A cloud computing platform-based predictive failure maintenance method and system

Abstract

The application discloses a kind of predictive failure maintenance method and system based on cloud computing platform, belong to cloud computing platform technical field;Device basic use state of target equipment end is obtained from the edge node in cloud platform;Data acquisition cycle and task execution number set and task execution time point set are constructed, and the task density of target equipment in single data acquisition cycle is calculated;The health sub-score of task density, the health sub-score of current, the health sub-score of voltage and the health sub-score of temperature of target equipment in single data acquisition cycle are calculated;The health score change rate of target equipment in single data acquisition cycle is calculated;The residual operational time of target equipment in single data acquisition cycle is analyzed and predicted;Threshold value is preset, and early warning maintenance is analyzed and carried out, the timeliness, accuracy and resource scheduling efficiency of maintenance are improved, equipment management is changed from traditional passive repair mode to active predictive maintenance mode.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of cloud computing platforms, and particularly to a predictive failure maintenance method and system based on a cloud computing platform. BACKGROUND

[0002] With the continuous advancement of industrial equipment intelligence and manufacturing system digitization, equipment failure prediction and maintenance management has gradually become an important research direction in the field of intelligent manufacturing and industrial Internet of Things (IIoT). Traditional equipment maintenance modes mainly include two ways of post-maintenance and regular maintenance. The former relies on manual or automatic alarm after the occurrence of failure, which is easy to cause uncontrollable loss. The latter is based on fixed time interval to implement maintenance, which improves the failure prevention ability, but also has problems such as resource waste and inefficiency. In recent years, with the rapid development of cloud computing, edge computing and big data analysis technology, predictive maintenance (PdM) has gradually emerged. It can give early warning before the equipment fails by real-time collection and data modeling analysis of the running state of the equipment, thereby effectively reducing the unplanned downtime rate and improving the equipment utilization rate and management intelligence level. Especially under the support of cloud computing platform, large-scale data processing, model unified deployment and state information sharing between devices become possible, which provides a solid foundation for building a multi-source data-driven predictive maintenance method.

[0003] However, the existing technology mostly relies on a single data source or low-dimensional indicators, and fails to fully integrate task execution behavior and multi-dimensional sensor data, resulting in low accuracy of the prediction results. Secondly, there is a lack of systematic modeling of the relationship between the running task density of the equipment and its running health, which is not conducive to judging the real load state of the equipment and its change trend. Thirdly, the current predictive maintenance system usually only focuses on the current state, ignoring the change rate and trend prediction of the state, resulting in lagging maintenance strategy and lack of forward-looking early warning mechanism. In addition, the maintenance scheduling between multiple devices often relies on human judgment, lacks intelligent clustering and strategy unification mechanism, and is difficult to maintain efficient response as the scale of the equipment expands. SUMMARY

[0004] The present application aims to provide a predictive failure maintenance method and system based on a cloud computing platform to solve the problems raised in the background art.

[0005] To solve the above technical problems, the present application provides the following technical solutions:

[0006] The application discloses a predictive failure maintenance method based on a cloud computing platform, and the method comprises the following steps: S1, obtaining device basic usage states of a target device from edge nodes in the cloud platform; constructing a data collection cycle and a task execution number set; S2, constructing a task execution time point set corresponding to each task, and calculating a task density of the target device in a single data collection cycle; based on the task density, calculating a task density health sub-score of the target device in the single data collection cycle; S3, based on the device basic usage states, calculating a current health sub-score, a voltage health sub-score and a temperature health sub-score of the target device in the single data collection cycle, and combining the task density health sub-score to calculate a comprehensive health score of the target device in the single data collection cycle; S4, calculating a health score change rate of the target device in the single data collection cycle; analyzing and predicting a remaining operation time of the target device in the single data collection cycle; and presetting a threshold value to analyze and perform early warning maintenance.

[0007] As a preferred scheme of the predictive failure maintenance method based on the cloud computing platform, the device basic usage states of the target device are obtained from the edge nodes in the cloud platform, wherein one edge node corresponds to one target device, the device basic usage states comprise task execution numbers, task execution times, task execution state data and sensor data of the device, the task execution state data comprises current data and voltage data of the device during task execution, and the sensor data comprises temperature data of the device during task execution.

[0008] The data collection cycle and the task execution number set are constructed, and the task execution number set collected in the i-th data collection cycle is denoted as NTE i i,a |a∈[1,A]} wherein TE i,a represents the a-th task in the i-th data collection cycle, and A represents the total number of task execution in the i-th data collection cycle.

[0009] It should be noted that the device basic usage states are obtained from the edge nodes in the cloud platform, the data collection cycle and the task execution number set are constructed, the structured collection and organization of the key operation behaviors (such as task execution frequency, time and operation state) and sensor data (current, voltage and temperature) of the target device in different operation cycles are realized, so that the comprehensiveness, timeliness and continuity of the data source are ensured, and finally high-quality basic data support is provided for the task density analysis and health score calculation in subsequent stages, and a reliable data perception foundation is constructed.

