Intelligent fault diagnosis method and system for stereo garage

Fault prediction and health status inference are carried out through dynamic Bayesian network and MDP model, combined with U-Net to identify damage characteristics, optimize maintenance plans, and solve the problem of lag in the fault response in the existing technology, and realize the accurate prediction of three-dimensional garage failures and early identification improves the scientificity and timeliness of maintenance decisions and ensures the stability and efficiency of garage operation.

CN119026022AInactive Publication Date: 2024-11-26BEIJING MASTEEL INTELLIGENT PARKING MANAGEMENT CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411514369.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-10-29
Publication Date
2024-11-26
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The existing technology cannot effectively integrate real-time monitoring data, resulting in lag in fault response, lack of instant updates and prediction models, and the inability to detect and prevent failures in the early stage, leading to small problems turning into large failures, affecting the stability of garage operation and increasing maintenance costs.

Method used

Dynamic Bayesian network and MDP model are used to predict potential faults and infer health status, combine U-Net to identify damage characteristics, generate maintenance optimization plans, optimize maintenance and maintenance time intervals, and achieve accurate prediction and real-time monitoring of three-dimensional garage failures.

Benefits of technology

Improves the prospective and responsiveness of fault diagnosis, avoids the high maintenance costs caused by sudden failures, ensures maximum efficiency and safety of garage operation, and improves prevention through optimized maintenance cycles and on-site verification of maintenance results. Maintenance effectiveness.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119026022A_ABST
    Figure CN119026022A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of fault diagnosis, in particular to an intelligent fault diagnosis method and system for a stereo garage, and the method comprises the following steps: carrying out the data integration and state evaluation based on collected garage operation data, determining the current working condition of the stereo garage, and calculating the conversion probability through employing a dynamic Bayesian network. According to the invention, through combination of the dynamic Bayesian network and the MDP model, accurate prediction and real-time health state monitoring of fault development of the stereo garage are realized, the perspectiveness and the response speed of fault diagnosis are improved, instant fault analysis is provided based on existing data, and through continuous state updating, the maintenance decision is more scientific and time-efficient, and the maintenance efficiency of the stereo garage is improved. The high maintenance cost caused by sudden faults is avoided, the maximum efficiency and safety of operation of the stereo garage are ensured by optimizing the maintenance period and verifying the maintenance effect on site, deep recognition of damage characteristics is conducted by introducing U-Net, and the early-stage discovery capacity of tiny damage is enhanced.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of fault diagnosis, and in particular to an intelligent fault diagnosis method and system for a stereo garage. Background Art

[0002] Fault diagnosis technology mainly focuses on the status monitoring and anomaly detection of machines or equipment during operation, as well as timely fault identification and root cause analysis, covering everything from basic fault detection methods to complex predictive maintenance strategies. With the development of information technology and artificial intelligence, fault diagnosis technology has gradually been integrated with technologies such as big data analysis, machine learning algorithms, sensor technology, and cloud computing to achieve more accurate and automated fault prediction and processing.

[0003] Among them, the intelligent fault diagnosis method of the multi-story parking garage is a technical method for fault detection and diagnosis of mechanized parking facilities such as multi-story parking garages. Due to their high degree of automation and complex mechanical structure, multi-story parking garages need to rely on accurate fault diagnosis to ensure operational efficiency and safety. Its uses include real-time monitoring of the operating status of the multi-story parking garage, automatic detection and diagnosis of potential faults, and providing maintenance suggestions.

[0004] Existing technologies are unable to effectively integrate real-time monitoring data, resulting in delayed responses to faults. The lack of instant updates and predictive models means that maintenance is mostly passive, making it impossible to detect and prevent faults at an early stage. Existing technologies are also limited in the early identification of small-scale damage, causing small problems to evolve into major faults, affecting the overall operational stability of the garage, increasing maintenance costs, and making it difficult for the maintenance team to develop efficient preventive maintenance plans, thus affecting the overall long-term health and operational efficiency of the garage. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose an intelligent fault diagnosis method and system for a stereoscopic car park.

