changing over time and improves the model's ability to capture time series features; the gradient descent method is used to automatically adjust the model parameters θ = {a, b, c}, η is the learning rate, is the loss function, y(t) is the actual operation and maintenance data label at time, and the actual operation and maintenance data label includes fault or non-fault. By continuously adjusting parameters to minimize the prediction error, the model can adapt to new operation and maintenance environments and data changes.

[0051] As a preferred solution of the fault diagnosis and prediction method based on the operation and maintenance scene of the present invention, wherein: the warning information of the equipment is judged according to the value output by the adaptive prediction algorithm, and the scope of measures that the operator should take is respectively corresponded according to the degree of the warning information, which includes constructing a warning level decision according to the output of the adaptive prediction model, and calculating the warning level, and converting the prediction result into a specific warning level, such as normal, attention, warning and emergency, by quantifying the probability value; and then generating an operation suggestion for the current equipment state according to the warning level and the actions taken at similar levels in history, and combining the self-adaptive prediction model to obtain a warning and a feasible operation decision, which respectively adopts the following expressions:

[0052]

[0053] Action=f(Warning l Eve, Historical a ctions)

[0054] Where k is a coefficient used to adjust the warning sensitivity; P adaptive(t) is the adaptive algorithm prediction model; Action is the warning level decision; Historical a Actions are actions taken at similar levels in history.

[0055] As a preferred solution of the fault diagnosis and prediction method based on operation and maintenance scenarios of the present invention, the prediction algorithm is automatically adjusted according to the specific operation and maintenance scenarios and real-time data, and the health status of the equipment is updated in real time. The output of the algorithm is used to judge the warning information of the equipment, and according to the different warning levels, corresponding operation suggestions are put forward. These warning information are classified according to predefined risk levels and guide corresponding operation behaviors:

[0056] Low risk stage (P adaptive <first threshold); when the warning probability output by the adaptive prediction model is lower than the first threshold, a warning message is sent to relevant departments and the public that may be affected, and a low-intensity virtual failure scenario is displayed for risk awareness training.

[0057] Maintain normal monitoring and routine maintenance procedures;

[0058] Medium risk stage (first threshold ≤ P adaptive ≤ the second threshold); the warning probability output by the adaptive prediction model is between the first threshold and the second threshold; send enhanced warning information to relevant departments, form and display medium-intensity virtual failure scenarios for risk communication and emergency drills, activate emergency response plans, and prepare necessary evacuation and rescue resources.

[0059] High risk stage (P adaptive > second threshold); when the warning probability output by the adaptive prediction model exceeds the second threshold; issue an emergency warning to all relevant departments and the public,

[0060] Generate and display high-intensity virtual failure scenarios for evacuation route planning and rescue strategy formulation, and immediately initiate evacuation procedures and emergency response measures.

[0061] In a second aspect, an embodiment of the present invention provides a fault diagnosis and prediction system based on an operation and maintenance scenario, which includes a collection module for collecting equipment data, building a scenario-based fault diagnosis model, and monitoring the operation of the equipment in real time through the scenario-based fault diagnosis model;

[0062] The fusion module integrates the collected operation and maintenance data and the historical maintenance record data of the equipment to obtain real-time data fusion. The output of the scenario-based fault diagnosis model is used as the input of the real-time data fusion. The health index and the current equipment status and potential failure risk of the current equipment are obtained based on the real-time data fusion.

[0063] The prediction module obtains an adaptive prediction algorithm based on the interaction between the scenario-based fault diagnosis model and the real-time data fusion, determines the warning information of the equipment according to the value output by the adaptive prediction algorithm, and corresponds to the scope of measures that the operator should take according to the degree of the warning information.

[0064] In a third aspect, an embodiment of the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program instructions are executed by the processor, the steps of the fault diagnosis and prediction method based on the operation and maintenance scenario as described in the first aspect of the present invention are implemented.

[0065] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program instructions are executed by a processor, the steps of the fault diagnosis and prediction method based on the operation and maintenance scenario as described in the first aspect of the present invention are implemented.

[0066] The beneficial effects of the present invention are as follows: by collecting comprehensive equipment data including basic equipment information, operating status data, environmental factors, etc., a scenario-based fault diagnosis model is constructed. This model can monitor the operation of the equipment in real time, and fully considers the particularity of the operation and maintenance scene. This step realizes the accurate capture of the dynamic characteristics of the equipment, and effectively improves the accuracy and applicability of fault diagnosis.

[0067] By fusing the collected operation and maintenance data with historical maintenance records, real-time data fusion results are obtained. This step uses nonlinear data processing methods such as power transformation and exponential weighting to effectively handle complex nonlinear relationships and time series dependencies. This method significantly improves the quality of data fusion, making the assessment of health index and potential failure risks more accurate and reliable.