[0010] As a preferred scheme of the predictive failure maintenance method based on the cloud computing platform, the task TE i,a ​The corresponding task execution time is evenly divided into N task execution time points, and a task execution time point set TET = {Y} is constructed. n (TE i,a )|n∈[1,N]}, where Y n (TE i,a ) indicates task (TE) i,a The corresponding nth task execution time point in the task execution time, where N represents the total number of task execution time points; assign task execution time points Y respectively n (TE i,a The current, voltage, and temperature data under ( ) are denoted as DL[Y n (TE i,a )]、DY[Y n (TE i,a )] and WD[Y n (TE i,a )];

[0011] The task density of the target device during the i-th data acquisition cycle is calculated using the following formula: Among them, TD i ΔT represents the task density of the target device during the i-th data acquisition cycle. i This represents the duration of the i-th data acquisition cycle;

[0012] Based on the task density TD of the target device in the i-th data acquisition cycle i Calculate the task density health sub-score for the i-th data collection period using the following formula: in, κ represents the task density health sub-score for the i-th data collection cycle. TD TD represents the preset task density scaling factor. ref This represents the preset task density standard reference value, δ TD This indicates the preset task density deviation value.

[0013] It should be noted that by constructing a set of task execution time points and calculating task density and its health sub-score, a quantitative assessment of the task load level of the equipment within a specific collection period is achieved. By utilizing the deviation relationship between density and standard value, scaling factors and deviation terms are introduced, which can identify potential operational anomalies of the equipment under high-frequency use, low-frequency use, or uneven load. Ultimately, this improves the sensitivity of the health assessment system to the task behavior level and lays the foundation for more refined status diagnosis and prediction.

[0014] As a preferred embodiment of the predictive fault maintenance method based on a cloud computing platform described in this invention, based on the task execution time point Y... n (TE i,aCurrent data DL[Y] under ) n (TE i,a Voltage data DY[Y] n (TE i,a )] and temperature data WD[Y n (TE i,a The current health sub-score, voltage health sub-score, and temperature health sub-score are calculated for the i-th data acquisition period, as follows:

[0015]

[0016] in, and Let κ represent the current health sub-score, voltage health sub-score, and temperature health sub-score respectively within the i-th data acquisition period. DL This represents the scaling factor for the preset current data. DL represents the average current data. ref Indicates the preset current reference value, α DL This represents the preset current fluctuation weight, σ. DL δ represents the standard deviation of current data. DL This represents the preset current deviation value, κ. DY This represents the scaling factor for the preset voltage data. DY represents the average voltage data. ref Indicates the preset voltage reference value, α DY This represents the preset voltage fluctuation weight, σ. DY δ represents the standard deviation of voltage data. DY This represents the preset voltage deviation value, κ. WD This represents the scaling factor for the preset temperature data. This represents the average temperature data, WD ref Indicates the preset temperature reference value, α WD This represents the preset temperature fluctuation weight, σ. WD δ represents the standard deviation of temperature data. WD This indicates the preset temperature deviation value;

[0017] The task density health sub-score for the i-th data collection period is respectively... Current health sub-score Voltage health sub-score Temperature and health sub-score Weighted fusion is performed to calculate the comprehensive health score of the target device within the i-th data acquisition cycle, denoted as HS. i .

[0018] It should be noted that by calculating the health sub-score of the three types of operating parameters of current, voltage and temperature, and weighting and fusing with the task density health sub-score, the current working state of the equipment is evaluated from multiple dimensions, especially considering the cross influence between the operating condition and the execution intensity, thereby effectively avoiding the one-sidedness of single index evaluation, improving the comprehensive expression ability of the health score, and finally building a set of comprehensive health score index reflecting the real state of the equipment, providing a quantitative basis for predictive maintenance.

[0019] As a preferred scheme of the predictive fault maintenance method based on the cloud computing platform, the comprehensive health score HS of the target equipment in the i-1th data collection period is obtained i-1 , and based on the comprehensive health score HS of the target equipment in the i th data collection period i and the cycle time length ΔT of the i th data collection period i , the health score change rate of the target equipment in the i th data collection period is calculated, and the calculation formula is: Wherein, HDR i represents the health score change rate of the target equipment in the i th data collection period.

[0020] It should be noted that the formula calculates the change speed of the health score with time, and reflects the trend of the deterioration (or improvement) of the equipment health. If HDR i is negative, it means that the health is declining (the smaller the negative value, the faster the deterioration); if HDR i is positive, it means that the health is recovering. The formula can identify the deterioration speed. For example, the previous period HS i-1 =0.8, the current HS i =0.6, and the cycle ΔT i =1h, then indicates that the health decreases by 0.2 per hour, and if the speed continues, it can be predicted when the failure threshold will be reached.

