Digital twin driven car washing equipment failure prediction and cloud maintenance scheduling method

Abstract

The application discloses a digital twin driven car washing equipment prediction fault and cloud maintenance scheduling method, relates to the technical field of equipment maintenance management, and comprises the following steps: obtaining a digital twin model of a target car washing equipment, obtaining a currently arranged maintenance task sequence as a maintenance scheduling link based on a normal state interval; mapping real-time collected equipment operation data to each hierarchical node of the digital twin model, performing layer-by-layer state verification, marking a fault risk point and determining a corresponding risk level, traversing idle time periods between adjacent maintenance tasks in the maintenance scheduling link based on a maintenance task estimated duration and the risk level, screening idle gaps that can accommodate maintenance tasks and performing maintenance task insertion, and determining a maintenance scheduling scheme and outputting based on the maintenance scheduling link after the insertion is completed. When the overall level is abnormal, the module level verification is triggered, and when the module level is abnormal, the component level verification is triggered, so that the fault position is accurately positioned layer by layer.

Description

Technical Field

[0001] This invention relates to the field of equipment maintenance management technology, specifically a digital twin-driven method for predicting faults and scheduling cloud-based maintenance for car wash equipment. Background Technology

[0002] As a frequently used automated operation equipment, the reliability of car wash equipment directly affects service efficiency and user experience. Car wash equipment typically includes multiple functional modules such as high-pressure spraying, foam coating, roller brush cleaning, and air drying. Each module has numerous actuators, resulting in a complex equipment structure and a variety of failure types.

[0003] Currently, the most common equipment maintenance methods are regular inspections and post-incident repairs. Regular inspections check and maintain the equipment according to a fixed cycle, but cannot detect sudden performance degradation or early signs of failure during the inspection interval. Post-incident repairs are carried out only after the equipment has obvious failures or has stopped. At this time, the failure has often caused actual damage to the equipment, resulting in high repair costs, long downtime, and possible disruption to ongoing car wash operations.

[0004] With the development of digital twin technology, some equipment manufacturers have attempted to build digital twin models of their devices. These models collect real-time operational data from sensors and map it onto the digital twin, enabling visualized monitoring of the equipment's operating status. However, most existing digital twin applications remain at the level of status display. When monitoring data shows anomalies, it is still necessary to manually determine the specific location and severity of the fault, and then manually find available time slots in the maintenance schedule to arrange maintenance tasks. This manual approach to fault location relies on personnel experience and makes it difficult to trace the faulty component layer by layer from the overall equipment status. Summary of the Invention

[0005] (a) Technical problems to be solved To address the shortcomings of existing technologies, this invention provides a digital twin-driven method for predicting faults and scheduling cloud-based maintenance for car wash equipment.

[0006] (II) Technical Solution To achieve the above objectives, the present invention provides the following technical solution: a digital twin-driven method for predicting faults and scheduling cloud-based maintenance of car wash equipment, specifically including: S1. Obtain the digital twin model of the target car wash equipment, generate the normal state interval of each node based on the historical normal operation parameters of each node in the digital twin model, and obtain the currently arranged maintenance task sequence as the maintenance scheduling link. S2. Map the real-time collected equipment operation data to the nodes of each level of the digital twin model, perform layer-by-layer status verification in the order from the first level to the third level, mark the fault risk points and determine the corresponding risk level. S3. Based on the estimated duration and risk level of the maintenance task corresponding to the fault risk point, traverse the idle time periods between adjacent maintenance tasks in the maintenance scheduling link, filter out the idle gaps that can accommodate maintenance tasks, and insert maintenance tasks. S4. Based on the maintenance scheduling link after insertion, determine the maintenance scheduling scheme and output it.

[0007] Preferably, in step S1, a digital twin model of the target car wash equipment is obtained. The first level of the digital twin model includes an overall node representing the overall operating status of the equipment, the second level includes module nodes representing the operating status of each functional module of the equipment, and the third level includes component nodes representing the operating status of each execution component of the equipment. Each functional module node has a subordinate relationship with its overall node, and each execution component node has a subordinate relationship with its functional module node. For each node in each level, extract the sequence of operating parameters corresponding to the historical normal operation period of each node from the digital twin model. The operating parameters include at least one of the vibration amplitude, operating current value and operating temperature value of the corresponding equipment unit of each node during operation. Calculate the arithmetic mean and standard deviation of each operating parameter value in the operating parameter sequence. Use the arithmetic mean minus the first multiple standard deviation as the lower limit of the normal state interval, and use the arithmetic mean plus the first multiple standard deviation as the upper limit of the normal state interval. All currently scheduled maintenance tasks are retrieved from the equipment maintenance management system. Each maintenance task includes the task execution period, task duration, and task priority. All maintenance tasks are arranged in chronological order of their execution periods. The time periods between adjacent tasks and between a task and the current time constitute idle periods, forming a maintenance scheduling link.