[0068] Through the interaction of scenario-based fault diagnosis model and real-time data fusion, an adaptive prediction algorithm is constructed. The algorithm uses sigmoid function and gradient descent method to automatically adjust parameters according to new operation and maintenance environment and data changes. This adaptive mechanism greatly improves the model's ability to adapt to dynamic changes in equipment status and environment, making fault prediction more accurate and timely.

[0069] A multi-level warning information system is established based on the output values ​​of the adaptive prediction algorithm. This step not only quantifies the risk of failure, but also provides corresponding operational suggestions based on the risk level. This fine-grained risk classification and decision support mechanism enables maintenance personnel to quickly make appropriate responses, effectively reducing the possibility of failure and potential losses.

[0070] By considering the interactions between devices, a system-wide impact model was constructed. This step evaluates the potential impact of a single device failure on the entire production system, making the leap from single-point failure prediction to system-level risk assessment. This holistic consideration significantly improves the effectiveness of fault prevention and risk management, and helps ensure the stable operation of the entire production system.

[0071] By introducing virtual failure scenarios, risk awareness training, emergency drills and evacuation strategy formulation at different risk levels were achieved. This innovative training method greatly enhanced the risk awareness and emergency response capabilities of operators, laying the foundation for rapid response when actual failures occur.

[0072] In summary, this fault diagnosis and prediction method based on operation and maintenance scenarios achieves accurate prediction and effective prevention of equipment failures through innovative steps such as multi-dimensional data collection, nonlinear data fusion, adaptive algorithm prediction, multi-level warning system and overall impact assessment. It not only improves the accuracy and timeliness of fault diagnosis, but also provides clear decision-making support for maintenance personnel, ultimately achieving the goal of improving equipment reliability, reducing maintenance costs and ensuring production efficiency. This method has broad application prospects in the fields of industrial automation and intelligent manufacturing, and is of great significance to improving the overall stability and economic benefits of production systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0074] Figure 1 This is a flow chart of a fault diagnosis and prediction method based on operation and maintenance scenarios;

[0075] Figure 2 Computer equipment diagram of the fault diagnosis and prediction method based on operation and maintenance scenarios. DETAILED DESCRIPTION

[0076] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0077] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0078] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor is it a separate or selective embodiment that is mutually exclusive with other embodiments.

[0079] Example 1

[0080] Reference Figure 1-2 , which is the first embodiment of the present invention, and provides a fault diagnosis and prediction method based on an operation and maintenance scenario, comprising:

[0081] S100: Collect equipment data, build a scenario-based fault diagnosis model, and monitor the equipment operation status in real time through the scenario-based fault diagnosis model;

[0082] S101: Collect equipment data including basic equipment information, operating status data, environmental factors, maintenance history records, maintenance operation data, load conditions, related equipment status, and early warning information;

[0083] Basic equipment information: equipment model, manufacturer, service life; equipment specifications and parameters; normal operating parameter range;

[0084] Operation status data: temperature, pressure, vibration, current, voltage, power; operation time, start and stop times;

[0085] Environmental factors: ambient temperature and humidity; dust concentration and corrosive gas concentration; geographical factors such as altitude and air pressure;

[0086] Maintenance history records: fault type and frequency, maintenance measures and effects, and replacement parts records;

[0087] Maintenance operation data: maintenance personnel's operation steps and sequence, maintenance tool usage, maintenance duration;

[0088] Load conditions: current load level of the equipment, load change trend, and peak load record;

[0089] Related equipment status: operating status of upstream and downstream equipment, and overall system operating parameters;

[0090] Early warning information: warning information automatically generated by the system and abnormal phenomena observed by operators.

[0091] S102: Collect equipment data, build a scenario-based fault diagnosis model, and monitor the equipment operation status in real time through the scenario-based fault diagnosis model. The steps are as follows:

[0092] Preprocess the data, normalize and synthesize the features of the actual operation and maintenance data, and use logarithmic changes to synthesize multiple sensor data. UC Huang Anqi's data includes temperature, pressure, and vibration. Multidimensional feature representation emphasizes extreme values ​​and nonlinear relationships. Multidimensional feature synthesis can be expressed as follows:

[0093]

[0094] Among them, α i is the feature weight, ∈ is a small amount to prevent the logarithm from being negative; calculate the environmental adjustment factor, which emphasizes the nonlinear effect of environmental factors on device performance through an exponential function. The environmental adjustment factor can be expressed as follows:

[0095]

[0096] According to data preprocessing, a dynamic health status model can be constructed, which includes equipment health score and load sensitivity analysis. The dynamic health status model dynamically reflects the equipment status based on real-time data; the environmental adjustment factor and maintenance history are brought into the equipment health score to obtain the performance of the equipment itself after evaluation. The equipment health score can be expressed as follows:

[0097]

[0098] Among them, the logistic function model is used to calculate the equipment health score, Maintenance score Among them, the gamma coefficient is calculated based on the maintenance history and replacement parts record to control the influence of various factors;