[0021] Based on the comprehensive health score HS of the target equipment in the i th data collection period i , the health score change rate HDR i and the task density TD i , the remaining operational time of the target equipment in the i th data collection period is analyzed and predicted, which is as follows:

[0022]

[0023] Wherein, RUL i represents the remaining operational time of the target equipment in the i th data collection period, HS threpresents a preset comprehensive health score threshold of the target device in the ith data collection period, β represents a preset calibration coefficient, TD ref represents a preset task density standard reference value;

[0024] represents a preset residual operational time threshold of the target device in the ith data collection period, if the residual operational time RUL of the target device in the ith data collection period is less than the residual operational time threshold, it is determined that the target device has a life health abnormal failure in the ith data collection period, and a warning is issued and a relevant staff is reminded to maintain; i

[0025] The residual operational time threshold of the target device in the ith data collection period is uploaded to a cloud platform, and different devices are clustered based on the device basic usage state, and the clustered devices are uniformly maintained;

[0026] Let i=i+1, iterate the data collection period, and perform dynamic predictive maintenance of the device.

[0027] It should be noted that by introducing the health score change rate, the residual operational time of the device is calculated by combining the current score and the task density, which realizes the transition from "static state evaluation" to "dynamic running trend analysis". At the same time, the health threshold and the warning mechanism are set, so that maintenance instructions can be issued in time in the early stage of device performance deterioration, avoiding running interruption or failure spread, and finally improving the life cycle management efficiency and safety guarantee ability of the device. Provide a technical foundation for the cloud platform to realize batch device cluster-level intelligent predictive maintenance.

[0028] A predictive failure maintenance system based on a cloud computing platform, the system comprises: a data acquisition and set construction module, a task density and health score calculation module, a sub-score calculation and comprehensive score calculation module, and a change rate calculation and analysis warning module;

[0029] The data acquisition and set construction module: acquires the device basic usage state of the target device from the edge node in the cloud platform; constructs a data collection period and a task execution timeset;

[0030] The task density and health score calculation module: constructs a task execution time point set corresponding to each task, and calculates the task density of the target device in a single data collection period; based on the task density, the task density health sub-score of the target device in a single data collection period is calculated;

[0031] ​The sub-score calculation and comprehensive score calculation module: based on the device basic usage state, calculate the current health sub-score, voltage health sub-score and temperature health sub-score of the target device in a single data acquisition period, and combine the task density health sub-score to calculate the comprehensive health score of the target device in a single data acquisition period;

[0032] The change rate calculation and analysis warning module: calculate the health score change rate of the target device in a single data acquisition period; analyze and predict the remaining operating time of the target device in a single data acquisition period; preset a threshold value, analyze and perform early warning maintenance.

[0033] Further, the data acquisition and set construction module includes a data acquisition unit and a set construction unit;

[0034] The data acquisition unit: acquires the device basic usage state of the target device from the edge node in the cloud platform, wherein one edge node corresponds to one target device, the device basic usage state includes the task execution times, task execution time, task execution state data and sensor data of the device, the task execution state data includes the current data and voltage data of the device during task execution, and the sensor data includes the temperature data of the device during task execution;

[0035] The set construction unit: constructs a data acquisition period and a task execution times set.

[0036] Further, the task density and health score calculation module includes a task density calculation unit and a health score calculation unit;

[0037] The task density calculation unit: divides the task execution time corresponding to the task into N task execution time points, constructs a task execution time point set, and calculates the task density of the target device in the i th data acquisition period;

[0038] The health score calculation unit: based on the task density of the target device in the i th data acquisition period, calculates the task density health sub-score of the i th data acquisition period.

[0039] Further, the sub-score calculation and comprehensive score calculation module includes a sub-score calculation unit and a comprehensive score calculation unit;

[0040] The sub-score calculation unit: based on the current data, voltage data and temperature data at the task execution time point, respectively calculates the current health sub-score, voltage health sub-score and temperature health sub-score in the i th data acquisition period;

[0041] The comprehensive score calculation unit: respectively weights and fuses the task density health sub-score, the current health sub-score, the voltage health sub-score and the temperature health sub-score of the i-th data acquisition period, and calculates the comprehensive health score of the target device in the i-th data acquisition period.

[0042] Further, the change rate calculation and analysis early warning module comprises a change rate calculation unit and an analysis early warning unit.

[0043] The change rate calculation unit: obtains the comprehensive health score of the target device in the i-1th data acquisition period, and calculates the health score change rate of the target device in the i-th data acquisition period based on the comprehensive health score of the target device in the i-th data acquisition period and the cycle time length of the i-th data acquisition period.

[0044] The analysis early warning unit: based on the comprehensive health score, the health score change rate and the task density of the target device in the i-th data acquisition period, analyzes and predicts the remaining operational time of the target device in the i-th data acquisition period; a remaining operational time threshold of the target device in the i-th data acquisition period is preset, if the remaining operational time of the target device in the i-th data acquisition period is less than the remaining operational time threshold, it is determined that the target device has a life health abnormal failure in the i-th data acquisition period, and an early warning is issued and the relevant staff is reminded to maintain; the remaining operational time threshold of the target device in the i-th data acquisition period is uploaded to the cloud platform, and different devices are clustered based on the basic use state of the device, and the clustered devices are uniformly maintained; let i=i+1, iterate the data acquisition period, and dynamically predict and maintain the device.