[0006] In order to achieve the above object, the present invention adopts the following technical solution: a method for intelligent fault diagnosis of a stereo garage, comprising the following steps: S1: Based on the collected garage operation data, data integration and status evaluation are performed to determine the current working conditions of the stereo garage, and the dynamic Bayesian network is used to calculate the transition probability to obtain potential fault prediction information; S2: Based on the potential fault prediction information, the health status of the stereo garage is inferred online using the MDP model, the instant health status of the stereo garage is identified, the fault development trend is analyzed, and the fault development timeline is determined to generate a health status timeline; S3: Based on the health status timeline, analyze the maintenance cycle of the lifting mechanism, evaluate the impact on the operating efficiency of the stereo garage, optimize the maintenance and maintenance time intervals, verify the maintenance effect, and generate a maintenance optimization plan; S4: Based on the maintenance optimization plan, the garage status is analyzed from the monitoring video, and the damage characteristics of key components are identified using U-Net. The damage types are classified, the severity is determined, and the damage identification results are generated.

[0007] The present invention is improved in that the step of determining the current working condition of the stereo garage is specifically as follows: S111: Based on the collected garage operation data, clean the data to remove abnormal and duplicate records, and obtain a cleaned operation data set; S112: based on the cleaned operation data set, distinguishing normal and abnormal operation modes, identifying distribution characteristics of the modes, and generating an operation mode classification result; S113: Based on the operation mode classification result, the formula is used: Calculate current working status indicators , get the current working conditions of the garage, where Represents the total number of data points in normal operating mode, Represents the total number of data points in the abnormal mode, , and is the adjustment factor.

[0008] The present invention is improved in that the step of calculating the conversion probability is specifically as follows: S121: Based on the current working conditions, record the number of occurrences of each state, including the number of transitions from one state to another, to obtain a state transition frequency matrix; S122: Based on the state transition frequency matrix, the historical state transition data is analyzed using the Bayesian network to calculate the future state transition probability using the formula: Calculate from the state To status The conversion probability , and get the state transition probability result, where Represents the state Transition to state The number of observations, is the smoothing parameter, Yes Status Transition to any other state The number of observations.

[0009] The present invention is improved in that the step of identifying the immediate health status is specifically as follows: S211: Based on the potential fault prediction information, real-time operation data of the three-dimensional parking garage is collected, including lifting operation frequency, parking space utilization rate and start-stop times, and noise is filtered out and data deviation is corrected to obtain a processed real-time data set; S212: Based on the processed real-time data set, the health status is inferred using the MDP model, using the formula: Evaluate the garage from its current state To the next state The transition probability , and get the health status prediction result, where is in state Next Action plans, Indicates taking action Back from state Transfer to state The probability of Represents a collection of actions; S213: Based on the health status prediction result and the maintenance record, diagnose the immediate health status of the garage to obtain the immediate health status.

[0010] The present invention is improved in that the step of determining the fault development timeline is specifically as follows: S221: Based on the instant health status, identify the fault type and development speed, and analyze the fault frequency and severity to obtain fault trend analysis information; S222: extract key information from the fault trend analysis information, and calculate the timeline of the fault progressing to the critical state, using the formula: Get the fault development timeline, where Represents the time to reach the critical state, is the fault development rate, is the initial fault severity, is the remaining tolerance threshold.

[0011] The present invention is improved in that the analysis steps of the maintenance cycle of the lifting mechanism are specifically as follows: S311: Based on the health status timeline, collect maintenance and fault records of the garage lifting mechanism, including the fault occurrence date, maintenance activities and post-fault recovery time, to obtain a summarized event log; S312: Based on the summarized event log, analyze the time interval between the maintenance activity and the fault occurrence, using the formula: Get the optimal maintenance cycle information, where: Indicates the recommended maintenance cycle, is the time when the failure occurred, is the running time since the last maintenance, is the number of failures during the monitoring period.