[0099] The load sensitivity analysis further adjusts the health score. The load sensitivity analysis adjusts the health score through the hyperbolic tangent function. The load sensitivity analysis expresses the relationship between the load level and the optimal load of the device. It uses the following expression:

[0100] DHS adjusted =DHS·(1-δ·tanh(Load level -Load optimal ))

[0101] Among them, δ is the adjustment coefficient, DHS adjusted For load sensitivity analysis;

[0102] Potential faults are predicted through the output values ​​of health status assessment, and a fault probability model is constructed through load sensitivity analysis. The overall system impact model is constructed through the status of related equipment. The fault probability model uses the square ratio formula to increase the sensitivity to the intermediate value of the health score, and improves the accuracy of the fault probability model in predicting subtle changes. The overall system impact model considers the interaction between upstream and downstream equipment. The fault probability model and the overall system impact model use the following expressions respectively:

[0103]

[0104] Where P(failure) is the failure probability model, P system (failure) is the overall impact model of the system, σ is the normalization function, κ ij Interaction is the intensity of interaction between devices. ij The interaction between the states of the devices.

[0105] S200: The collected operation and maintenance data and the historical maintenance record data of the equipment are integrated to obtain real-time data fusion, and the output of the scenario-based fault diagnosis model is used as the input of the real-time data fusion. The health index and the current equipment status and the potential failure risk of the current equipment are obtained according to the real-time data fusion;

[0106] S201: The collected operation and maintenance data and the historical maintenance record data of the equipment are integrated to obtain real-time data fusion, and the output of the scenario-based fault diagnosis model is used as the input of the real-time data fusion. The health index and the current equipment status and the potential failure risk of the current equipment are obtained according to the real-time data fusion;

[0107] The real-time collected operation and maintenance data is subjected to power transformation and weighted summation, feature fusion is constructed according to the equipment model and maintenance history record, and feature fusion is combined with exponentially weighted environmental factors to obtain the impact of nonlinear reaction environment on equipment performance, which is expressed as follows:

[0108]

[0109] Among them, the parameter α i and β i Dynamic adjustment based on equipment model and historical maintenance records; γ j is the weight of environmental factors;

[0110] The logistic regression model is used to calculate the health index based on the environmentally adjusted characteristics and construct the basic health index. Considering the deviation between the current load and the optimal load and the historical maintenance effect, the basic health index is further adjusted and the adjustment model is constructed. The health index and the adjustment model are expressed as follows:

[0111]

[0112] BHI final =BHI·(1-λ·tanh(LL opt )+ξ·M effect )

[0113] Among them, BHI is the health index, BHI final is the adjustment model, k and θ are model parameters used to adjust the shape and position of the curve; parameters λ and ξ adjust the influence of load deviation and maintenance history; L is the current load; L opt is the optimal load; M effect It is the historical maintenance effect;

[0114] The square function of BHI is used to enhance the sensitivity of the model to the median of the health index, thereby improving the accuracy of predicting the probability of failure and building a failure probability prediction model. At the same time, historical and real-time data are combined to increase the model's response to the current state of the equipment to build real-time data fusion, where the failure probability prediction model and real-time data fusion are expressed as follows:

[0115]

[0116] Data fusion =ω·P(failure)+(1-ω)·R hist

[0117] Where P(failure) is the probability of failure; Data fusion is real-time data fusion; R hist is the historical maintenance record, and the weight ω is dynamically adjusted according to the recent operation and maintenance frequency.

[0118] S300: An adaptive prediction algorithm is obtained based on the interaction between the scenario-based fault diagnosis model and the real-time data fusion, and the warning information of the device is determined based on the value output by the adaptive prediction algorithm. The scope of measures that the operator should take corresponds to the degree of the warning information.

[0119] S301: obtaining an adaptive prediction algorithm based on the interaction between the scenario-based fault diagnosis model and the real-time data fusion, determining the warning information of the device based on the value output by the adaptive prediction algorithm, and corresponding to the scope of measures that the operator should take based on the degree of the warning information;

[0120] Among them, by combining the results of data fusion and data trends, the adaptive algorithm prediction model can capture the dynamic changes of equipment status over time. The adaptive adjustment mechanism of parameters ensures that the model can continue to learn and adapt to the new operation and maintenance environment. The sigmoid function is used to establish a basic prediction model; at the same time, the adaptive adjustment mechanism of parameters is adopted, which uses the following expressions:

[0121] P adaptive (t) = sigmoid(a·Data fusion (t)+b·trend(Data fusion ,t)+c)

[0122]

[0123] Where a, b, c are model parameters, Data fusion (t) is the output of real-time data fusion; in addition, trend (Data fusion ,t) Trend function This function analyzes the trend of data changing over time and improves the model's ability to capture time series features; the gradient descent method is used to automatically adjust the model parameters θ = {a, b, c}, η is the learning rate, is the loss function, y(t) is the actual operation and maintenance data label at time, and the actual operation and maintenance data label includes fault or non-fault. By continuously adjusting parameters to minimize the prediction error, the model can adapt to new operation and maintenance environments and data changes.