[0045] Compared with the prior art, the present application has the beneficial effects that: in the predictive failure maintenance method and system based on a cloud computing platform provided by the present application, the task execution times, state data and sensor data of a target device are obtained from the edge nodes of the cloud platform, a data acquisition cycle and a task execution times set are constructed, the structured perception of the device running state is realized, and high-quality data basis is provided for subsequent evaluation; then, by constructing a task execution time point set and calculating the task density and its health sub-score, the quantitative monitoring of the device load level is realized, so as to identify potential abnormalities under high load or low frequency operation; on this basis, the sub-score evaluation of various working condition data such as current, voltage and temperature is further carried out, and the comprehensive health score is generated by weighted fusion with the task density score, so as to realize the multi-dimensional accurate description of the device state; then, by analyzing the change rate of the health score and combining the task density to calculate the remaining operating time of the device, the dynamic trend analysis and threshold early warning mechanism are introduced, not only the early identification of device performance degradation is realized, but also the clustering of different devices and the deployment of centralized maintenance strategies are supported, which significantly improves the timeliness, accuracy and resource scheduling efficiency of maintenance, promotes the transformation of device management from the traditional passive repair mode to the active predictive maintenance mode, and has wide application value and promotion prospect. BRIEF DESCRIPTION OF DRAWINGS

[0046] The accompanying drawings are included to provide a further understanding of the present application, and constitute a part of the specification, illustrate the present application together with the embodiments thereof, and explain the principles of the present application, and do not constitute a limitation of the present application.

[0047] Figure 1 is a step schematic diagram of a predictive failure maintenance method based on a cloud computing platform of the present application;

[0048] Figure 2 is a structure schematic diagram of a predictive failure maintenance system based on a cloud computing platform of the present application. DETAILED DESCRIPTION

[0049] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0050] Please refer to Figure 1 In the present embodiment one: a predictive failure maintenance method based on a cloud computing platform is provided, which comprises the following steps:

[0051] Step S1: obtaining the device basic usage state of the target device end from the edge nodes in the cloud platform; constructing a data collection cycle and a task execution number set.

[0052] Specifically, the device basic usage state of the target device end is obtained from the edge nodes in the cloud platform, wherein one edge node corresponds to one target device, the device basic usage state includes the task execution number, the task execution time, the task execution state data and the sensor data of the device, the task execution state data includes the current data and the voltage data of the device during task execution, and the sensor data includes the temperature data of the device during task execution;

[0053] A data collection cycle and a task execution number set are constructed, and the task execution number set collected in the i th data collection cycle is denoted as NTE i i,a |a∈[1,A]},wherein TE i,a represents the a th task in the i th data collection cycle, and A represents the total number of task executions in the i th data collection cycle.

[0054] Step S2: constructing a task execution time point set corresponding to each task, and calculating the task density of the target device in a single data collection cycle; based on the task density, calculating the task density health sub-score of the target device in a single data collection cycle.

[0055] Specifically, the task TE i,a is divided into N task execution time points, and a task execution time point set TET is constructed. n i,a |n∈[1,N]},wherein Y n (TE i,a ) represents the n th task execution time point in the task execution time corresponding to the task TE i,a , and N represents the total number of task execution time points; the current data, the voltage data and the temperature data under the task execution time point Y n (TE i,a ) are respectively denoted as DL[Y n (TE i,a )], DY[Y n (TE i,a )] and WD[Y n (TE i,a )].

[0056] The task density of the target device in the i th data collection cycle is calculated, and the calculation formula is: wherein TD i represents the task density of the target device in the i th data collection cycle, and ΔT i ​​denotes the length of the cycle time of the i-th data collection cycle;

[0057] It should be noted that the task density reflects the load intensity of the device in unit time, and is a core index for measuring the degree of busyness of the device. Excessive load of the device (excessive task density) will accelerate the aging of components (such as temperature rise caused by frequent starting of the motor), and excessive load (excessive task density) may cause idle loss (such as rusting of mechanical components), so the association between load and health needs to be quantified by task density. In actual scenarios, for example, a mechanical arm on a production line, if it performs 10 handling tasks (A = 10, ΔT i = 1h) in 1 hour, then the task density TD i = 10 times / hour. If the reasonable load of the mechanical arm is 8 times / hour, the excessive task density will prompt that there may be an overload risk, providing a load basis for subsequent health scoring.

[0058] Based on the task density TD i of the target device in the i-th data collection cycle, the task density health sub-score of the i-th data collection cycle is calculated, and the calculation formula is: wherein, denotes the task density health sub-score of the i-th data collection cycle, κ TD denotes a preset task density scaling factor, TD ref denotes a preset task density standard reference value, δ TD denotes a preset task density deviation value.