[0008] Preferably, in step S2, the operating data is collected in real time by sensors deployed on each execution component, each functional module and the whole machine of the target car wash equipment according to the collection cycle, and the operating data collected by each sensor is mapped to the corresponding whole node, module node and component node according to the correspondence between the sensor and each level node in the digital twin model. For each global node in the first level, the real-time running data mapped to the global node is compared with the normal state range corresponding to the global node. If the real-time running data is within the normal state range, the global node is determined to be in a normal state. If the real-time running data exceeds the normal state range, the global node is determined to be in an abnormal state. When all nodes in the first level are in normal condition, the equipment is determined to be without fault risk, and the current round of layer-by-layer status verification ends. When an entire node in the first level is in an abnormal state, the status verification of all second-level module nodes to which the abnormal overall node belongs is triggered. Triggering a status check on all second-level module nodes to which the abnormal overall node belongs includes the following steps: Read the node identifier of the overall node in an abnormal state, and search for all second-level module nodes in the digital twin model that have a subordinate relationship with the abnormal overall node; For each second-level module node found, the real-time running data mapped to that module node is compared with the normal state range corresponding to that module node; When the real-time running data of all the verified second-level module nodes are within their respective normal state ranges, the anomaly of the first level is determined to be caused by instantaneous data fluctuations or sensor noise. No fault risk points are marked, and the current round of layer-by-layer state verification ends. When the real-time running data of a second-level module node exceeds its normal state range, the module node is marked as an abnormal module, triggering a state check on all third-level component nodes to which the abnormal module belongs. Triggering a status check on all third-level component nodes to which the abnormal module belongs includes the following steps: Read the node identifier of the module node marked as an abnormal module, and search for all third-level component nodes in the digital twin model that have a subordinate relationship with the abnormal module. For each third-level component node found, the real-time running data mapped to that component node is compared with the normal state range corresponding to that component node; Mark the third-level component nodes whose real-time operating data exceeds their normal range as fault risk points; For each fault risk point, the risk level is determined based on the ratio of the deviation of the real-time operating data of the component node from the normal state range to the standard deviation of the corresponding component node in the normal state range. The larger the ratio of the deviation to the standard deviation, the higher the risk level.

[0009] Preferably, in step S3, based on the component node identifier corresponding to the fault risk point, the maintenance procedure library corresponding to the component node is read from the digital twin model. The maintenance procedure library records the standard maintenance procedures for different fault types of the component and the estimated operation time of each procedure. The estimated operation time of each procedure is summarized to obtain the estimated time of the maintenance task. The priority of maintenance tasks is determined based on the risk level; higher risk levels correspond to higher priority maintenance tasks. Read each idle time slot in the maintenance scheduling link in chronological order, compare the duration of the idle time slot with the estimated duration of the maintenance task, and mark the idle time slot as a candidate idle gap if the duration of the idle time slot is greater than or equal to the estimated duration of the maintenance task. Arrange all candidate idle gaps in chronological order to form a candidate gap set. Select the first available gap from the candidate gap set, insert the maintenance task into the available gap, and set the start time of the maintenance task to the start time of the available gap. The end time of the maintenance task is the start time plus the estimated duration of the maintenance task. The execution time of all maintenance tasks located after the insertion position in the maintenance scheduling link is shifted backward as a whole. The shift time is the difference between the estimated duration of the maintenance task and the original idle duration of the candidate idle gap. If the difference is less than or equal to zero, there is no need to shift the subsequent maintenance tasks. Check if there is any overlap in the occurrence time of any two adjacent maintenance tasks in the maintenance scheduling link after the shift. If there is no overlap, it is determined that there is no timing conflict and the maintenance task insertion is completed. If there is a time overlap, the current candidate empty gap is removed from the candidate gap set, and the next candidate empty gap is selected for insertion verification until an insertion position without time conflict is found or the candidate gap set is empty. When the candidate gap set is empty, that is, there is no idle time period in the maintenance scheduling link that can accommodate the estimated duration of the maintenance task, forced insertion is performed according to the priority of the maintenance task: if the priority of the maintenance task is the highest priority, the maintenance task is inserted into the current position of the maintenance scheduling link, and all scheduled maintenance tasks after the current time are shifted backward as a whole, with the shifting time being the estimated duration of the maintenance task. If the maintenance task has a non-highest priority, add the maintenance task to the end of the maintenance scheduling chain, after all other scheduled maintenance tasks.

[0010] Preferably, in step S4, the maintenance scheduling link that has been inserted and has no timing conflict is used as the maintenance scheduling scheme. The maintenance scheduling scheme includes the task identifier, execution component name, task start time, task end time, task priority, and task execution content description for each maintenance task. The maintenance scheduling plan is sent to the mobile terminal devices of maintenance personnel through the communication network. At the same time, the fault risk point information marked in this round of layer-by-layer status verification is updated to the status attribute field of the corresponding component node in the digital twin model, so as to continuously track the status of the node in the next round of status verification.

[0011] (III) Beneficial Effects This invention provides a digital twin-driven method for predicting faults and scheduling cloud-based maintenance for car wash equipment, which has the following advantages: This invention divides the device into three levels: the whole, functional modules, and execution components. When an anomaly occurs at the whole level, a module-level verification is triggered. When an anomaly occurs at the module level, a component-level verification is triggered. This enables precise, layer-by-layer fault location and narrows the scope of fault investigation.