[0124] S302: Determine the warning information of the device according to the value output by the adaptive prediction algorithm, and correspond the scope of measures that the operator should take according to the degree of the warning information. It includes building a warning level decision according to the output of the adaptive prediction model, and calculating the warning level. By quantifying the probability value, the prediction result is converted into a specific warning level, such as normal, attention, warning and emergency; then generate operation suggestions for the current device status according to the warning level and the actions taken at similar levels in history, and combine the adaptive prediction model to obtain warnings and feasible operation decisions, which are respectively expressed as follows:

[0125]

[0126] Action=f(Warning l Eve, Historical a ctions)

[0127] Where k is a coefficient used to adjust the warning sensitivity; P adaptive (t) is the adaptive algorithm prediction model; Action is the warning level decision; Historical aActions are actions taken at similar levels in history.

[0128] S303: Automatically adjust the prediction algorithm according to the specific operation and maintenance scenario and real-time data, and update the health status of the equipment in real time. The output of the algorithm is used to determine the warning information of the equipment, and put forward corresponding operation suggestions according to the different warning levels. These warning information are classified according to the predefined risk level and guide the corresponding operation behavior:

[0129] Low risk stage (P adaptive (first threshold); when the warning probability output by the adaptive prediction model is lower than the first threshold, a warning message is sent to relevant departments and the public that may be affected, and a low-intensity virtual failure scenario is displayed for risk awareness training.

[0130] Maintain normal monitoring and routine maintenance procedures;

[0131] Medium risk stage (first threshold ≤ P adaptive ≤ the second threshold); the warning probability output by the adaptive prediction model is between the first threshold and the second threshold; send enhanced warning information to relevant departments, form and display medium-intensity virtual failure scenarios for risk communication and emergency drills, activate emergency response plans, and prepare necessary evacuation and rescue resources.

[0132] High risk stage (P adaptive 〉Second threshold); when the warning probability output by the adaptive prediction model exceeds the second threshold; issue an emergency warning to all relevant departments and the public,

[0133] Generate and display high-intensity virtual failure scenarios for evacuation route planning and rescue strategy formulation, and immediately initiate evacuation procedures and emergency response measures.

[0134] The value ranges of each threshold value can be as follows:

[0135] Low risk threshold (first threshold): P adaptive <0.3

[0136] Medium risk threshold (first threshold ≤ P adaptive ≤ second threshold): 0.3≤P adaptive ≤0.7 high risk threshold (second threshold): P adaptive >0.7

[0137] These thresholds are set based on the importance of the equipment, the extent of the loss that a failure may cause, and the difficulty of repair. A warning probability below 0.3 is considered low risk because the probability of a failure occurring is low, or even if it occurs, it will not cause serious consequences.

[0138] An alert probability between 0.3 and 0.7 is considered a medium risk, requiring attention and preparation.

[0139] A warning probability higher than 0.7 is considered high risk and requires immediate action. This threshold setting can effectively prevent major failures and avoid excessive maintenance, achieving a balance between the economy and reliability of equipment maintenance.

[0140] Scenario-based fault diagnosis model: This model builds a dynamic and adaptable diagnostic framework by integrating multi-dimensional equipment data, such as basic equipment information, operating status data, and environmental factors. This model can accurately capture the characteristics of equipment in different operation and maintenance scenarios, significantly improving the accuracy and reliability of fault diagnosis.

[0141] Real-time data fusion: nonlinear data processing methods such as power transformation and exponential weighting are used to effectively handle complex nonlinear relationships and time series dependencies. This method overcomes the limitations of traditional linear fusion and provides a more accurate health index assessment.

[0142] Adaptive prediction algorithm: Using the sigmoid function and gradient descent method, the model parameters are dynamically adjusted. This adaptive mechanism enables the algorithm to quickly respond to changes in device status and environment, greatly improving the accuracy and real-time performance of predictions.

[0143] Multi-level warning system: By quantifying the prediction results, a refined risk level classification is established. This system not only provides a more accurate risk assessment, but also provides maintenance personnel with corresponding operational suggestions, improving the efficiency and effectiveness of fault response.

[0144] System overall impact model: Considering the interaction between devices, it realizes the assessment from single point failure to system-level risk. This holistic consideration significantly improves the comprehensiveness and accuracy of risk management.

[0145] Furthermore, this embodiment also provides a fault diagnosis and prediction system based on operation and maintenance scenarios, including:

[0146] The acquisition module collects equipment data, builds a scenario-based fault diagnosis model, and monitors the equipment operation status in real time through the scenario-based fault diagnosis model;

[0147] The fusion module integrates the collected operation and maintenance data and the historical maintenance record data of the equipment to obtain real-time data fusion. The output of the scenario-based fault diagnosis model is used as the input of the real-time data fusion. The health index and the current equipment status and potential failure risk of the current equipment are obtained based on the real-time data fusion.

