Active maintenance method based on device life cycle management

Abstract

The invention discloses an active maintenance method based on device life cycle management, and relates to the technical field of electronic equipment management, and the method comprises the following steps: S1, obtaining the initial operation state data of a device, the normal operation state data of the device, and the overload operation state data of the device; s2, performing data model division according to the data acquired in the step S1, establishing a multi-modal state model, and establishing a state model parameter set; s3, performing detection through an anomaly detection function; s4, scoring the comprehensive health degree of the device through an active maintenance decision algorithm, and sending a maintenance instruction; and S5, actively maintaining a decision algorithm to be connected with the cloud to realize real-time data interaction. Through active maintenance, device faults can be effectively prevented, and the reliability of equipment is improved; through comprehensive health degree scoring, the use state of the device is scored, so that unnecessary maintenance can be reduced, the downtime of equipment failure can be reduced, and the production efficiency can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of electronic device management, and particularly to an active maintenance method based on device life cycle management. Background Art

[0002] With the continuous development of industrial technology, modern devices are developing towards being large-scale, complex, and automated, and the difficulty and cost of device maintenance are also increasing accordingly. The traditional device maintenance methods are mainly divided into two types: corrective maintenance and preventive maintenance. Corrective maintenance means repairing the device after a failure occurs. Although this method is simple and direct, it will cause the device to stop running, resulting in production losses and even possibly triggering safety accidents. Preventive maintenance means maintaining the device at predetermined time intervals. This method can prevent the occurrence of some failures, but there are problems of insufficient or excessive maintenance, resulting in increased maintenance costs. In order to overcome the deficiencies of traditional maintenance methods, predictive maintenance has emerged. Predictive maintenance monitors the operating status of the device, predicts the remaining service life of the device, and performs maintenance before the device fails, thereby avoiding device downtime and reducing maintenance costs.

[0003] A device is composed of many components, and each component has its own life cycle. The failure of a component often leads to the overall failure of the device. Therefore, managing the life cycle of key components inside the device and realizing the active maintenance of components are of great significance for improving device reliability and reducing maintenance costs.

[0004] However, existing maintenance technologies mostly adopt regular inspections or corrective maintenance after failures, lacking the ability to monitor and predict the status of components in real time, resulting in prominent problems of maintenance lag;

[0005] Traditional methods rely on single-parameter (such as temperature or current) threshold alarms and are difficult to reflect the comprehensive degradation trend of components. For example, the coupling effect of vibration frequency and current fluctuation may accelerate component failure, but there is a lack of multi-modal data fusion analysis means. Therefore, we propose an active maintenance method based on device life cycle management to solve the above-mentioned technical problems. Summary of the Invention

[0006] The purpose of the present invention is to provide an active maintenance method based on device life cycle management to solve the problems raised in the above background art.

[0007] To achieve the above purpose, the present invention adopts the following technical solutions: including the following steps:

[0008] S1. Obtain the initial operating status data of the component, obtain the normal operating status data of the component, and obtain the data of the component operating under overload;

[0009] S2. Based on the data obtained in step S1, perform data model partitioning, establish a multi-modal state model, and establish a state model parameter set;

[0010] S3. Detect the devices in abnormal operating states through an anomaly detection function;

[0011] S4. Score the comprehensive health of the devices through an active maintenance decision algorithm, and issue a maintenance instruction based on the score;

[0012] S5. The active maintenance decision algorithm connects to the cloud to achieve real-time data interaction.

[0013] As a preferred solution, in step S1, the operating state data includes the temperature, time, current, and vibration frequency during device operation.

[0014] As a preferred solution, in step S1, the temperature, current, and vibration frequency of the device are respectively detected by a temperature sensor, a current tester, and a speed sensor.