[0012] The present invention is improved in that the steps of obtaining the maintenance optimization plan are specifically as follows: S321: Evaluate the impact of maintenance activities on garage operating efficiency, collect and analyze data on the relationship between maintenance time and downtime, and generate preliminary maintenance impact records; S322: Based on the preliminary maintenance impact record, optimize the maintenance and maintenance time intervals using the formula: Get a maintenance optimization plan, where For the new maintenance cycle, is the maintenance efficiency influence coefficient, Indicates the change in efficiency before and after maintenance, For the current efficiency, Target efficiency.

[0013] The present invention is improved in that the step of obtaining the damage identification result is specifically as follows: S411: Based on the maintenance optimization plan, U-Net is used to extract features according to the monitoring video data, key textures in the image are captured, and a preliminary damage feature mapping set is obtained; S412: Based on the preliminary damage feature mapping set, the feature mapping is clarified by binarization and morphological operations, the damage area is defined, and the accuracy of the damage contour is verified to obtain a refined damage feature mapping set; S413: Based on the refined damage feature map set, calculate the regional statistical characteristics within the feature map set, including the region size and morphological characteristics, using the formula: Optimize the accuracy of damage type classification and classify the damage features to obtain damage identification results, among which, represents the distance between damage features, are the components of the eigenvector, is a similarity score based on morphological features, is the weight coefficient used to adjust the influence of morphological features, is the characteristic dimension.

[0014] An intelligent fault diagnosis system for a stereo garage, the system comprising: The fault assessment and prediction module integrates data based on the collected garage operation data, evaluates the current working conditions of the garage, uses the state assessment results, calculates the transition probability using a dynamic Bayesian network, and generates potential fault prediction information; The health status analysis module uses the MDP model to perform online inference of the health status of the garage based on the potential fault prediction information, identifies the immediate health status, analyzes the fault development trend, determines the fault development timeline, and generates a health status timeline; The maintenance cycle optimization module analyzes the maintenance cycle of the garage lifting mechanism according to the health status timeline, evaluates the impact on the operation efficiency of the stereo garage, optimizes the maintenance and maintenance time intervals, verifies the maintenance effect, and generates a maintenance optimization plan; The damage detection module analyzes the garage status from the monitoring video based on the maintenance optimization plan, applies the U-Net model to identify the damage characteristics of key components, determines the severity of the damage, and generates damage identification results; The maintenance execution and feedback module classifies the damage type based on the damage identification result, formulates a corresponding maintenance plan, performs maintenance on the damaged location, verifies the effectiveness of the maintenance plan, and generates a maintenance effect verification result.

[0015] Compared with the prior art, the advantages and positive effects of the present invention are: In the present invention, by combining the dynamic Bayesian network and the MDP model, accurate prediction of the development of faults in the stereogram parking garage and real-time health status monitoring are achieved, the foresight and response speed of fault diagnosis are improved, not only instant fault analysis is provided based on existing data, but also maintenance decisions are made more scientific and timely through continuous status updates, the high maintenance costs caused by sudden failures are avoided, and the maximum efficiency and safety of the operation of the stereogram parking garage are ensured by optimizing the maintenance cycle and verifying the maintenance effect on the spot. By introducing U-Net for deep identification of damage characteristics, the early detection capability of minor damages is enhanced, and the effectiveness of preventive maintenance is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A flow chart of an intelligent fault diagnosis method for a stereo garage is proposed for the present invention; Figure 2 A flow chart for determining the current working conditions of the stereo garage in the present invention; Figure 3 A flow chart for calculating conversion probability in the present invention; Figure 4 This is a flow chart for identifying the instant health status in the present invention; Figure 5 A flow chart for determining a fault development timeline in the present invention; Figure 6It is a flow chart for analyzing the maintenance cycle of the lifting mechanism in the present invention; Figure 7 A flowchart for obtaining a maintenance optimization plan in the present invention; Figure 8 This is a flowchart for obtaining damage identification results in the present invention. DETAILED DESCRIPTION

[0017] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0018] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc., indicating positions or positional relationships, are based on the positions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, the meaning of "multiple" is two or more, unless otherwise clearly and specifically defined.