[0148] The prediction module obtains an adaptive prediction algorithm based on the interaction between the scenario-based fault diagnosis model and the real-time data fusion, determines the warning information of the equipment according to the value output by the adaptive prediction algorithm, and corresponds to the scope of measures that the operator should take according to the degree of the warning information.

[0149] This embodiment also provides a computer device, which is suitable for the fault diagnosis and prediction method based on operation and maintenance scenarios, including a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the fault diagnosis and prediction method based on operation and maintenance scenarios proposed in the above embodiment.

[0150] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0151] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the fault diagnosis and prediction method based on the operation and maintenance scenario proposed in the above embodiment is implemented.

[0152] In summary, a scenario-based fault diagnosis model was constructed by collecting comprehensive equipment data including basic equipment information, operating status data, environmental factors, etc. This model can monitor the operation of equipment in real time and fully consider the particularity of operation and maintenance scenarios. This step achieves accurate capture of the dynamic characteristics of the equipment and effectively improves the accuracy and applicability of fault diagnosis.

[0153] By fusing the collected operation and maintenance data with historical maintenance records, real-time data fusion results are obtained. This step uses nonlinear data processing methods such as power transformation and exponential weighting to effectively handle complex nonlinear relationships and time series dependencies. This method significantly improves the quality of data fusion, making the assessment of health index and potential failure risks more accurate and reliable.

[0154] Through the interaction of scenario-based fault diagnosis model and real-time data fusion, an adaptive prediction algorithm is constructed. The algorithm uses sigmoid function and gradient descent method to automatically adjust parameters according to new operation and maintenance environment and data changes. This adaptive mechanism greatly improves the model's ability to adapt to dynamic changes in equipment status and environment, making fault prediction more accurate and timely.

[0155] A multi-level warning information system is established based on the output values ​​of the adaptive prediction algorithm. This step not only quantifies the risk of failure, but also provides corresponding operational suggestions based on the risk level. This fine-grained risk classification and decision support mechanism enables maintenance personnel to quickly make appropriate responses, effectively reducing the possibility of failure and potential losses.

[0156] By considering the interactions between devices, a system-wide impact model was constructed. This step evaluates the potential impact of a single device failure on the entire production system, making the leap from single-point failure prediction to system-level risk assessment. This holistic consideration significantly improves the effectiveness of fault prevention and risk management, and helps ensure the stable operation of the entire production system.

[0157] By introducing virtual failure scenarios, risk awareness training, emergency drills and evacuation strategy formulation at different risk levels were achieved. This innovative training method greatly enhanced the risk awareness and emergency response capabilities of operators, laying the foundation for rapid response when actual failures occur.

[0158] In summary, this fault diagnosis and prediction method based on operation and maintenance scenarios achieves accurate prediction and effective prevention of equipment failures through innovative steps such as multi-dimensional data collection, nonlinear data fusion, adaptive algorithm prediction, multi-level warning system and overall impact assessment. It not only improves the accuracy and timeliness of fault diagnosis, but also provides clear decision-making support for maintenance personnel, ultimately achieving the goal of improving equipment reliability, reducing maintenance costs and ensuring production efficiency. This method has broad application prospects in the fields of industrial automation and intelligent manufacturing, and is of great significance to improving the overall stability and economic benefits of production systems.

[0159] Example 2

[0160] Reference Figure 1 - Figure 2 , which is the second embodiment of the present invention, and this embodiment provides a fault diagnosis and prediction method based on an operation and maintenance scenario. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0161] In order to verify the effectiveness of the fault diagnosis and prediction method based on operation and maintenance scenarios proposed in this invention, a set of key equipment of a large petrochemical enterprise was selected for testing. The device contains multiple key equipment, such as reactors, heat exchangers, compressors, etc. In the test preparation stage, firstly, comprehensive data collection is carried out on all key equipment in the device, including basic equipment information (model, service life, etc.), operating status data (temperature, pressure, vibration, etc.), environmental factors (ambient temperature, humidity, dust concentration, etc.), maintenance history records, maintenance operation data, load conditions and related equipment status, etc. The data collection lasted for 6 months, and a total of about 5 million pieces of original data were collected.

[0162] Next, the collected data is preprocessed and feature extracted. Using the multidimensional feature synthesis method proposed in the present invention, the sensor data such as temperature, pressure, and vibration are nonlinearly transformed and fused to obtain more representative comprehensive features. At the same time, the environmental adjustment factor is calculated by an exponential function to reflect the nonlinear impact of environmental factors on equipment performance. Based on the preprocessed data, a dynamic health status model is constructed, including equipment health score and load sensitivity analysis. The health score adopts a logical function model to comprehensively consider factors such as environmental adjustment factors and maintenance history; the load sensitivity analysis adjusts the health score through the hyperbolic tangent function to reflect the relationship between the load level and the optimal load of the equipment.