[0012] After determining the fault risk points and risk levels, the system automatically generates corresponding maintenance tasks and iterates through the idle time slots in the maintenance scheduling link. It then matches an idle gap that can accommodate the duration of the maintenance task for insertion. After insertion, it automatically checks for timing conflicts. If there are no conflicts, the scheduling is completed. If there are conflicts, it automatically switches to the next candidate gap. This achieves an automated closed-loop connection from fault prediction to maintenance scheduling. If there are no idle time slots in the maintenance scheduling link, it performs forced insertion or end-point appending based on the priority of the maintenance task, prioritizing the timely handling of high-risk faults while preventing low-risk tasks from crowding out emergency maintenance resources. Attached Figure Description

[0013] Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is a flowchart of the forced insertion logic of the present invention. Detailed Implementation

[0014] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0015] Please see Figure 1 This invention provides a digital twin-driven method for predicting faults and scheduling cloud-based maintenance for car wash equipment, comprising the following steps: S1. Obtain the digital twin model of the target car wash equipment, generate the normal state interval of each node based on the historical normal operation parameters of each node in the digital twin model, and obtain the currently arranged maintenance task sequence as the maintenance scheduling link. Furthermore, in S1, a digital twin model of the target car wash equipment is obtained. The first level of the digital twin model includes an overall node representing the overall operating status of the equipment, the second level includes module nodes representing the operating status of each functional module of the equipment, and the third level includes component nodes representing the operating status of each execution component of the equipment. Each functional module node has a subordinate relationship with its overall node, and each execution component node has a subordinate relationship with its functional module node. For each node in each level, extract the sequence of operating parameters corresponding to the historical normal operation period of each node from the digital twin model. The operating parameters include at least one of the vibration amplitude, operating current value and operating temperature value of the corresponding equipment unit of each node during operation. Calculate the arithmetic mean and standard deviation of each operating parameter value in the operating parameter sequence. Use the arithmetic mean minus the first multiple standard deviation as the lower limit of the normal state interval, and use the arithmetic mean plus the first multiple standard deviation as the upper limit of the normal state interval. All currently scheduled maintenance tasks are retrieved from the equipment maintenance management system. Each maintenance task includes the task execution period, task duration, and task priority. All maintenance tasks are arranged in chronological order of their execution periods. The time periods between adjacent tasks and between a task and the current time constitute idle periods, forming a maintenance scheduling link.

[0016] It should be noted that the first level contains an overall node, which represents the overall operating status of the entire device. The second level contains multiple module nodes, each corresponding to a functional module of the device, and each module node is subordinate to the overall node in the first level.

[0017] The third level contains multiple component nodes. Each component node corresponds to a specific execution component under a functional module. Each component node establishes a subordinate association with its corresponding second-level module node.

[0018] The digital twin model stores a complete hierarchical relationship table, which records the parent-child correspondence between the overall node identifier and the node identifiers of each module, as well as the parent-child correspondence between the node identifiers of each module and the node identifiers of each component.

[0019] For each node in each level, extract the sequence of operating parameters of that node during its historical normal operation period from the digital twin model.

[0020] Historical normal operating period refers to the continuous operating period from the time the equipment was last maintained until the time of the current maintenance until the current time, during which no fault alarms were confirmed by the equipment operation log.

[0021] The operating parameters include at least one of the vibration amplitude, operating current, and operating temperature values ​​of the corresponding equipment unit during operation.

[0022] Taking a component node in the third level as an example, this component node corresponds to a drive motor. The vibration amplitude sequence of the drive motor recorded by the vibration sensor according to the acquisition cycle during the historical normal operation period is extracted from the digital twin model.

[0023] Calculate the sum of all vibration amplitude samples and divide by the number of samples to obtain the arithmetic mean of the vibration amplitude. Calculate the square of the difference between each vibration amplitude sample and the arithmetic mean. Sum the squares of all differences and divide by the number of samples. Then take the square root of the quotient to obtain the standard deviation of the vibration amplitude.

[0024] The arithmetic mean of the vibration amplitude minus twice the standard deviation is used as the lower limit of the normal range of the vibration amplitude of the drive motor, and the arithmetic mean of the vibration amplitude plus twice the standard deviation is used as the upper limit of the normal range of the vibration amplitude of the drive motor. The interval between the lower limit and the upper limit constitutes the normal range of the vibration amplitude of the drive motor.

[0025] In the same manner, extract the sequence of operating current values ​​of the drive motor recorded by the current transformer during the same historical normal operation period, calculate the arithmetic mean and standard deviation of the operating current values, and generate the normal state range of the operating current values.

[0026] Extract the sequence of operating temperature values ​​recorded by the temperature sensor, calculate the arithmetic mean and standard deviation of the operating temperature values, and generate the normal operating temperature range. The normal operating temperature range of the drive motor is composed of the normal operating temperature range, the normal operating current range, and the normal operating temperature range.

[0027] For each type of operating parameter of all nodes at all levels in the digital twin model, the corresponding normal state interval is generated one by one in the above manner, and the normal state interval of each node is stored in the attribute field of the corresponding node in the digital twin model.

[0028] The current equipment maintenance management system has arranged multiple maintenance tasks. The first maintenance task has a duration of 1 and a priority of general priority. The second maintenance task has a duration of 2 and a priority of general priority. The third maintenance task has a duration of 3 and a priority of medium priority.

[0029] The three scheduled maintenance tasks are arranged in chronological order of their start times. The idle time periods are defined between adjacent tasks and between the current time and the start time of the first task.