[0019] Example See also Figure 1 The present invention provides a technical solution: a method for intelligent fault diagnosis of a stereo garage, comprising the following steps: S1: Based on the collected garage operation data, data integration and status evaluation are performed to determine the current working conditions of the stereo garage, and the dynamic Bayesian network is used to calculate the transition probability to obtain potential fault prediction information; the garage operation data includes sensor data, maintenance records, monitoring data, alarm records, and vehicle entry and exit times.

[0020] S2: Based on the potential fault prediction information, the MDP model is used to perform online inference on the health status of the stereo garage, identify the immediate health status of the stereo garage, analyze the fault development trend, determine the fault development timeline, and generate the health status timeline; S3: Based on the health status timeline, analyze the maintenance cycle of the lifting mechanism, evaluate the impact on the operating efficiency of the stereo garage, optimize the maintenance and maintenance time intervals, verify the maintenance effect, and generate a maintenance optimization plan; S4: Based on the maintenance optimization plan, the garage status is analyzed from the surveillance video, and the U-Net is used to identify the damage characteristics of key components, classify the damage type, determine the severity, and generate damage identification results.

[0021] Potential fault prediction information includes failure probability, failure impact level, and predicted failure location. The health status timeline includes health score, expected maintenance time point, and trend change indicator. The maintenance optimization plan includes maintenance cycle adjustment results, key component inspection frequency, and maintenance cost estimation results. Damage identification results include damage type classification, damage severity level, and damage impact area.

[0022] See also Figure 2 , the specific steps for determining the current working conditions of the stereo garage are: S111: Based on the collected garage operation data, clean the data to remove abnormal and duplicate records, and obtain a cleaned operation data set; The collected data covers information such as vehicle entry and exit time, usage frequency, fault records, etc. Through data cleaning technology, abnormal data caused by system errors, duplicate records or non-standard operations are removed. This process ensures the purity and reliability of the data set. The operating data set obtained after cleaning not only reflects the daily operation status of the garage, but also provides high-quality basic data for subsequent data analysis. Through the operating data, we can better understand the real-time status of the garage operation.

[0023] S112: based on the cleaned operation data set, distinguishing normal and abnormal operation modes, identifying distribution characteristics of the modes, and generating operation mode classification results; The clustering algorithm can effectively distinguish the operating modes in the garage, including clearly distinguishing the normal operating vehicle flow from abnormal situations. Through statistical analysis, the distribution characteristics of various modes can be identified, which helps to understand the performance of the garage at different times and under different conditions. It also reveals operational problems and risks. The use of data can take preventive measures, optimize operational strategies, and generate operating mode classification results with clear classification.

[0024] S113: Based on the operation mode classification results, the formula is used: Calculate current working status indicators , get the current working conditions of the garage, where Represents the total number of data points in normal operating mode, Represents the total number of data points in the abnormal mode, , and is the adjustment coefficient, reflecting the weight of the difference pattern on the working conditions; Currently there are Normal operation data points and Abnormal data points, the adjustment coefficient is , , , the calculation process is as follows: The results show that the current working conditions have a high degree of stability in the normal operating mode, and the impact of abnormal conditions is very limited, indicating that the garage is operating in a good working condition. The numerical results show that the current working state is good and can effectively support daily operations.

[0025] See also Figure 3 , the steps to calculate the conversion probability are as follows: S121: Based on the current working conditions, record the number of occurrences of each state, including the number of transitions from one state to another, to obtain a state transition frequency matrix; According to the current working conditions, each vehicle status is recorded, such as the number of occurrences of idle, occupied, and under maintenance. The data is obtained through the monitoring of the garage management system. The system records each state change once. The records are used to analyze the state transition frequency. By counting the number of transitions from each state to another, the state transition matrix can be obtained. The matrix reflects the probability of transition from any specific state to another. In order to ensure the accuracy of the data, data verification and updating are performed regularly to keep the information up to date and accurate, and generate a state transition frequency matrix.