[0163] In the real-time data fusion stage, the power transformation and weighted summation methods are used to process the real-time collected operation and maintenance data, and combined with the exponentially weighted environmental factors to build a feature fusion model. The health index is calculated through the logistic regression model, and the deviation between the current load and the optimal load and the historical maintenance effect are considered to further adjust the health index. In order to increase the sensitivity near the median of the health index, the square function is used to build a fault probability prediction model.

[0164] The adaptive prediction algorithm uses the sigmoid function to establish a basic prediction model and introduces a trend function to analyze the trend of data changes over time. The gradient descent method is used to automatically adjust the model parameters to adapt to the new operation and maintenance environment and data changes. According to the prediction results, a multi-level warning system is established to divide the risk level into four levels: normal, attention, warning, and emergency. Based on the actions taken at similar levels in history, specific operation suggestions for the current equipment status are generated.

[0165] To fully evaluate the performance of this method, it was compared with traditional fixed threshold-based methods and general machine learning methods (such as support vector machines (SVMs)). The test lasted for 3 months, during which the number of warnings, accuracy, false alarm rate, and advance warning time of potential failures of each method were recorded. At the same time, maintenance costs and equipment downtime were calculated to evaluate the economic benefits of the method. The test results are shown in the following table:

[0166]

[0167] It can be seen from the test results that the fault diagnosis and prediction method based on operation and maintenance scenarios proposed in the present invention is significantly superior to traditional methods and general machine learning methods in multiple key indicators. First, in terms of warning accuracy, the accuracy of this method reaches 94.1%, which is 17.8 percentage points higher than the traditional fixed threshold method and 8.4 percentage points higher than the general machine learning method. This significant improvement is mainly due to the scenario-based fault diagnosis model and nonlinear data fusion technology used in this method, which can more accurately capture the dynamic characteristics of equipment in different operation and maintenance environments.

[0168] Secondly, the false alarm rate of this method is only 4.7%, which is much lower than that of the other two methods. This result fully demonstrates the advantages of this method in dealing with complex nonlinear relationships and timing dependencies. By comprehensively considering multi-dimensional data such as environmental factors, load conditions, and maintenance history, this method can more accurately distinguish between normal fluctuations and potential failures, thereby significantly reducing false alarms.

[0169] In terms of early warning timeliness, the average advance warning time of this method reaches 19.8 hours, which is 13.6 hours earlier than the traditional method and 7.3 hours earlier than the general machine learning method. This significant advance time reserves sufficient response time for maintenance personnel and greatly reduces the risk of sudden failures. This is mainly due to the adaptive prediction algorithm used in this method, which can adjust parameters in real time and quickly adapt to dynamic changes in equipment status and environment.

[0170] From the perspective of economic benefits, this method also performs well in reducing maintenance costs and reducing equipment downtime. Compared with general machine learning methods, this method reduces maintenance costs by an additional 12.3% and reduces equipment downtime by 16.3%. These improvements directly translate into significant economic benefits, not only improving production efficiency but also extending the service life of equipment.

[0171] It is worth noting that the number of warnings (85) of this method is lower than that of the other two methods, but the accuracy and advance warning time are higher. This shows that this method does not simply increase the frequency of warnings, but issues warnings when they are really needed through more accurate analysis. This "precise warning" capability greatly reduces the workload of maintenance personnel, allowing them to allocate time and resources more effectively.

[0172] In summary, the method proposed in the present invention has shown significant advantages in terms of early warning accuracy, timeliness and economic benefits. It not only overcomes the problem of insufficient adaptability of traditional methods to complex scenarios, but also surpasses general machine learning methods in dealing with nonlinear relationships and dynamic changes. This comprehensive performance improvement makes the method have broad application prospects in the fields of industrial automation and intelligent manufacturing, and provides strong support for improving the overall stability and economic benefits of the production system.

[0173] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should be included in the scope of the claims of the present invention.

Claims

1. A fault diagnosis and prediction method based on operation and maintenance scenarios, characterized by: include, Collect equipment data, build scenario-based fault diagnosis models, and use scenario-based fault diagnosis models to monitor equipment operation in real time; The collected operation and maintenance data and the historical maintenance record data of the equipment are integrated to obtain real-time data fusion. The output of the scenario-based fault diagnosis model is used as the input of the real-time data fusion. The health index and the current equipment status and potential failure risk of the current equipment are obtained based on the real-time data fusion. An adaptive prediction algorithm is obtained based on the interaction between the scenario-based fault diagnosis model and real-time data fusion. The warning information of the equipment is judged according to the value output by the adaptive prediction algorithm. The scope of measures that the operator should take corresponds to the degree of the warning information.