[0030] The first idle period is between the current time and the start time of the first task, and its duration is determined by the difference between the start time of the first task and the current time. The second idle period is between the end time of the first task and the start time of the second task, and its duration is determined by the difference between the end time of the first task and the start time of the second task. The third idle period is between the end time of the second task and the start time of the third task. The fourth idle period is after the end time of the third task. All idle periods and scheduled maintenance tasks are arranged alternately in chronological order to form a maintenance scheduling link.

[0031] It should be noted that the first multiplier is a multiplier factor used to determine the boundary range of the interval during the generation of the normal state interval. The product of the first multiplier and the standard deviation determines the width of the normal state interval from the arithmetic mean to both sides. The first multiplier is calculated by extracting the operating parameter sequence of the node during the historical normal operation period from the device's digital twin model, and calculating the skewness and kurtosis of the operating parameter sequence. The skewness reflects the symmetry of the distribution of the operating parameter sequence data, and the kurtosis reflects the degree of concentration or dispersion of the operating parameter sequence data distribution relative to the normal distribution. When the absolute value of the skewness is less than the skewness judgment threshold and the kurtosis is less than the kurtosis judgment threshold, it indicates that the data distribution of the operating parameter sequence approximately follows a normal distribution. At this time, the first multiplier is set to two, so that the normal state interval statistically covers about 95% of the normal operation data points.

[0032] When the absolute value of the skewness value is greater than or equal to the skewness threshold, or the kurtosis value is greater than or equal to the kurtosis threshold, it indicates that the data distribution of the operating parameter sequence has skewed or heavy-tailed characteristics and does not meet the normal distribution assumption. In this case, the adjustment method for the first multiple is as follows: arrange all data values ​​in the operating parameter sequence from smallest to largest, take the data value located at the second quantile after arrangement as the lower quantile value, take the data value located at the ninety-eighth quantile after arrangement as the upper quantile value, take the difference between the lower quantile value and the upper quantile value and the arithmetic mean respectively, take the absolute value of the two differences and divide it by the standard deviation to obtain the lower quantile multiple and the upper quantile multiple, and take the smaller value of the lower quantile multiple and the upper quantile multiple as the first multiple.

[0033] The skewness and kurtosis thresholds are determined as follows: take the historical normal operation parameter sequences of all nodes at the same level in the digital twin model, calculate the skewness and kurtosis values ​​of each node parameter sequence, and take the median of all skewness values ​​as the skewness threshold and the median of all kurtosis values ​​as the kurtosis threshold.

[0034] When a node has multiple types of operating parameters, the above distribution test and first multiple determination process are performed for each type of operating parameter. Each type of operating parameter uses its own independently determined first multiple to generate the corresponding normal state interval. When the number of operating parameter sequence samples in the historical normal operation period is insufficient to support the statistical calculation of skewness and kurtosis, the first multiple is directly taken as two.

[0035] S2. Map the real-time collected equipment operation data to the nodes of each level of the digital twin model, perform layer-by-layer status verification in the order from the first level to the third level, mark the fault risk points and determine the corresponding risk level. Furthermore, in S2, the sensors deployed on each execution component, each functional module and the whole machine of the target car wash equipment collect operating data in real time according to the collection cycle, and the operating data collected by each sensor is mapped to the corresponding whole node, module node and component node according to the correspondence between the sensor and each level node in the digital twin model. For each global node in the first level, the real-time running data mapped to the global node is compared with the normal state range corresponding to the global node. If the real-time running data is within the normal state range, the global node is determined to be in a normal state. If the real-time running data exceeds the normal state range, the global node is determined to be in an abnormal state. When all nodes in the first level are in normal condition, the equipment is determined to be without fault risk, and the current round of layer-by-layer status verification ends. When an entire node in the first level is in an abnormal state, the status verification of all second-level module nodes to which the abnormal overall node belongs is triggered. Triggering a status check on all second-level module nodes to which the abnormal overall node belongs includes the following steps: Read the node identifier of the overall node in an abnormal state, and search for all second-level module nodes in the digital twin model that have a subordinate relationship with the abnormal overall node; For each second-level module node found, the real-time running data mapped to that module node is compared with the normal state range corresponding to that module node; When the real-time running data of all the verified second-level module nodes are within their respective normal state ranges, the anomaly of the first level is determined to be caused by instantaneous data fluctuations or sensor noise. No fault risk points are marked, and the current round of layer-by-layer state verification ends. When the real-time running data of a second-level module node exceeds its normal state range, the module node is marked as an abnormal module, triggering a state check on all third-level component nodes to which the abnormal module belongs. Triggering a status check on all third-level component nodes to which the abnormal module belongs includes the following steps: Read the node identifier of the module node marked as an abnormal module, and search for all third-level component nodes in the digital twin model that have a subordinate relationship with the abnormal module. For each third-level component node found, the real-time running data mapped to that component node is compared with the normal state range corresponding to that component node; Mark the third-level component nodes whose real-time operating data exceeds their normal range as fault risk points; For each fault risk point, the risk level is determined based on the ratio of the deviation of the real-time operating data of the component node from the normal state range to the standard deviation of the corresponding component node in the normal state range. The larger the ratio of the deviation to the standard deviation, the higher the risk level.

[0036] It should be noted that the operating data is collected in real time by sensors deployed on various execution components, functional modules and the whole machine according to the collection cycle. In each collection cycle, each sensor packages the operating data collected at the current moment together with the sensor identifier and timestamp into a data record and sends it to the data acquisition gateway. The data acquisition gateway summarizes the data records of each sensor and transmits them to the control system.