[0026] S122: Based on the state transition frequency matrix, the Bayesian network is used to analyze the historical state transition data and calculate the future state transition probability using the formula: Calculate from the state To status The conversion probability , and get the state transition probability result, where Represents the state Transition to state The number of observations, is a smoothing parameter used to deal with data sparsity problems. Yes Status Transition to any other state The number of observations; The following data were monitored: The number of observations from state i to state j The total number of times from state i to any state Smoothing parameter , substitute the numerical value to calculate: The result shows that the transition probability from state i to state j is 15.56%, which shows the possibility of state change under given conditions and helps to predict the change of garage state in the future.

[0027] See also Figure 4 , the specific steps for identifying the immediate health status are: S211: Based on the potential fault prediction information, real-time operation data of the stereo garage is collected, including lifting operation frequency, parking space utilization rate and start-stop times, and noise is filtered out and data deviation is corrected to obtain a processed real-time data set; Key operating information is obtained from the monitoring system of the multi-story parking garage, including elevator usage frequency, parking space occupancy rate, door operation times, etc., and noise and outliers are removed. The pre-processed data is organized into an accurate real-time data set, which not only provides data support for the immediate health status of the multi-story parking garage, but also provides a basis for further analysis. Through the data collection and processing process, it is ensured that the data set can truly reflect the operating status of the garage, providing an input source for subsequent health status analysis and fault prediction.

[0028] S212: Based on the processed real-time data set, the health status is inferred using the MDP model using the formula: Evaluate the garage from its current state To the next state The transition probability , and get the health status prediction result, where is in state Next Action plans, Indicates taking action Back from state Transfer to state The probability of Represents a collection of actions; Monitor the current status , the probability of garage failure , , then the next state The probability of (gate failure to continue) is calculated as follows: This means that at the end of the current monitoring cycle, the probability that the door will continue to fail is 1%, which shows that the current state has little impact on door failure.

[0029] S213: Based on the health status prediction result and the maintenance record, diagnose the immediate health status of the garage to obtain the immediate health status; Combining recent maintenance records and actual operation logs, in-depth health status diagnosis is carried out to determine the key to the garage's immediate health status. By comparing model predictions and actual maintenance data, the severity and actual impact of potential problems can be verified. The generated health status report describes all identified problems and their impact on garage operations, which not only provides real-time health status for the garage's daily operations, but also guides future maintenance and operation decisions.

[0030] See also Figure 5 , the specific steps for determining the fault development timeline are: S221: Based on the immediate health status, identify the fault type and development speed, analyze the fault frequency and severity, and obtain fault trend analysis information; Using real-time health status reports and historical fault data, through time series analysis and fault pattern recognition, the type of fault and its development speed are determined, including the frequency of occurrence of faults and the serious impact on operations. After rigorous analysis, the data is integrated into a comprehensive fault trend analysis report. The report describes the behavior pattern of each fault and predicts potential risks and the development path of the fault, providing a scientific basis for the next stage of fault timeline prediction and ensuring the depth and breadth of the analysis.

[0031] S222: Extract key information from the fault trend analysis information and calculate the timeline of the fault progressing to the critical state using the formula: Get the fault development timeline, where Represents the time to reach the critical state, is the fault development rate, is the initial fault severity, is the residual tolerance threshold; Monitor the current fault development rate Days, initial fault severity , the remaining tolerance threshold , substitute the numerical value to calculate: The results show that if the current trend continues, failures will reach a critical state in about 14 days, allowing maintenance planning to be carried out in a more scientific and orderly manner.

[0032] See also Figure 6 , the analysis steps of the maintenance cycle of the lifting mechanism are as follows: S311: Based on the health status timeline, the maintenance and fault records of the garage lifting mechanism are collected, including the fault occurrence date, maintenance activities and post-fault recovery time, to obtain a summarized event log; Collect historical maintenance and fault records of the lifting mechanism based on the health status timeline, including the specific date and time of the fault, the type of maintenance activity, and the details of the fault recovery. By summarizing the information, create an event log to provide necessary data support for subsequent maintenance cycle analysis.