[0015] As a preferred solution, in step S2;

[0016] Partition the data by time: The initial operating state data of the device is the operating state data from 0 to 0.15 hours after the device starts. The normal operating state data is the operating state data from 0.5 hours to 2 hours after the device starts. The overloaded operating state data of the device is the operating state data after the device starts for more than 8 hours;

[0017] Partition the data by current: The initial operating data of the device is that the current during device operation is more than 10% lower than its rated current. The normal operating state data is that the current fluctuation during device operation is within ±2% of its rated current. The overloaded operating state data is that the current during device operation exceeds 10% of its rated current;

[0018] Partition the data by vibration frequency: The initial operating data of the device is combined with the data partitioned by time. When the device starts from 0 to 0.15 hours, the vibration frequency between 10 and 60 Hz is normal operating data. When the device starts from 0.5 hours to 2 hours, the vibration frequency less than 15 Hz is normal operating data. When the device starts for more than 8 hours, the vibration frequency less than 20 Hz is normal operating data. The vibration frequency includes the power frequency vibration of the device.

[0019] As a preferred solution, in step S2, partition the data model:

[0020] Define the device operating time domain: Let \(t\in R\) + be the continuous operating time (hours) of the device, then the data is partitioned as:

[0021] Current constraint condition (I rated is the rated current):

[0022]

[0023] Vibration frequency constraint (f is the measured vibration frequency):

[0024]

[0025] As a preferred solution, in the step S2, the state model parameter set:

[0026] Initial state model: Θ init ={α t , β I , γ f};

[0027] Normal state model: Θ norm ={μt, σ I 2 , λf};

[0028] Overload state model: Θ over ={φ t , ψ I , ω f};

[0029] Anomaly detection function: where ρt is the model prediction value, and χ 2 0.95 is the critical value of the chi-square distribution.

[0030] As a preferred solution, in the step S4, the proactive maintenance decision algorithm: Comprehensive health score:

[0031]

[0032] The weight coefficients satisfy:

[0033] W1 + W2 + W3 = 1

[0034] Maintenance departure condition:

[0035] where t represents the continuous operation time, in hours; I represents the real-time current, in amperes; f represents the main vibration frequency, in hertz; χ 2 0.95 is the critical value at a significance level of 5%.

[0036] As a preferred solution, in the step S4, for the device with abnormal data, data sorting and collection are performed, the device position is marked, and the located device position is fed back to the active maintenance algorithm system.

[0037] As a preferred solution, in the step S2, a memory is further included for storing the operation data of the device.

[0038] Compared with the prior art, the present invention has obvious advantages and beneficial effects. Specifically, as can be seen from the above technical solutions, the main ones are:

[0039] Through active maintenance, the present invention can effectively prevent device failures and improve the reliability of the equipment;

[0040] By comprehensively scoring the health status, the usage status of the device can be scored, unnecessary maintenance can be reduced, and the maintenance cost can be lowered;

[0041] Moreover, it can also reduce the equipment failure downtime and improve the production efficiency.

[0042] To more clearly illustrate the structural features and functions of the present invention, the following combines the drawings with specific embodiments to detail the present invention. Description of the Drawings

[0043] Figure 1 It is a flowchart of the active maintenance method of the embodiment of the present invention. Detailed Embodiments

[0044] In order to make the purpose, technical solutions and advantages of the present invention clearer, the following further details the present invention in conjunction with the drawings and implementation examples. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0045] It should be noted that when an element is referred to as being "fixed to" another element, it can be directly on the other element or there can also be an intermediate element. When an element is considered to be "connected" to another element, it can be directly connected to the other element or there may be an intermediate element at the same time. The terms "vertical", "horizontal", "left", "right" and similar expressions used herein are only for the purpose of illustration.