[0037] The control system maps each sensor data record to the corresponding overall node, module node, and component node according to the correspondence table between sensor identifiers and nodes at each level in the digital twin model, and updates the real-time operation data fields of each node in the digital twin model.

[0038] When the current data collection cycle arrives, the layer-by-layer status verification process is initiated. First, the status verification is performed on the first-level overall node. The calculation method for the real-time comprehensive operating status value of the overall node is as follows: read the real-time operating data of all module nodes of the second level at the current moment from the digital twin model, sum the real-time operating data of each module node according to the preset weight, and compare the calculated real-time comprehensive operating status value with the normal state interval corresponding to the overall node.

[0039] If the real-time comprehensive operating status value is greater than or equal to the lower limit of the normal status range of the entire node and less than or equal to the upper limit, the entire node is determined to be in a normal state. At this time, the device is determined to have no fault risk, the result of this round of verification is recorded as normal, the current layer-by-layer status verification process ends, and the verification is restarted in the next collection cycle.

[0040] If the real-time integrated operating status value is less than the lower limit of the normal state range of the entire node, or greater than the upper limit, the entire node is determined to be in an abnormal state. The control system records the determination result of the abnormality of the entire node and the time when the abnormality occurred, and triggers the status verification of all module nodes in the second level.

[0041] The control system reads the subordinate relationship table in the digital twin model, obtains the node identifier list of all second-level module nodes that are subordinate to the first-level overall node, and sequentially reads the real-time running data of each module node in the node identifier list at the current moment, and compares it with the normal state range of the corresponding module node. The real-time running data of the module node is also a comprehensive value obtained by weighting the real-time running data of each component node under that module.

[0042] For module nodes whose real-time operating data is within the normal range, the module is determined to be normal, and the normal determination result is recorded. For module nodes whose real-time operating data exceeds the normal range, they are marked as abnormal modules, and the node identifier and abnormal determination result of the abnormal modules are recorded. The list of node identifiers of all third-level component nodes to which the abnormal module belongs is used as the verification target.

[0043] The control system reads the dependency table and obtains a list of node identifiers for all third-level component nodes that are dependent on the abnormal module.

[0044] The real-time operating data mapped to each component node in the list is compared with its corresponding normal state range.

[0045] For each component node, check whether its vibration amplitude, operating current value, and operating temperature value are all within their respective normal operating ranges.

[0046] For component nodes where all operating parameters are within the normal range, the component is considered normal. For component nodes where operating parameters exceed the normal range, the component node is marked as a fault risk point.

[0047] Calculate the deviation of abnormal operating parameters at fault risk points from the upper or lower limit of their normal state range. Calculate the ratio of the deviation to the standard deviation of the operating parameters. The deviation is the difference between the current value of the abnormal operating parameter and the upper limit of the normal state range (when it exceeds the upper limit), or the difference between the lower limit of the normal state range and the current value of the abnormal operating parameter (when it is below the lower limit). Compare this ratio with the first proportional judgment value and the second proportional judgment value.

[0048] The method for determining the first and second proportional judgment values ​​is as follows: extract the ratio of the deviation of the first abnormal data of all components of the equipment recorded in the historical fault case database of the equipment maintenance management system to the standard deviation of the corresponding parameter, arrange all ratios from smallest to largest, take the ratio at the third position after arrangement as the first proportional judgment value, and take the ratio at the two-thirds position after arrangement as the second proportional judgment value.

[0049] When the ratio of deviation to standard deviation is greater than the second proportional judgment value, the risk level is determined to be high risk. When the ratio is greater than the first proportional judgment value and less than or equal to the second proportional judgment value, the risk level is determined to be medium risk. When the ratio is less than or equal to the first proportional judgment value, the risk level is determined to be low risk. Record the component node identifier of the fault risk point, the abnormal operating parameter type, the current value of the abnormal operating parameter, the boundary value of the normal state interval, the deviation, the ratio of deviation to standard deviation, and the risk level.

[0050] It should be noted that the weights of each module node used in the calculation of the real-time integrated operating status value of the overall node are determined as follows: The weight of each module node is determined based on the degree of impact of the module on the overall operational safety of the equipment. The degree of impact includes: whether the failure of the module will lead to the shutdown of the entire machine, the average time required to repair the module after the failure, and the level of secondary safety risks that may be caused by the failure of the module.

[0051] The specific process for determining the weights is as follows: For each module node, extract the downtime data caused by each failure of the module in the past year from the historical fault record database of the equipment maintenance management system. Calculate the proportion of the total downtime caused by the failure of this module to the total downtime caused by all module failures. This proportion is used as the downtime impact component of the module. At the same time, extract the safety risk level score of the module from the equipment operation safety assessment report. After normalizing the safety risk level score, use it as the safety risk component of the module. Add the downtime impact component and the safety risk component and take the arithmetic mean to obtain the initial weight value of the module.

[0052] The initial weight values ​​of all second-level module nodes are normalized by dividing the initial weight value of each module node by the sum of the initial weight values ​​of all module nodes, so that the sum of the weights of all module nodes equals one. The normalized weights of each module node are the preset weights used for weighted summation calculation.

[0053] When the historical fault records or equipment operation safety assessment reports in the equipment maintenance management system are updated, the weights of each module node are recalculated according to the above process, and the stored values ​​in the digital twin model are updated.