[0033] S312: Based on the summarized event log, analyze the time interval between maintenance activities and failure occurrence, using the formula: Get the optimal maintenance cycle information, where: Indicates the recommended maintenance cycle, is the time when the failure occurred, is the running time since the last maintenance, is the number of failures during the monitoring period; In the past year, the operation time after maintenance was 30 days, 45 days, 40 days, 35 days and 50 days respectively, and the failure occurred on the 31st, 46th, 41st, 36th and 51st day of operation respectively. The calculation is as follows: This suggests that maintenance every 40 days would be the ideal cycle.

[0034] See also Figure 7 , the specific steps for obtaining the maintenance optimization plan are: S321: Evaluate the impact of maintenance activities on garage operating efficiency, collect and analyze data on the relationship between maintenance time and downtime, and generate preliminary maintenance impact records; Based on the health status timeline and historical maintenance data, the data set is collated and analyzed, and the factors affecting the operating efficiency are quantified through statistical methods, such as the correlation analysis between maintenance frequency and system downtime, to build an impact assessment model. This model provides a preliminary scientific basis for the adjustment of the garage maintenance strategy based on actual data, ensures the practicality and effectiveness of the optimization, and generates a preliminary maintenance impact record.

[0035] S322: Based on the preliminary maintenance impact record, optimize the maintenance and maintenance intervals using the formula: Get a maintenance optimization plan, where For the new maintenance cycle, is the maintenance efficiency impact coefficient, which represents the average impact of maintenance activities on efficiency. Indicates the change in efficiency before and after maintenance, For the current efficiency, To target efficiency; The following data was collected: is 0.95; is 0.85; is 0.03; is 0.10.

[0036] Substituting the values ​​into: The results show that the maintenance cycle is about 12 days, and efficiency can be improved while maintaining a lower maintenance frequency, achieving the dual goals of saving costs and improving efficiency.

[0037] See also Figure 8 , the specific steps for obtaining damage identification results are: S411: Based on the maintenance optimization plan, U-Net is used to extract features based on the surveillance video data to capture key textures in the image and obtain a preliminary damage feature map set; A preliminary image of the damaged area is obtained through monitoring video, and the U-Net network is used for feature extraction. The image is first normalized to unify the image format and size, and then the color space conversion is performed to enhance the contrast of the damaged area. Through the preprocessing steps, the damage features in the image can be more accurately identified. The captured features are then converted into a feature mapping set, which provides a basis for further identification and classification of damage types and ensures the accuracy and efficiency of feature extraction.

[0038] S412: Based on the preliminary damage feature mapping set, the feature mapping is clarified through binarization and morphological operations, the damage area is defined, and the accuracy of the damage contour is verified to obtain a refined damage feature mapping set; The damaged area in the image is distinguished from the non-damaged area through binarization processing. The selection of an appropriate threshold is the key. The selection of the threshold is based on the grayscale histogram analysis of the image. After determining the optimal threshold, morphological operations such as dilation and erosion are applied to help clarify the damage edge and make the damage contour more obvious. At the same time, small noise points are removed and small cracks in the contour are filled to ensure the continuity and integrity of the damaged area. This not only improves the recognition accuracy of the damaged area, but also provides a basis for judging the degree of damage. Through refinement processing, a clearly defined damage feature mapping set is generated.

[0039] S413: Based on the refined damage feature map set, calculate the regional statistical characteristics within the feature map set, including the region size and morphological characteristics, using the formula: Optimize the accuracy of damage type classification and classify the damage features to obtain damage identification results, among which, represents the distance between damage features, are the components of the eigenvector, is a similarity score based on morphological features, is the weight coefficient used to adjust the influence of morphological features, is the characteristic dimension; The following data was collected: ; ; ; ; .

[0040] Compute the squared difference of the Euclidean distance between two vectors: Add a weighted value to the morphological similarity score: Taking the square root gives the distance: The results show that combining geometric and morphological features of damage can more accurately identify and classify damage in garages, supporting accurate assessment of damage severity.