2. The fault diagnosis and prediction method based on operation and maintenance scenario according to claim 1 is characterized in that: The collected equipment data includes basic equipment information, operating status data, environmental factors, maintenance history records, maintenance operation data, load conditions, related equipment status, and early warning information; Basic equipment information: equipment model, manufacturer, service life; equipment specifications and parameters; normal operating parameter range; Operation status data: temperature, pressure, vibration, current, voltage, power; operation time, start and stop times; Environmental factors: ambient temperature and humidity; dust concentration and corrosive gas concentration; geographical factors such as altitude and air pressure; Maintenance history records: fault type and frequency, maintenance measures and effects, and replacement parts records; Maintenance operation data: maintenance personnel's operation steps and sequence, maintenance tool usage, maintenance duration; Load conditions: current load level of the equipment, load change trend, and peak load record; Related equipment status: operating status of upstream and downstream equipment, and overall system operating parameters; Early warning information: warning information automatically generated by the system and abnormal phenomena observed by operators.

3. The fault diagnosis and prediction method based on operation and maintenance scenario according to claim 2 is characterized in that: The steps of collecting device data, building a scenario-based fault diagnosis model, and monitoring the device operation status in real time through the scenario-based fault diagnosis model are as follows: The data is preprocessed, the actual operation and maintenance data is normalized and feature synthesized, and multiple sensor data are synthesized using logarithmic changes. The UC Huang Anqi data includes temperature, pressure, and vibration. The multidimensional feature characterization emphasizes extreme values ​​and nonlinear relationships. The multidimensional feature synthesis can be expressed as follows: Among them, α i is the feature weight, ∈ is a small amount to prevent the logarithm from being negative; calculate the environmental adjustment factor, which emphasizes the nonlinear effect of environmental factors on device performance through an exponential function. The environmental adjustment factor can be expressed as follows: According to data preprocessing, a dynamic health status model can be constructed, which includes equipment health score and load sensitivity analysis. The dynamic health status model dynamically reflects the equipment status based on real-time data; the environmental adjustment factor and maintenance history are brought into the equipment health score to obtain the performance of the equipment itself after evaluation. The equipment health score can be expressed as follows: Among them, the logistic function model is used to calculate the equipment health score, Maintenance score Among them, the gamma coefficient is calculated based on the maintenance history and replacement parts record to control the influence of various factors; The load sensitivity analysis further adjusts the health score. The load sensitivity analysis adjusts the health score through a hyperbolic tangent function. The load sensitivity analysis expresses the relationship between the load level and the optimal load of the device, which uses the following expression: DHS adjusted =DHS·(1-δ·tanh(Load level -Load optimal )) Among them, δ is the adjustment coefficient, DHS adjusted For load sensitivity analysis; Potential faults are predicted through the output values ​​of health status assessment, and a fault probability model is constructed through load sensitivity analysis. The overall system impact model is constructed through the status of related equipment. The fault probability model uses the square ratio formula to increase the sensitivity to the intermediate value of the health score, and improves the accuracy of the fault probability model in predicting subtle changes. The overall system impact model considers the interaction between upstream and downstream equipment. The fault probability model and the overall system impact model are expressed as follows: Where P(failure) is the failure probability model, P system (failure) is the overall impact model of the system, σ is the normalization function, κ ij Interaction is the intensity of interaction between devices. ij The interaction between the states of the devices.

4. The fault diagnosis and prediction method based on operation and maintenance scenario according to claim 3 is characterized in that: The collected operation and maintenance data and the historical maintenance record data of the equipment are integrated to obtain real-time data fusion, and the output of the scenario-based fault diagnosis model is used as the input of the real-time data fusion. The health index and the status of the current equipment and the potential failure risk of the current equipment are obtained according to the real-time data fusion; The real-time collected operation and maintenance data is subjected to power transformation and weighted summation, feature fusion is constructed according to the equipment model and maintenance history record, and feature fusion is combined with exponentially weighted environmental factors to obtain the impact of nonlinear reaction environment on equipment performance, which is expressed as follows: Among them, the parameter α i and β i Dynamic adjustment based on equipment model and historical maintenance records; γ j is the weight of environmental factors; The logistic regression model is used to calculate the health index based on the environmentally adjusted characteristics and construct the basic health index. Considering the deviation between the current load and the optimal load and the historical maintenance effect, the basic health index is further adjusted and the adjustment model is constructed. The health index and the adjustment model are expressed as follows: BHI final =BHI·(1-λ·tanh(LL opt )+ξM effect , Among them, BHI is the health index, BHI final is the adjustment model, k and θ are model parameters used to adjust the shape and position of the curve; parameters λ and ξ adjust the influence of load deviation and maintenance history; L is the current load; L opt is the optimal load; M effect It is the historical maintenance effect; The square function of BHI is used to enhance the sensitivity of the model to the median value of the health index, thereby improving the accuracy of predicting the probability of failure and building a failure probability prediction model. At the same time, historical and real-time data are combined to increase the responsiveness of the model to the current state of the equipment and build real-time data fusion. The failure probability prediction model and real-time data fusion are expressed as follows: Data fusion =ω·P(failure)+(1-ω)·R hist Where P(failure) is the failure probability; Data fusion is real-time data fusion; R hist is the historical maintenance record, and the weight ω is dynamically adjusted according to the recent operation and maintenance frequency.