[0054] S3. Based on the estimated duration and risk level of the maintenance task corresponding to the fault risk point, traverse the idle time periods between adjacent maintenance tasks in the maintenance scheduling link, filter out the idle gaps that can accommodate maintenance tasks, and insert maintenance tasks. Furthermore, in S3, based on the component node identifier corresponding to the fault risk point, the maintenance procedure library corresponding to the component node is read from the digital twin model. The maintenance procedure library records the standard maintenance procedures for different fault types of the component and the estimated operation time of each procedure. The estimated operation time of each procedure is summarized to obtain the estimated time of the maintenance task. The priority of maintenance tasks is determined based on the risk level; higher risk levels correspond to higher priority maintenance tasks. Read each idle time slot in the maintenance scheduling link in chronological order, compare the duration of the idle time slot with the estimated duration of the maintenance task, and mark the idle time slot as a candidate idle gap if the duration of the idle time slot is greater than or equal to the estimated duration of the maintenance task. Arrange all candidate idle gaps in chronological order to form a candidate gap set. Select the first available gap from the candidate gap set, insert the maintenance task into the available gap, and set the start time of the maintenance task to the start time of the available gap. The end time of the maintenance task is the start time plus the estimated duration of the maintenance task. The execution time of all maintenance tasks located after the insertion position in the maintenance scheduling link is shifted backward as a whole. The shift time is the difference between the estimated duration of the maintenance task and the original idle duration of the candidate idle gap. If the difference is less than or equal to zero, there is no need to shift the subsequent maintenance tasks. Check if there is any overlap in the occurrence time of any two adjacent maintenance tasks in the maintenance scheduling link after the shift. If there is no overlap, it is determined that there is no timing conflict and the maintenance task insertion is completed. If there is a time overlap, the current candidate empty gap is removed from the candidate gap set, and the next candidate empty gap is selected for insertion verification until an insertion position without time conflict is found or the candidate gap set is empty. When the candidate gap set is empty, that is, there is no idle time period in the maintenance scheduling link that can accommodate the estimated duration of the maintenance task, forced insertion is performed according to the priority of the maintenance task: if the priority of the maintenance task is the highest priority, the maintenance task is inserted into the current position of the maintenance scheduling link, and all scheduled maintenance tasks after the current time are shifted backward as a whole, with the shifting time being the estimated duration of the maintenance task. If the priority of a maintenance task is not the highest priority, the maintenance task will be added to the end of the maintenance scheduling chain, after all other scheduled maintenance tasks.

[0055] It should be noted that, based on the component node identifier corresponding to the fault risk point, the maintenance procedure library corresponding to the component node is read from the digital twin model. The maintenance procedure library is a subset of associated data of the component node in the digital twin model, which stores the standard maintenance procedures corresponding to different types of abnormal operating parameters of the component. Each record in the maintenance procedure library includes a description of the fault phenomenon, the corresponding operating parameter type, the name of the maintenance procedure, and the estimated operation time of each procedure. The estimated operation time is determined based on the arithmetic mean of the actual time spent on each procedure in the component's previous maintenance records.

[0056] Using the abnormal operating parameter type of the fault risk point as the search condition, the system searches the maintenance procedure database for records whose fault phenomenon description field matches the current abnormal operating parameter type. If a match is found, the record contains multiple maintenance procedures, each with its own estimated operation time. The estimated operation time of all maintenance procedures is added together to obtain the estimated time of the maintenance task.

[0057] The priority of maintenance tasks is determined based on the risk level. Maintenance tasks with high risk are given the highest priority, those with medium risk are given medium priority, and those with low risk are given general priority.

[0058] The duration of each idle period is read sequentially from the maintenance scheduling link in chronological order. The estimated duration of the maintenance task is compared with the duration of each idle period. Idle periods with a duration greater than or equal to the estimated duration of the maintenance task are marked as candidate idle gaps. All candidate idle gaps are arranged in chronological order to form a candidate gap set.

[0059] Select the candidate available gap that appears first in the candidate gap set and insert the maintenance task into that gap. The start time of the maintenance task is set to the start time of the candidate available gap, and the end time is set to the start time plus the estimated duration of the maintenance task.

[0060] The execution time of all scheduled maintenance tasks located after the insertion position in the maintenance scheduling link is shifted backward as a whole. The shift time is the difference between the estimated duration of the maintenance task and the original idle duration of the candidate idle gap. If the difference is less than or equal to zero, no subsequent tasks need to be shifted.

[0061] Check if any two adjacent maintenance tasks overlap in time in the shifted maintenance scheduling link. Starting from the first maintenance task along the shifted maintenance scheduling link, check each maintenance task and its next adjacent maintenance task for time overlap. If the end time of a maintenance task is later than or equal to the start time of its next adjacent maintenance task, it is determined that there is a time overlap. If there is no time overlap between all adjacent task pairs, it is determined that there is no timing conflict and the maintenance task insertion is completed.

[0062] If a timing conflict occurs after inserting the current candidate vacancy gap, the candidate vacancy gap is removed from the candidate gap set, and the next candidate vacancy gap in the next order is selected. The insertion, shifting, and timing conflict checking operations are repeated until an insertion position without timing conflict is found.