[0041] An intelligent fault diagnosis system for a stereo garage, the system comprising: The fault assessment and prediction module integrates data based on the collected garage operation data, evaluates the current working conditions of the garage, uses the state assessment results, calculates the transition probability using a dynamic Bayesian network, and generates potential fault prediction information; The health status analysis module uses the MDP model to perform online inference of the garage health status based on potential fault prediction information, identify the immediate health status, analyze the fault development trend, determine the fault development timeline, and generate the health status timeline; The maintenance cycle optimization module analyzes the maintenance cycle of the garage lifting mechanism based on the health status timeline, evaluates the impact on the operating efficiency of the stereo garage, optimizes the maintenance and maintenance time intervals, verifies the maintenance effect, and generates a maintenance optimization plan; The damage detection module analyzes the garage status from the surveillance video based on the maintenance optimization plan, applies the U-Net model to identify the damage characteristics of key components, determines the severity of the damage, and generates damage identification results; The maintenance execution and feedback module classifies the damage types based on the damage identification results, formulates corresponding maintenance plans, performs maintenance on the damaged locations, verifies the effectiveness of the maintenance plans, and generates maintenance effect verification results.

[0042] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A method for intelligent fault diagnosis of a stereo garage, characterized in that: The following steps are involved: Based on the collected garage operation data, data integration and status evaluation are performed to determine the current working conditions of the stereo garage, and the dynamic Bayesian network is used to calculate the conversion probability to obtain potential fault prediction information; Based on the potential fault prediction information, the MDP model is used to perform online inference on the health status of the stereo garage, identify the immediate health status of the stereo garage, analyze the fault development trend, determine the fault development timeline, and generate a health status timeline; Based on the health status timeline, analyze the maintenance cycle of the lifting mechanism, evaluate the impact on the operating efficiency of the stereo garage, optimize the maintenance and maintenance time intervals, verify the maintenance effect, and generate a maintenance optimization plan; Based on the maintenance optimization plan, the garage status is analyzed from the surveillance video, and the U-Net is used to identify the damage characteristics of key components. The damage type is classified, the severity is determined, and the damage identification results are generated.

2. The intelligent fault diagnosis method for a stereo garage according to claim 1 is characterized in that: The steps for determining the current working conditions of the stereo garage are specifically as follows: Based on the collected garage operation data, the data is cleaned to remove abnormal and duplicate records, and a cleaned operation data set is obtained; Based on the cleaned operation data set, distinguish normal and abnormal operation modes, identify distribution characteristics of the modes, and generate operation mode classification results; Based on the operation mode classification results, the formula is adopted: Calculate current working status indicators , get the current working conditions of the garage, where Represents the total number of data points in normal operating mode, Represents the total number of data points in the abnormal mode, , and is the adjustment factor.

3. The intelligent fault diagnosis method for a stereo garage according to claim 1 is characterized in that: The step of calculating the conversion probability is specifically as follows: Based on the current working conditions, the number of occurrences of each state is recorded, including the number of transitions from one state to another, to obtain a state transition frequency matrix; Based on the state transition frequency matrix, the Bayesian network is used to analyze the historical state transition data and calculate the future state transition probability using the formula: Calculate from the state To status The conversion probability , and get the state transition probability result, where Represents the state Transition to state The number of observations, is the smoothing parameter, Yes Status Transition to any other state The number of observations.

4. The intelligent fault diagnosis method for a stereo garage according to claim 1 is characterized in that: The steps for identifying the instant health status are specifically as follows: Based on the potential fault prediction information, real-time operation data of the stereo garage is collected, including lifting operation frequency, parking space utilization rate and start-stop times, and noise is filtered out and data deviation is corrected to obtain a processed real-time data set; Based on the processed real-time data set, the health status is inferred using the MDP model, using the formula: Evaluate the garage from its current state To the next state The transition probability , and get the health status prediction result, where is in state Next Action plans, Indicates taking action Back from state Transfer to state The probability of Represents a collection of actions; Based on the health status prediction result and the maintenance record, the immediate health status of the garage is diagnosed to obtain the immediate health status.