5. The fault diagnosis and prediction method based on operation and maintenance scenario according to claim 4 is characterized in that: The adaptive prediction algorithm is obtained based on the interaction between the scenario-based fault diagnosis model and the real-time data fusion, the warning information of the equipment is judged according to the value output by the adaptive prediction algorithm, and the scope of measures that the operator should take corresponds to the degree of the warning information; Among them, by combining the results of data fusion and data trends, the adaptive algorithm prediction model can capture the dynamic changes of equipment status over time. The adaptive adjustment mechanism of parameters ensures that the model can continue to learn and adapt to the new operation and maintenance environment. The sigmoid function is used to establish a basic prediction model; at the same time, the adaptive adjustment mechanism of parameters is adopted, which uses the following expressions: P adaptive (t)=sigmoid(a·Data fusion (t)+b·trend(Data fusion ,t)+c) Where a, b, c are model parameters, Data fusion (t) is the output of real-time data fusion; in addition, trend (Data fusion ,t) Trend function This function analyzes the trend of data changing over time and improves the model's ability to capture time series features; the gradient descent method is used to automatically adjust the model parameters θ = {a, b, c}, η is the learning rate, is the loss function, y(t) is the actual operation and maintenance data label at time, and the actual operation and maintenance data label includes fault or non-fault. By continuously adjusting parameters to minimize the prediction error, the model can adapt to new operation and maintenance environments and data changes.

6. The fault diagnosis and prediction method based on operation and maintenance scenario according to claim 5 is characterized in that: The method comprises determining the warning information of the device according to the value output by the adaptive prediction algorithm, and corresponding to the scope of measures that the operator should take according to the degree of the warning information, which includes constructing a warning level decision according to the output of the adaptive prediction model, and calculating the warning level, and converting the prediction result into a specific warning level, such as normal, attention, warning and emergency, by quantifying the probability value; Then, according to the warning level and the actions taken at similar levels in history, the operation suggestions for the current equipment status are generated, and the warning and feasible operation decisions are obtained by combining the adaptive prediction model, which are expressed as follows: Action=f(Warning l evel,Historical a ctions) Where k is a coefficient used to adjust the warning sensitivity; P adaptive (t) is the adaptive algorithm prediction model; Action is the warning level decision; Historical a Actions are actions taken at similar levels historically.

7. The fault diagnosis and prediction method based on operation and maintenance scenario according to claim 6 is characterized in that: The prediction algorithm is automatically adjusted according to the specific operation and maintenance scenario and real-time data, and the health status of the equipment is updated in real time. The output of the algorithm is used to determine the warning information of the equipment and make corresponding operation suggestions according to the different warning levels. These warning information are classified according to the predefined risk level and guide the corresponding operation behavior: Low risk stage (P adaptive <first threshold); when the warning probability output by the adaptive prediction model is lower than the first threshold, a warning message is sent to relevant departments and the public that may be affected, and a low-intensity virtual failure scenario is displayed for risk awareness training. Maintain normal monitoring and routine maintenance procedures; Medium risk stage (first threshold ≤ P adaptive ≤ the second threshold); the warning probability output by the adaptive prediction model is between the first threshold and the second threshold; send enhanced warning information to relevant departments, form and display medium-intensity virtual failure scenarios for risk communication and emergency drills, activate emergency response plans, and prepare necessary evacuation and rescue resources. High risk stage (P adaptive > second threshold); When the warning probability output by the adaptive prediction model exceeds the second threshold; Issue emergency warnings to all relevant departments and the public. Generate and display high-intensity virtual failure scenarios for evacuation route planning and rescue strategy formulation, and immediately initiate evacuation procedures and emergency response measures.

8. A fault diagnosis and prediction system based on an operation and maintenance scenario, based on the fault diagnosis and prediction method based on an operation and maintenance scenario according to any one of claims 1 to 7, characterized in that: Also includes, The acquisition module collects equipment data, builds a scenario-based fault diagnosis model, and monitors the equipment operation status in real time through the scenario-based fault diagnosis model; The fusion module integrates the collected operation and maintenance data and the historical maintenance record data of the equipment to obtain real-time data fusion. The output of the scenario-based fault diagnosis model is used as the input of the real-time data fusion. The health index and the current equipment status and potential failure risk of the current equipment are obtained based on the real-time data fusion. The prediction module obtains an adaptive prediction algorithm based on the interaction between the scenario-based fault diagnosis model and the real-time data fusion, determines the warning information of the equipment according to the value output by the adaptive prediction algorithm, and corresponds to the scope of measures that the operator should take according to the degree of the warning information.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the fault diagnosis and prediction method based on the operation and maintenance scenario described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the fault diagnosis and prediction method based on the operation and maintenance scenario described in any one of claims 1 to 7 are implemented.

Images