[0063] When the candidate gap set is empty, that is, when there is no idle period in the maintenance scheduling link that can accommodate the expected duration of the maintenance task, forced insertion is performed according to the priority of the maintenance task.

[0064] If the maintenance task has the highest priority, the maintenance task will be inserted into the current position of the maintenance scheduling link. All scheduled maintenance tasks after the current time will be shifted forward, and the shifting time will be the estimated duration of the maintenance task.

[0065] If the priority of a maintenance task is not the highest priority, the maintenance task will be added to the end of the maintenance scheduling link, after all scheduled maintenance tasks. The start time of the maintenance task will be the end time of the last scheduled maintenance task.

[0066] S4. Based on the maintenance scheduling link after insertion, determine the maintenance scheduling scheme and output it.

[0067] Furthermore, in S4, the maintenance scheduling link that has been inserted and has no timing conflict is used as the maintenance scheduling scheme. The maintenance scheduling scheme includes the task identifier, execution component name, task start time, task end time, task priority, and task execution content description for each maintenance task. The maintenance scheduling plan is sent to the mobile terminal devices of maintenance personnel through the communication network. At the same time, the fault risk point information marked in this round of layer-by-layer status verification is updated to the status attribute field of the corresponding component node in the digital twin model, so as to continuously track the status of the node in the next round of status verification.

[0068] It should be noted that the maintenance scheduling scheme uses maintenance scheduling links that do not have timing conflicts after insertion. This scheme includes both existing scheduled maintenance tasks and newly inserted maintenance tasks. Each maintenance task includes a task identifier, the name of the executing component, the task start time, the task end time, the task priority, and a description of the task execution content.

[0069] The task identifier is a unique number automatically generated by the system, the execution component name is the component node name corresponding to the fault risk point, the task start time and end time are determined according to the inserted scheduling link, the task priority is the priority corresponding to the determined risk level, and the task execution content is described as a text of the sequential combination of the names of each maintenance procedure matched in the maintenance procedure library.

[0070] The control system converts the maintenance scheduling plan into a data message in a predetermined format and sends it to the mobile terminal device of the maintenance personnel through the communication network. The data message includes the plan generation time, a complete list of maintenance tasks, and detailed information for each maintenance task.

[0071] At the same time, the fault risk point information marked in this round of layer-by-layer status verification will be written into the status attribute field of the corresponding component node in the digital twin model.

[0072] The information written includes the verification timestamp, the type of abnormal operating parameter, the current value of the abnormal operating parameter, the deviation, the ratio of the deviation to the standard deviation, the risk level, and the corresponding maintenance task identifier.

[0073] The information written in the status attribute field will be read in the next round of layer-by-layer status verification and compared with the real-time running data of the component node to determine whether the abnormal status of the component node has been alleviated or further deteriorated.

[0074] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0075] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0076] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0077] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0078] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of protection of the described technical solution.

Claims

1. A digital twin-driven method for predicting faults and scheduling cloud-based maintenance in car wash equipment, characterized in that: Includes the following steps: S1. Obtain the digital twin model of the target car wash equipment, generate the normal state interval of each node based on the historical normal operation parameters of each node in the digital twin model, and obtain the currently arranged maintenance task sequence as the maintenance scheduling link. S2. Map the real-time collected equipment operation data to the nodes of each level of the digital twin model, perform layer-by-layer status verification in the order from the first level to the third level, mark the fault risk points and determine the corresponding risk level. S3. Based on the estimated duration and risk level of the maintenance task corresponding to the fault risk point, traverse the idle time periods between adjacent maintenance tasks in the maintenance scheduling link, filter out the idle gaps that can accommodate maintenance tasks, and insert maintenance tasks. S4. Based on the maintenance scheduling link after insertion, determine the maintenance scheduling scheme and output it.

2. The digital twin-driven car wash equipment fault prediction and cloud maintenance scheduling method according to claim 1, characterized in that, In S1, a digital twin model of the target car wash equipment is obtained. The first level of the digital twin model includes an overall node representing the overall operating status of the equipment, the second level includes module nodes representing the operating status of each functional module of the equipment, and the third level includes component nodes representing the operating status of each execution component of the equipment. For each node in each level, extract the sequence of operating parameters corresponding to the historical normal operation period of each node from the digital twin model. The operating parameters include at least one of the vibration amplitude, operating current value and operating temperature value of the corresponding equipment unit of each node during operation. Calculate the arithmetic mean and standard deviation of each operating parameter value in the operating parameter sequence. Use the arithmetic mean minus the first multiple standard deviation as the lower limit of the normal state interval, and use the arithmetic mean plus the first multiple standard deviation as the upper limit of the normal state interval. All currently scheduled maintenance tasks are retrieved from the equipment maintenance management system. Each maintenance task includes the task execution period, task duration, and task priority. All maintenance tasks are arranged in chronological order of their execution periods. The time periods between adjacent tasks and between a task and the current time constitute idle periods, forming a maintenance scheduling link.