5. The intelligent fault diagnosis method for a stereo garage according to claim 1 is characterized in that: The steps for determining the fault development timeline are specifically as follows: Based on the instant health status, identify the fault type and development speed, and analyze the frequency and severity of the fault to obtain fault trend analysis information; Extract key information from the fault trend analysis information and calculate the timeline of the fault progressing to the critical state using the formula: Get the fault development timeline, where Represents the time to reach the critical state, is the fault development rate, is the initial fault severity, is the remaining tolerance threshold.

6. The intelligent fault diagnosis method for a stereo garage according to claim 1 is characterized in that: The analysis steps of the maintenance cycle of the lifting mechanism are specifically as follows: Based on the health status timeline, the maintenance and fault records of the garage lifting mechanism are collected, including the fault occurrence date, maintenance activities and post-fault recovery time, to obtain a summarized event log; Based on the summarized event log, the time interval between maintenance activities and failure occurrence is analyzed using the formula: Get the optimal maintenance cycle information, where: Indicates the recommended maintenance cycle, is the time when the failure occurred, is the running time since the last maintenance, is the number of failures during the monitoring period.

7. The intelligent fault diagnosis method for a stereo garage according to claim 1 is characterized in that: The steps for obtaining the maintenance optimization plan are specifically as follows: Evaluate the impact of maintenance activities on garage operating efficiency, collect and analyze maintenance time versus downtime data, and generate preliminary maintenance impact records; Based on the preliminary maintenance impact record, optimize the service and maintenance intervals using the formula: Get a maintenance optimization plan, where For the new maintenance cycle, is the maintenance efficiency influence coefficient, Indicates the change in efficiency before and after maintenance, For the current efficiency, Target efficiency.

8. The intelligent fault diagnosis method for a stereo garage according to claim 1 is characterized in that: The steps for obtaining the damage identification result are specifically as follows: Based on the maintenance optimization plan, U-Net is used to extract features based on the surveillance video data to capture key textures in the image and obtain a preliminary damage feature mapping set; Based on the preliminary damage feature map set, the feature map is clarified by binarization and morphological operations, the damage area is defined, and the accuracy of the damage contour is verified to obtain a refined damage feature map set; Based on the refined damage feature map set, the regional statistical characteristics within the feature map set are calculated, including the region size and morphological characteristics, using the formula: Optimize the accuracy of damage type classification and classify the damage features to obtain damage identification results, among which, represents the distance between damage features, are the components of the eigenvector, is a similarity score based on morphological features, is the weight coefficient used to adjust the influence of morphological features, is the characteristic dimension.

9. An intelligent fault diagnosis system for a stereo garage, characterized in that: According to any one of claims 1 to 8, the intelligent fault diagnosis method for a stereo garage is implemented, and the system comprises: The fault assessment and prediction module integrates data based on the collected garage operation data, evaluates the current working conditions of the garage, uses the state assessment results, calculates the transition probability using a dynamic Bayesian network, and generates potential fault prediction information; The health status analysis module uses the MDP model to perform online inference of the health status of the garage based on the potential fault prediction information, identifies the immediate health status, analyzes the fault development trend, determines the fault development timeline, and generates a health status timeline; The maintenance cycle optimization module analyzes the maintenance cycle of the garage lifting mechanism according to the health status timeline, evaluates the impact on the operation efficiency of the stereo garage, optimizes the maintenance and maintenance time intervals, verifies the maintenance effect, and generates a maintenance optimization plan; The damage detection module analyzes the garage status from the monitoring video based on the maintenance optimization plan, applies the U-Net model to identify the damage characteristics of key components, determines the severity of the damage, and generates damage identification results; The maintenance execution and feedback module classifies the damage type based on the damage identification result, formulates a corresponding maintenance plan, performs maintenance on the damaged location, verifies the effectiveness of the maintenance plan, and generates a maintenance effect verification result.