[0046] Please refer to Figure 1 , the embodiment of the present invention provides an active maintenance method based on device life cycle management, including the following steps:

[0047] S1. Obtain the initial operation state data of the device, obtain the normal operation state data of the device, and obtain the data of the device in overloaded operation;

[0048] The collection of this data is implemented by detecting the temperature, current, and vibration frequency of the device through a temperature sensor, a current tester, and a speed sensor respectively;

[0049] S2. According to the data obtained in step S1, perform data model partitioning, establish a multi-modal state model, and establish a state model parameter set;

[0050] S3. Detect the device in an abnormal operating state through an anomaly detection function;

[0051] S4. Score the comprehensive health of the device through an active maintenance decision algorithm and issue a maintenance instruction according to the score;

[0052] S5. The active maintenance decision algorithm is connected to the cloud to achieve real-time data interaction;

[0053] In step S5, the model parameter Θ is updated dynamically by online learning and is updated according to the following formula:

[0054] μ new = μ old + η(x new - μ old )

[0055] σ 2 new = σ 2 old + η((x new - μ old ) 2 - σ 2 old )

[0056] In the above steps, the data is divided into three categories. Specifically, one category is to divide the data according to time: the initial operating state data of the device is the operating state data of the device from startup to 0.15 hours, the normal operating state data is the operating state data of the device from startup to 2 hours, and the overloaded operating state data of the device is the operating state data of the device after startup for more than 8 hours;

[0057] The second category is to divide the data according to current: the initial operating data of the device is that the current during device operation is more than 10% lower than its rated current, the normal operating state data is that the current fluctuation during device operation is within ±2% of its rated current, and the overloaded operating state data is that the current during device operation exceeds 10% of its rated current;

[0058] Three categories, data divided according to vibration frequency: The initial operation data of the device is combined with the data divided according to time. When the device starts up for 0 to 0.15 hours, the vibration frequency between 10 and 60 Hz is normal operation data. When the device starts up for 0.5 hours to 2 hours, the vibration frequency less than 15 Hz is normal operation data. When the device starts up for more than 8 hours, the vibration frequency less than 20 Hz is normal operation data. The vibration frequency includes the power frequency vibration of the device;

[0059] After the data division is completed, the data is processed by the following formula: Define the device operation time domain: Let t ∈ R + be the continuous operation time (hours) of the device, then the data is divided into:

[0060]

[0061] Current constraint condition (I rated is the rated current):

[0062]

[0063] Vibration frequency constraint (f is the measured vibration frequency):

[0064]

[0065] And establish a state model parameter set: Initial state model: Θ init ={α t ,β I ,γ f};

[0066] Normal state model: Θ norm ={μt,σ I 2 ,λf};

[0067] Overload state model: Θ over ={φ t ,ψ I ,ω f}; After the parameters are established, the collected data of the device is detected by the anomaly detection function, so as to realize the active detection and maintenance of the device, which can help users discover the existence of device problems earlier and provide guarantee for the use of the device. The anomaly detection function is as follows:

[0068] where ρt is the model prediction value, χ 2 0.95 is the chi-square distribution critical value.

[0069] Please refer to Figure 1, in the step S1, the temperature, current and vibration frequency of the device are respectively detected by a temperature sensor, a current tester and a speed sensor. The temperature sensor, the current tester and the speed sensor are all existing mature technologies and will not be elaborated further in this article.

[0070] Please refer to Figure 1 , in the step S2;

[0071] Divide the data according to time: The initial operating state data of the device is the operating state data from 0 to 0.15 hours after the device starts. The normal operating state data is the operating state data from 0.5 hours to 2 hours after the device starts. The overloaded operating state data of the device is the operating state data after the device starts for more than 8 hours;

[0072] Divide the data according to current: The initial operating data of the device is that the current during the device operation is more than 10% lower than its rated current. The normal operating state data is that the current fluctuation during the device operation is within ±2% of its rated current. The overloaded operating state data is that the current during the device operation exceeds 10% of its rated current;

[0073] Divide the data according to vibration frequency: The initial operating data of the device is combined with the data divided according to time. When the device starts from 0 to 0.15 hours, the vibration frequency between 10 and 60 Hz is normal operating data. When the device starts from 0.5 hours to 2 hours, the vibration frequency less than 15 Hz is normal operating data. When the device starts for more than 8 hours, the vibration frequency less than 20 Hz is normal operating data. The vibration frequency includes the power frequency vibration of the device.