[0059] It should be noted that the formula uses the Sigmoid function to convert the deviation of the task density from the standard value into a health score. When TD i is within the range of TD ref ± δ TD , the deviation term (|TD i -TD ref |- δ TD ) ≤ 0, the exponential term tends to 0, and the score is close to 1 (healthy); when TD i exceeds the allowed range, the deviation term is positive, the exponential term increases, and the score rapidly decreases (unhealthy);

[0060] The formula quantifies the impact of load on health. For example, the task density standard reference value TD ref = 8 times / hour, the task density deviation value δ TD = 2, and the allowed range is 6-10 times / hour, if the task density TD i = 12 times / hour in a certain cycle, the deviation term (|TD i -TD ref |- δ TD ) = 12-8-2 = 2, and κ TD= 1, by K TD After scaling, the exponential term exp(K TD × (|TD i -TD ref |-D TD )) = exp(1x2) = 7.389, task density health sub-score If the task density health sub-score of the i-th data collection period can be predicted in advance, it can be prompted that the overload risk needs attention.

[0061] Step S3: Based on the device basic usage state, the current health sub-score, the voltage health sub-score and the temperature health sub-score of the target device in a single data collection period are calculated, and the comprehensive health score of the target device in a single data collection period is calculated in combination with the task density health sub-score.

[0062] Specifically, based on the current data DL[Y n (TE i,α )], the voltage data DY[Y n (TE i,α )] and the temperature data WD[Y n (TE i,a )] at the task execution time point Y n (TE i,a )], the current health sub-score, the voltage health sub-score and the temperature health sub-score in the i-th data collection period are calculated respectively, and the specific calculation is as follows:

[0063]

[0064] Among them, and respectively represent the current health sub-score, the voltage health sub-score and the temperature health sub-score in the i-th data collection period, K DL represents a preset scaling factor of current data, represents the mean of current data, DL ref represents a preset current reference value, a DL represents a preset current fluctuation weight, s DL represents the standard deviation of current data, D DL represents a preset current deviation value, K DY represents a preset scaling factor of voltage data, represents the mean of voltage data, DY ref represents a preset voltage reference value, a DY represents a preset voltage fluctuation weight, s DY represents the standard deviation of voltage data, D DY represents a preset voltage deviation value, K WD represents a preset scaling factor of temperature data, This represents the average temperature data, WD ref Indicates the preset temperature reference value, α WD This represents the preset temperature fluctuation weight, σ. WD δ represents the standard deviation of temperature data. WD This indicates the preset temperature deviation value;

[0065] It should be noted that, taking current data as an example, Used to measure the degree to which the overall current deviates from the standard (e.g., if the normal current of a motor is 10A and the average is 12A, then the deviation is 2A); α DL ×σ DL Used to measure the impact of current fluctuations (e.g., frequent fluctuations in current between 8 and 14 A with a large standard deviation may still damage equipment due to impact, even if the mean is close to the standard); the sum of the two minus the allowable deviation δ DL Then, by using the Sigmoid function, it can be converted into a health score, which can detect abnormal electrical parameters.

[0066] For example, suppose a certain motor current reference value DL ref =10A, scaling factor κ for current data DL =1, current deviation value δ DL =1A, preset current fluctuation weight α DL =0.5, mean value of current data Standard deviation of current data σ DL =2A, then the current health sub-score in the i-th data acquisition period is...

[0067] The task density health sub-score for the i-th data collection period is respectively... Current health sub-score Voltage health sub-score Temperature and health sub-score Weighted fusion is performed to calculate the comprehensive health score of the target device within the i-th data acquisition cycle, denoted as HS. i .

[0068] Step S4: Calculate the rate of change of the health score of the target device within a single data acquisition cycle; analyze and predict the remaining uptime of the target device within a single data acquisition cycle; preset thresholds, analyze and perform early warning maintenance.

[0069] Specifically, obtain the comprehensive health score (HS) of the target device during the (i-1)th data acquisition cycle. i-1 And based on the comprehensive health score HS of the target device within the i-th data acquisition cycle i and the cycle time length ΔT of the i-th data acquisition cycle i, calculate the health score change rate of the target device in the i-th data collection period, and the calculation formula is: wherein, HDR i represents the health score change rate of the target device in the i-th data collection period;

[0070] Based on the comprehensive health score HS i of the target device in the i-th data collection period, the health score change rate HDR i and the task density TD i , the remaining operational time of the target device in the i-th data collection period is analyzed and predicted, and the specific process is as follows:

[0071]

[0072] wherein, RUL i represents the remaining operational time of the target device in the i-th data collection period, HS th represents the preset comprehensive health score threshold of the target device in the i-th data collection period, and β represents the preset calibration coefficient. TD ref represents the preset task density standard reference value.

