CN113298278B - Power equipment with self-health state prediction function, self-health state prediction method and cloud server - Google Patents

Abstract

The application provides power equipment with a self-health state prediction function, a self-health state prediction method thereof and a cloud server suitable for a plurality of power equipment. The power equipment obtains a health index prediction model and a health index threshold value of the power equipment from the cloud server through the communication module of the power equipment, and is used for carrying the obtained plurality of sensing data and basic data of the power equipment into the health index prediction model to operate so as to obtain a health index. The power equipment is also used for obtaining health state prediction data of the power equipment according to the comparison result of the health prediction index and the health prediction index threshold value so as to execute subsequent processing. The health index prediction model is established and trained by the cloud server through a machine learning mode.

Description

Power equipment with self-health state prediction function and its self-health state prediction Testing methods and cloud servers

Technical field

The present invention relates to the technical field of power supply, and in particular to a power equipment with a self-health state prediction function, a self-health state prediction method thereof, and a cloud server suitable for multiple power equipment.

Background technique

A power device, such as an uninterruptible power system (UPS), a power distribution unit (PDU) or an automatic transfer switch (ATS), is used to provide an operating power supply to at least one load , so that these loads can operate normally.

However, once the power device fails (for example, due to damage to its internal parts), it is likely that the power device will not be able to supply power to the loads normally, and thus the loads will not be able to operate normally. At this time, if important loads (such as critical medical equipment) cannot operate normally, the consequences will be unimaginable. Therefore, if the health status of the power device can be predicted, maintenance personnel can take preventive measures when the health status of the power device is not good, so that the aforementioned problems can be effectively prevented.

Contents of the invention

One of the objects of the present invention is to provide an electric power equipment with a self-health state prediction function.

Another object of the present invention is to provide a method for predicting the self-health state of electric power equipment.

Another object of the present invention is to provide a cloud server suitable for multiple power equipment.

In order to achieve the above object, the present invention provides a power equipment with a self-health state prediction function, which includes a plurality of sensors, a communication module, a control unit and a storage unit. The multiple sensors are used to obtain multiple sensing data. The control unit is used to obtain a health index prediction model and a health prediction index threshold of the power equipment from a cloud server through the communication module, and is used to bring the sensing data and basic data of the power equipment into the health index prediction. The model is used to perform calculations to obtain a health prediction index, and obtain a health status prediction data of the power equipment based on the comparison result between the health prediction index and the health prediction index threshold in order to perform a subsequent process, wherein the health index prediction model is as described Cloud servers are built and trained through machine learning. As for the storage unit, it is used to store basic data of power equipment, health index prediction models and health prediction index thresholds.

In order to achieve the above object, the present invention also provides a method for predicting the self-health state of electric power equipment. The electric power equipment includes a plurality of sensors and a communication module, and the method includes the following steps: from a cloud through the communication module The server obtains a health index prediction model and a health prediction index threshold of the power equipment, where the health index prediction model is established and trained by the cloud server through machine learning; a plurality of sensing data are obtained through the sensors; Bring the sensing data and the basic data of the power equipment into the health index prediction model for calculation to obtain a health prediction index; and obtain a health status of the power equipment based on the comparison result between the health prediction index and the health prediction index threshold. Predict data in order to perform a subsequent process.

In order to achieve the above object, the present invention further provides a cloud server, which is suitable for multiple power equipment. The cloud server includes a communication module, a database, a prediction model training module and a data collection module. The database is used to store basic data, sensing data, health status prediction data, abnormal event data, life data and maintenance data of each power equipment, and is used to store a health prediction corresponding to an equipment model of each power equipment. Exponential threshold. The prediction model training module is used to establish and train a health index prediction model for each equipment model. Each health index prediction model is based on the basic data and sensing of the power equipment with a corresponding equipment model in the database. Data, health status prediction data, abnormal event data, life data and maintenance data are created and trained through machine learning. As for the data collection module, it is used to collect basic data, sensing data, health status prediction data, and abnormal event data of each power equipment through the communication module, and obtain corresponding life data based on the basic data of each power equipment. . This data collection module is further used to provide a network user interface to collect the maintenance data of each electrical equipment and the health prediction index thresholds through the network user interface, and combine all the collected data, the health prediction index thresholds with The obtained life data are stored in the database.

In order to make the above purposes, technical features and gains after actual implementation more obvious and easy to understand, a more detailed description will be given below with the help of better implementation examples and corresponding related drawings.

Description of drawings

The drawings described here are used to provide a further understanding of the present invention and constitute a part of this application. The illustrative embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the attached picture:

Figure 1 depicts a plurality of power equipment according to an embodiment of the present invention and a cloud server according to an embodiment of the present invention.

FIG. 2 illustrates the internal architecture of each power device of FIG. 1 .

Figure 3 shows the establishment process of each health index prediction model.

Figure 4 shows the training process of each health index prediction model.

Figure 5 is a flowchart of a method for predicting the self-health state of electric power equipment according to an embodiment of the present invention.

Detailed ways

In order to have a clearer understanding of the characteristics, content, advantages and effects of the present invention, the present invention is described in detail below in conjunction with the accompanying drawings and in the form of embodiments. The drawings used are only for their main purpose. They are for illustration and auxiliary description purposes, and may not represent the actual proportions and precise configurations after implementation of the present invention. Therefore, the proportions and configuration relationships of the attached drawings should not be interpreted to limit the scope of rights of the present invention in actual implementation.