[0074] Please refer to Figure 1 , in the step S4, the active maintenance decision algorithm: comprehensive health score:

[0075]

[0076] The weight coefficients satisfy:

[0077] W1 + W2 + W3 = 1

[0078] Maintenance departure condition:

[0079] where t represents the continuous operation time, in hours; I represents the real-time current, in amperes; f represents the main vibration frequency, in hertz; χ 2 0.95 The critical value at a significance level of 5%.

[0080] Please refer to Figure 1In step S4, for the device with abnormal data, data is sorted and collected, the device position is marked, and the device position is located and fed back to the active maintenance algorithm system;

[0081] The original abnormal data collected by the sensor (such as sudden temperature increase, current fluctuation, and vibration exceeding the limit) is organized in a unified format, noise data is eliminated, and missing values ​​are filled in. The storage modules are divided according to the abnormality type (such as electrical failure, mechanical wear) to facilitate subsequent traceability and analysis. The installation location of the abnormal device is recorded through the equipment code or QR code. For example, "Cabinet A-Module 3-Slot 5" is synchronously updated in the maintenance system, and the coordinates of the faulty device in the virtual device are highlighted. The abnormal data packet (including timestamp, fault parameters, and location mark) is uploaded to the cloud maintenance platform in real time, triggering the algorithm update. A maintenance task queue is generated according to the severity of the abnormality (such as current overload 10% vs. temperature exceeding 5°C), and emergency tasks are automatically queued for processing.

[0082] See also Figure 1 In the step 2, a memory is also included for storing the operating data of the device.

[0083] Working process, sensor configuration: temperature sensor, current tester and vibration sensor are used to collect parameters such as temperature (T), current (I), vibration frequency (V) during device operation in real time;

[0084] Initial state: transient data within 0-0.15 hours after the device is started (such as cold start current impact, component running-in vibration), steady-state data in the stable operation stage (0.5-2 hours) (such as current fluctuation ±2%, vibration frequency <15Hz), overload state: performance degradation data after continuous operation for ≥8 hours (such as current exceeding the limit by 10%, temperature rise rate accelerating), so as to realize the monitoring of the device under various operating time and respective operating status;

[0085] Time dimension: define the state by segment according to the running time, such as initial (0-0.15h), normal (0.5-2h), overload (≥8h); current dimension: divide the low load (<90%), normal (±2%), and overload (>110%) intervals based on the rated current; vibration dimension: dynamically adjust the threshold, allowing 10-60Hz in the initial stage, limiting <15Hz in the normal stage, and ≤20Hz in the overload stage; parameter set construction: extract statistical features such as mean (μ), standard deviation (σ), extreme value, etc. in each state to form a multi-modal parameter library;

[0086] The Mahalanobis distance or similarity calculation (such as cosine similarity) is used to quantify the degree of deviation between the real-time data and the normal state, and a dynamic threshold is set: if the deviation value exceeds the critical value of the chi-square distribution (such as the confidence level of 95% corresponding to the chi-square distribution), the 2= 7.815), it is determined as abnormal. When the current suddenly increases by > 10% and the temperature rises by > 5°C / min, an overload warning is triggered. When the vibration frequency exceeds the threshold and lasts for more than 10 minutes, a mechanical failure risk is prompted;

[0087] Loss degree calculation: According to the parameter cumulative deviation and operation duration, quantify the independent losses of temperature, current, and vibration, and score the state health degree of the device according to the health degree scoring formula. According to the score result, trigger the maintenance logic of this method. When the H value is between 0.6 and 0.8, including 0.6 and 0.8, a warning prompt is given. At this time, device lubrication and maintenance can be carried out. When the H value is less than or equal to 0.6, stop the machine immediately. When the H value is greater than 0.6, continuously detect the device;

[0088] Upload the device status data to the cloud through the Wi-Fi / 5G module, support cross-device data sharing and remote monitoring, and dynamically update the state model parameters (such as adjusting the vibration frequency threshold) and health degree weights by using historical data in the cloud to improve the adaptability of the algorithm.