[0073] It should be noted that the remaining life of the device depends not only on the current health status, but also on the load intensity (high load will accelerate aging). The formula reflects the gap between the current health and the failure through HS i -HS th , corrects the impact of load on life through , reflects the health attenuation rhythm through , and multi-dimensional fusion makes the prediction more accurate. Different devices have individual differences due to manufacturing errors, service life, etc. (such as two motors of the same model, which may have different aging speeds). The calibration coefficient β in the formula can be adjusted through historical data (such as increasing β for devices that age faster), which corrects the deviation between theoretical prediction and actual life, so that the formula can still maintain high precision in batch devices.

[0074] A remaining operational time threshold of the target device in the i-th data collection period is preset. If the remaining operational time RUL i of the target device in the i-th data collection period is less than the remaining operational time threshold, it is determined that the target device has a life health abnormal failure in the i-th data collection period, and a warning is issued and the relevant staff is reminded to maintain;

[0075] The remaining operational time threshold of the target device in the i-th data collection period is uploaded to the cloud platform, and different devices are clustered based on the basic use state of the device, and the clustered devices are uniformly maintained;

[0076] Let i = i + 1, iterate the data acquisition period, and perform dynamic predictive maintenance of the equipment.

[0077] Please refer to Figure 2 In the second embodiment: a predictive failure maintenance system based on a cloud computing platform is provided, which comprises a data acquisition and set construction module, a task density and health score calculation module, a sub-score calculation and comprehensive score calculation module, and a change rate calculation and analysis warning module.

[0078] The data acquisition and set construction module: acquires the device basic usage state of the target device end from the edge node in the cloud platform; constructs the data acquisition period and the task execution times set;

[0079] The task density and health score calculation module: constructs the task execution time point set corresponding to each task, and calculates the task density of the target device in a single data acquisition period; based on the task density, calculates the task density health sub-score of the target device in a single data acquisition period;

[0080] The sub-score calculation and comprehensive score calculation module: based on the device basic usage state, calculates the current health sub-score, voltage health sub-score and temperature health sub-score of the target device in a single data acquisition period, and combines the task density health sub-score to calculate the comprehensive health score of the target device in a single data acquisition period;

[0081] The change rate calculation and analysis warning module: calculates the health score change rate of the target device in a single data acquisition period; analyzes and predicts the remaining operational time of the target device in a single data acquisition period; presets a threshold, analyzes and performs warning maintenance.

[0082] Further, the data acquisition and set construction module comprises a data acquisition unit and a set construction unit;

[0083] The data acquisition unit: acquires the device basic usage state of the target device end from the edge node in the cloud platform, wherein one edge node corresponds to one target device, the device basic usage state includes the task execution times, task execution time, task execution state data and sensor data of the device, the task execution state data includes the current data and voltage data of the device during task execution, and the sensor data includes the temperature data of the device during task execution;

[0084] The set construction unit: constructs the data acquisition period and the task execution times set.

[0085] Further, the task density and health score calculation module comprises a task density calculation unit and a health score calculation unit;

[0086] The task density calculation unit: evenly divides the task execution time corresponding to the task into N task execution time points, constructs a task execution time point set; calculates the task density of the target device in the i th data acquisition cycle;

[0087] The health score calculation unit: based on the task density of the target device in the i th data acquisition cycle, calculates the task density health sub-score of the i th data acquisition cycle.

[0088] Further, the sub-score calculation and comprehensive score calculation module includes a sub-score calculation unit and a comprehensive score calculation unit;

[0089] The sub-score calculation unit: based on the current data, voltage data and temperature data at the task execution time point, respectively calculates the current health sub-score, voltage health sub-score and temperature health sub-score in the i th data acquisition cycle;

[0090] The comprehensive score calculation unit: respectively weights and fuses the task density health sub-score, current health sub-score, voltage health sub-score and temperature health sub-score of the i th data acquisition cycle, and calculates the comprehensive health score of the target device in the i th data acquisition cycle.

[0091] Further, the change rate calculation and analysis warning module includes a change rate calculation unit and an analysis warning unit;

[0092] The change rate calculation unit: obtains the comprehensive health score of the target device in the i-1 th data acquisition cycle, and based on the comprehensive health score of the target device in the i th data acquisition cycle and the cycle time length of the i th data acquisition cycle, calculates the health score change rate of the target device in the i th data acquisition cycle;

[0093] The analysis warning unit: based on the comprehensive health score, health score change rate and task density of the target device in the i th data acquisition cycle, analyzes and predicts the remaining operational time of the target device in the i th data acquisition cycle; preset the remaining operational time threshold of the target device in the i th data acquisition cycle, if the remaining operational time of the target device in the i th data acquisition cycle is less than the remaining operational time threshold, it is determined that the target device in the i th data acquisition cycle exists life health abnormal failure, then issue a warning and remind the relevant staff to maintain; upload the remaining operational time threshold of the target device in the i th data acquisition cycle to the cloud platform, and based on the device basic use state, cluster different devices, and uniformly maintain the clustered devices; let i=i+1, iterate the data acquisition cycle, and perform dynamic prediction and maintenance of the device.