3. The digital twin-driven car wash equipment fault prediction and cloud maintenance scheduling method according to claim 1, characterized in that, In S2, the sensors deployed on each execution component, functional module and whole machine of the target car wash equipment collect operating data in real time according to the collection cycle, and the operating data collected by each sensor is mapped to the corresponding whole node, module node and component node according to the correspondence between the sensor and the nodes at each level in the digital twin model. For each global node in the first level, the real-time running data mapped to the global node is compared with the normal state range corresponding to the global node. If the real-time running data is within the normal state range, the global node is determined to be in a normal state. If the real-time running data exceeds the normal state range, the global node is determined to be in an abnormal state. When all nodes in the first level are in normal condition, the equipment is determined to be without fault risk, and the current round of layer-by-layer status verification ends. When an overall node in the first level is in an abnormal state, the status verification of all second-level module nodes to which the abnormal overall node belongs is triggered.

4. The digital twin-driven car wash equipment fault prediction and cloud maintenance scheduling method according to claim 3, characterized in that, Triggering a status check on all second-level module nodes to which the abnormal overall node belongs includes the following steps: Read the node identifier of the overall node in an abnormal state, and search for all second-level module nodes in the digital twin model that have a subordinate relationship with the abnormal overall node; For each second-level module node found, the real-time running data mapped to that module node is compared with the normal state range corresponding to that module node; When the real-time running data of all the verified second-level module nodes are within their respective normal state ranges, the anomaly of the first level is determined to be caused by instantaneous data fluctuations or sensor noise. No fault risk points are marked, and the current round of layer-by-layer state verification ends. When the real-time running data of a second-level module node exceeds its normal state range, the module node is marked as an abnormal module, triggering a state check on all third-level component nodes to which the abnormal module belongs.

5. The digital twin-driven car wash equipment fault prediction and cloud maintenance scheduling method according to claim 4, characterized in that, Triggering a status check on all third-level component nodes to which the abnormal module belongs includes the following steps: Read the node identifier of the module node marked as an abnormal module, and search for all third-level component nodes in the digital twin model that have a subordinate relationship with the abnormal module. For each third-level component node found, the real-time running data mapped to that component node is compared with the normal state range corresponding to that component node; Mark the third-level component nodes whose real-time operating data exceeds their normal range as fault risk points; For each fault risk point, the risk level is determined based on the ratio of the deviation of the real-time operating data of the component node from the normal state range to the standard deviation of the corresponding component node in the normal state range. The larger the ratio of the deviation to the standard deviation, the higher the risk level.

6. The digital twin-driven car wash equipment fault prediction and cloud maintenance scheduling method according to claim 1, characterized in that, In step S3, based on the component node identifier corresponding to the fault risk point, the maintenance procedure library corresponding to the component node is read from the digital twin model. The maintenance procedure library records the standard maintenance procedures for different fault types of the component and the estimated operation time of each procedure. The estimated operation time of each procedure is summarized to obtain the estimated time of the maintenance task. The priority of maintenance tasks is determined based on the risk level; higher risk levels correspond to higher priority maintenance tasks. Read each idle time slot in the maintenance scheduling link in chronological order, compare the duration of the idle time slot with the estimated duration of the maintenance task, and mark the idle time slot as a candidate idle gap if the duration of the idle time slot is greater than or equal to the estimated duration of the maintenance task. Arrange all candidate idle gaps in chronological order to form a candidate gap set. Select the first available gap from the candidate gap set, insert the maintenance task into the available gap, and set the start time of the maintenance task to the start time of the available gap. The end time of the maintenance task is the start time plus the estimated duration of the maintenance task.

7. The digital twin-driven car wash equipment fault prediction and cloud maintenance scheduling method according to claim 6, characterized in that, The execution time of all maintenance tasks located after the insertion position in the maintenance scheduling link is shifted backward as a whole. The shift time is the difference between the estimated duration of the maintenance task and the original idle duration of the candidate idle gap. If the difference is less than or equal to zero, there is no need to shift the subsequent maintenance tasks. Check if there is any overlap in the occurrence time of any two adjacent maintenance tasks in the maintenance scheduling link after the shift. If there is no overlap, it is determined that there is no timing conflict and the maintenance task insertion is completed. If there is a time overlap, the current candidate empty gap is removed from the candidate gap set, and the next candidate empty gap is selected for insertion verification until an insertion position without time conflict is found or the candidate gap set is empty.

8. The digital twin-driven car wash equipment fault prediction and cloud maintenance scheduling method according to claim 7, characterized in that, When the candidate gap set is empty, that is, there is no idle time period in the maintenance scheduling link that can accommodate the estimated duration of the maintenance task, forced insertion is performed according to the priority of the maintenance task: if the priority of the maintenance task is the highest priority, the maintenance task is inserted into the current position of the maintenance scheduling link, and all scheduled maintenance tasks after the current time are shifted backward as a whole, with the shifting time being the estimated duration of the maintenance task. If the maintenance task has a non-highest priority, add the maintenance task to the end of the maintenance scheduling chain, after all other scheduled maintenance tasks.

9. The digital twin-driven method for predicting faults and scheduling cloud-based maintenance of car wash equipment according to claim 1, characterized in that, In step S4, the maintenance scheduling link that has been inserted and has no timing conflict is used as the maintenance scheduling scheme. The maintenance scheduling scheme includes the task identifier, execution component name, task start time, task end time, task priority and task execution content description for each maintenance task. The maintenance scheduling plan is sent to the mobile terminal devices of maintenance personnel through the communication network. At the same time, the fault risk point information marked in this round of layer-by-layer status verification is updated to the status attribute field of the corresponding component node in the digital twin model, so as to continuously track the status of the node in the next round of status verification.

Images