[0089] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An active maintenance method based on device life cycle management, characterized by: The following steps are involved: S1, obtaining the initial operation state data of the device, obtaining the normal operation state data of the device, and obtaining the overload operation state data of the device; S2. According to the data obtained in step S1, data model division is performed, a multimodal state model is established, and a state model parameter set is established; S3, detecting the device in abnormal operation state through an abnormal detection function; S4. Score the comprehensive health of the device through the active maintenance decision algorithm and issue maintenance instructions based on the score; S5. Actively maintain the decision-making algorithm to connect to the cloud to achieve real-time data interaction.

2. The active maintenance method based on device lifecycle management according to claim 1, characterized in that: In the step S1, the operating status data includes the temperature, time, current and vibration frequency of the device during operation.

3. The active maintenance method based on device lifecycle management according to claim 2, characterized in that: In step S1 , the temperature, current and vibration frequency of the device are detected respectively by a temperature sensor, a current tester and a speed sensor.

4. The active maintenance method based on device lifecycle management according to claim 3, characterized in that: In the step S2; Divide the data by time: the initial operation status data of the device refers to the operation status data of the device from 0 to 0.15 hours after startup, the normal operation status data refers to the operation status data of the device from 0.5 to 2 hours after startup, and the overload operation status data of the device refers to the operation status data of the device for more than 8 hours after startup; Data divided by current: the initial operation data of the device refers to the current when the device is running being lower than 10% of its rated current; the normal operation data refers to the current fluctuation of the device when the device is running being within ±2% of its rated current; the overload operation data refers to the current when the device is running being higher than 10% of its rated current; Data divided by vibration frequency: The initial operating data of the device is combined with the data divided by time. When the device is started for 0 to 0.15 hours, the vibration frequency is 10 to 60 Hz, which is normal operating data. When the device is started for 0.5 to 2 hours, the vibration frequency is less than 15 Hz, which is normal operating data. When the device is started for more than 8 hours, the vibration frequency is less than 20 Hz, which is normal operating data. The vibration frequency includes the power supply frequency vibration of the device.

5. The active maintenance method based on device lifecycle management according to claim 1, characterized in that: In step S2, the data model is divided: Define the device operation time domain: Let t∈R + is the continuous operation time of the device (hours), the data is divided into: Current constraint (I rated is the rated current): Vibration frequency constraint (f is the measured vibration frequency):

6. The active maintenance method based on device lifecycle management according to claim 1, characterized in that: In step S2, the state model parameter set: Initial state model: Θ init ={α t ,β I ,γ f }; Normal state model: Θ norm = {μt,σ I 2 ,λf}; Overload state model: Θ over ={φ t , ψ I ,ω f }; Anomaly detection function: Where ρt is the model prediction value, χ 2 0.95 is the critical value of the chi-square distribution.

7. The active maintenance method based on device lifecycle management according to claim 1, characterized in that: In step S4, the active maintenance decision algorithm: comprehensive health score: The weight coefficient satisfies: W1+W2+W3=1 Maintenance departure conditions: Where, t represents the continuous running time in hours; I represents the real-time current in amperes; f represents the main frequency of vibration in hertz; 2 0.95 The critical value of the significance level is 5%.

8. The active maintenance method based on device lifecycle management according to claim 1, characterized in that: In step S4, data of the device with abnormal data is sorted and collected, the device position is marked, and the device position is located and fed back to the active maintenance algorithm system.

9. The active maintenance method based on device lifecycle management according to claim 1, characterized in that: In the step S2, a memory is also included for storing the operation data of the device.

Images