[0094] It is to be noted that, in the present text, relational terms such as first and second and the like can be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises", "comprising", or any other variations thereof, are intended to cover a non-exclusive inclusion such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can include other elements not expressly listed or inherent to such process, method, article, or apparatus.

[0095] Finally, it should be noted that the above-mentioned only constitutes the preferred embodiments of the present application and is not intended to limit the present application, and although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art can still make modifications to the technical solutions described in the foregoing embodiments or make equivalent replacements to some of the technical features. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A predictive fault maintenance method based on a cloud computing platform, characterized in that, The method includes the following steps: Step S1: Obtain the basic usage status of the target device from the edge nodes in the cloud platform. The basic usage status includes the number of task executions, task execution time, task execution status data, and sensor data. The task execution status data includes the current and voltage data of the device during task execution, and the sensor data includes the temperature data of the device during task execution. Construct a data acquisition cycle and task execution count set. Step S2: Construct a set of task execution time points corresponding to each task, and calculate the task density of the target device within a single data acquisition cycle; based on the task density, calculate the task density health sub-score of the target device within a single data acquisition cycle; Step S3: Based on the basic usage status of the equipment, calculate the current health sub-score, voltage health sub-score, and temperature health sub-score of the target equipment within a single data acquisition cycle, and combine them with the task density health sub-score to calculate the comprehensive health score of the target equipment within a single data acquisition cycle; Step S4: Calculate the rate of change of the health score of the target device within a single data acquisition cycle; analyze and predict the remaining uptime of the target device within a single data acquisition cycle; preset thresholds, analyze and perform early warning maintenance; The specific implementation process of step S2 includes: The a-th task within the i-th data acquisition cycle The corresponding task execution time is evenly divided into N task execution time points to construct a task execution time point set. ,in, Indicates task The nth task execution time point in the corresponding task execution time, where N represents the total number of task execution time points; the task execution time points are respectively... The current, voltage, and temperature data are recorded as follows: , and ; The task density of the target device during the i-th data acquisition cycle is calculated using the following formula: ,in, This represents the task density of the target device during the i-th data acquisition cycle. This represents the duration of the i-th data acquisition cycle; Based on the task density of the target device during the i-th data acquisition cycle Calculate the task density health sub-score for the i-th data collection period using the following formula: ,in, This represents the task density health sub-score for the i-th data collection cycle. This represents the preset task density scaling factor. This represents the preset task density standard reference value. This indicates the preset task density deviation value; The specific implementation process of step S3 includes: Based on task execution time point Current data below Voltage data and temperature data The current health sub-score, voltage health sub-score, and temperature health sub-score are calculated for the i-th data acquisition period, as follows: ； in, , and These represent the current health sub-score, voltage health sub-score, and temperature health sub-score, respectively, within the i-th data acquisition cycle. This represents the scaling factor for the preset current data. This represents the average value of the current data. This indicates the preset current reference value. This indicates the preset current fluctuation weight. This represents the standard deviation of the current data. This indicates the preset current deviation value. This represents the scaling factor for the preset voltage data. This represents the average voltage data. This indicates the preset voltage reference value. This indicates the preset voltage fluctuation weight. This represents the standard deviation of voltage data. This indicates the preset voltage deviation value. This represents the scaling factor for the preset temperature data. This represents the average temperature data. This indicates the preset temperature reference value. This indicates the preset temperature fluctuation weight. This represents the standard deviation of temperature data. This indicates the preset temperature deviation value; The task density health sub-score for the i-th data collection period is respectively... Current health sub-score Voltage health sub-score Temperature and health sub-score Weighted fusion is performed to calculate the comprehensive health score of the target device within the i-th data acquisition cycle, denoted as . ; The specific implementation process of step S4 includes: Obtain the comprehensive health score of the target device during the (i-1)th data acquisition cycle. And based on the comprehensive health score of the target device within the i-th data acquisition cycle. and the cycle time length of the i-th data acquisition cycle Calculate the rate of change of the health score of the target device within the i-th data acquisition period. The calculation formula is as follows: ,in, This represents the rate of change of the health score of the target device during the i-th data acquisition cycle; Based on the comprehensive health score of the target device during the i-th data acquisition cycle Health score change rate and task density The remaining uptime of the target device during the i-th data acquisition cycle is analyzed and predicted as follows: ； in, This represents the remaining uptime of the target device within the i-th data acquisition cycle. This represents the preset comprehensive health score threshold for the target device within the i-th data collection period. This indicates the preset calibration coefficient. This indicates the preset task density standard reference value; A preset threshold for the remaining operational time of the target device within the i-th data acquisition cycle is established. If the remaining operational time of the target device within the i-th data acquisition cycle... If the remaining runnable time is less than the threshold, it is determined that the target device has an abnormal lifespan health fault within the i-th data acquisition cycle, and an early warning is issued to remind relevant personnel to perform maintenance; The remaining uptime threshold of the target device within the i-th data collection cycle is uploaded to the cloud platform, and different devices are clustered based on the basic usage status of the devices, and the clustered devices are maintained in a unified manner. Let i = i + 1, iterate through the data acquisition cycle, and perform dynamic predictive maintenance of the equipment.

2. The predictive fault maintenance method based on a cloud computing platform according to claim 1, characterized in that, The specific implementation process of step S1 includes: Obtain the basic device usage status of the target device from the edge nodes in the cloud platform, where one edge node corresponds to one target device; Construct a set of data acquisition periods and task execution counts, and denote the set of task execution counts collected within the i-th data acquisition period as . ,in, Let A represent the a-th task within the i-th data acquisition period, and let A represent the total number of tasks executed within the i-th data acquisition period.

3. A predictive fault maintenance system based on a cloud computing platform, executing the predictive fault maintenance method based on a cloud computing platform as described in any one of claims 1-2, characterized in that, The system includes: a data acquisition and collection construction module, a task density and health score calculation module, a sub-score calculation and comprehensive score calculation module, and a change rate calculation, analysis and early warning module. The data acquisition and collection construction module: acquires the basic device usage status of the target device from the edge nodes of the cloud platform; and constructs a data collection cycle and task execution count set. The task density and health score calculation module: constructs a set of task execution time points corresponding to each task, and calculates the task density of the target device within a single data acquisition cycle; based on the task density, it calculates the task density health sub-score of the target device within a single data acquisition cycle. The sub-score calculation and comprehensive score calculation module calculates the current health sub-score, voltage health sub-score, and temperature health sub-score of the target device within a single data acquisition cycle based on the basic usage status of the device, and calculates the comprehensive health score of the target device within a single data acquisition cycle by combining the task density health sub-score. The change rate calculation and analysis early warning module calculates the change rate of the health score of the target device within a single data acquisition cycle; analyzes and predicts the remaining operating time of the target device within a single data acquisition cycle; presets thresholds, analyzes and performs early warning maintenance.

4. The predictive fault maintenance system based on a cloud computing platform according to claim 3, characterized in that: The data acquisition and collection construction module includes a data acquisition unit and a collection construction unit; The data acquisition unit acquires the basic usage status of the target device from the edge nodes in the cloud platform. Each edge node corresponds to one target device. The basic usage status of the device includes the number of task executions, task execution time, task execution status data, and sensor data. The task execution status data includes the current data and voltage data of the device during task execution, and the sensor data includes the temperature data of the device during task execution. The set construction unit: constructs a set of data collection cycles and task execution counts.

5. A predictive fault maintenance system based on a cloud computing platform according to claim 4, characterized in that: The task density and health score calculation module includes a task density calculation unit and a health score calculation unit; The task density calculation unit: evenly divides the task execution time corresponding to the task into N task execution time points to construct a task execution time point set; Calculate the task density of the target device during the i-th data acquisition cycle; The health score calculation unit calculates the task density health sub-score for the i-th data acquisition cycle based on the task density of the target device within the i-th data acquisition cycle.

6. The predictive fault maintenance system based on a cloud computing platform according to claim 5, characterized in that: The sub-score calculation and comprehensive score calculation module includes a sub-score calculation unit and a comprehensive score calculation unit; The sub-score calculation unit calculates the current health sub-score, voltage health sub-score, and temperature health sub-score for the i-th data acquisition cycle based on the current data, voltage data, and temperature data at the task execution time point. The comprehensive scoring calculation unit performs weighted fusion of the task density health sub-score, current health sub-score, voltage health sub-score, and temperature health sub-score for the i-th data acquisition cycle to calculate the comprehensive health score of the target device within the i-th data acquisition cycle.

7. A predictive fault maintenance system based on a cloud computing platform according to claim 6, characterized in that: The rate of change calculation and analysis early warning module includes a rate of change calculation unit and an analysis early warning unit; The rate of change calculation unit: obtains the comprehensive health score of the target device within the (i-1)th data acquisition cycle, and calculates the rate of change of the health score of the target device within the i-th data acquisition cycle based on the comprehensive health score of the target device within the i-th data acquisition cycle and the cycle time length of the i-th data acquisition cycle; The analysis and early warning unit: based on the comprehensive health score, health score change rate and task density of the target device in the i-th data acquisition cycle, analyzes and predicts the remaining runnable time of the target device in the i-th data acquisition cycle; A preset threshold for the remaining operable time of the target device within the i-th data acquisition cycle is set. If the remaining operable time of the target device within the i-th data acquisition cycle is less than the threshold, it is determined that the target device has an abnormal lifespan health fault within the i-th data acquisition cycle. An early warning is then issued and relevant personnel are reminded to perform maintenance. The remaining uptime threshold of the target device within the i-th data collection cycle is uploaded to the cloud platform, and different devices are clustered based on the basic usage status of the devices, and the clustered devices are maintained in a unified manner. Let i = i + 1, iterate through the data acquisition cycle, and perform dynamic predictive maintenance of the equipment.

Images