The advantages, features and technical methods achieved by the present invention will be described in more detail with reference to the exemplary embodiments and accompanying drawings to be more easily understood, and the present invention may be implemented in different forms, so it should not be understood to be limited thereto. The embodiments set forth herein are, on the contrary, provided embodiments that will make this disclosure more thorough and complete and fully convey the scope of the present invention to those of ordinary skill in the art, and the present invention will only be appended. defined by the scope of the patent application.

FIG. 1 depicts a plurality of power equipment according to an embodiment of the present invention and a cloud server according to an embodiment of the present invention, and FIG. 2 illustrates the internal architecture of each power equipment in FIG. 1 . Please refer to Figure 1 and Figure 2 at the same time. In this example, the cloud server 110 includes a prediction model training module 112, a database 114, a data collection module 116, a warning message push module 118 and a communication module 120, and each power equipment 130 It includes a control unit 132, a storage unit 134, a communication module 136, a plurality of sensors 138 and an alarm module 140. For convenience of explanation, the following description takes the example that these power equipment 130 are all uninterruptible power systems. These power equipment 130 may be all online uninterruptible power systems (On-line UPS), all off-line uninterruptible power systems (Off-line UPS), or all online interactive uninterruptible power systems (Line-interactive UPS). Or it is composed of a mixture of at least two of the above three uninterruptible power systems. In addition, FIG. 2 only illustrates the parts of each power equipment 130 related to the present invention in order to focus on the technical content of the present invention.

First, the operation of the power equipment 130 will be described. Referring to FIG. 2 , the control unit 132 of the power equipment 130 is used to obtain a plurality of sensing data through the sensors 138 . The method of obtaining the data may be a timed manner or an untimed manner. The obtained sensing data includes at least one of the input voltage, input current, output voltage, output current, battery voltage, battery charging current, battery discharging current, ambient temperature, and ambient humidity of the power device 130 . After obtaining these sensing data, the control unit 132 will determine whether an abnormal event has occurred (such as the input voltage is too low, the battery voltage is too low, the battery internal resistance is too high, the ambient temperature is too high, etc.) to obtain the abnormal event. data, and store the obtained sensing data and the obtained abnormal event data into the storage unit 134 . The storage unit 134 also stores basic data of the power equipment 130. The basic data includes at least one of the equipment model, manufacturing date, rated power, rated voltage and rated current of the power equipment 130.

The control unit 132 is also used to obtain a health index prediction model and a health prediction index threshold of the power equipment 130 from the cloud server 110 through the communication module 136, and store the obtained health index prediction model and health prediction index threshold in the storage. In unit 134, the health index prediction model is established and trained by the cloud server 110 through machine learning (details will be described later). After obtaining the health index prediction model and the health prediction index threshold, the control unit 132 brings the obtained sensing data and the basic data of the power equipment 130 into the health index prediction model for calculation to obtain a health prediction index ( For example, it is the probability that the power equipment 130 will fail within a predetermined time in the future), and a health status prediction data of the power equipment 130 is obtained based on the comparison result between the health prediction index and the health prediction index threshold, and the obtained health status prediction data is obtained. The state prediction data is also stored in the storage unit 134 .

In this example, the control unit 132 includes a health index prediction module 132-1 and a health status analysis module 132-2. The health index prediction module 132-1 is used to bring the obtained sensing data and the basic data of the power equipment 130 into the health index prediction model to perform calculations to obtain the above-mentioned health prediction index. The health status analysis module 132-2 is used to compare the obtained health prediction index with the health prediction index threshold, thereby obtaining the above-mentioned health status prediction data.

After obtaining the health state prediction data of the power equipment 130, the control unit 132 performs a subsequent process accordingly. For example, when the obtained health state prediction data shows that the health state of the power equipment 130 is poor, the subsequent processing may include notifying a mail server (not shown) through the communication module 136 to send an alarm letter, Notify a short message transmitter (not shown) to send an alarm message through the communication module 136 and control at least one of the alarm module 140 to send an alarm, so that the maintenance personnel can first perform preventive treatment, such as replacing the soon to be damaged. Component. Of course, regardless of the health status of the power equipment 130 represented by the obtained health status prediction data, the subsequent processing may also include transmitting the obtained health status prediction data to the cloud server 110 through the communication module 136 so that the cloud server The server 110 performs a corresponding process (described in detail later) based on the health status of the power equipment 130 presented by the health status prediction data.

In addition, in this example, the alarm module 140 includes at least one of a display device (not shown) and an audio warning device (not shown) to issue corresponding warning messages according to actual needs. In addition, in this example, the control unit 132 is further used to obtain an alarm execution script from the cloud server 110 through the communication module 136 . The alarm execution script has an alarm action and at least one judgment condition set by a user. The judgment conditions are, for example, whether the battery voltage is lower than a first default value, whether the output voltage of the power device 130 is lower than a second default value, and so on. When the control unit 132 determines that the conditions set in the alarm execution script are all met, it controls the alarm module 140 to execute the above-mentioned alarm action. The alarm action is, for example, controlling the display device in the alarm module 140 to emit a red flashing light, or controlling the sound warning device in the alarm module 140 to emit a series of alarm sounds, or executing the above two alarm methods at the same time.

Next, the operation of the cloud server 110 will be described. Please refer to FIG. 1 again. In this example, the database 114 of the cloud server 110 is used to store basic data, sensing data, health state prediction data, abnormal event data, life data and maintenance data of each power equipment 130, and is used to store A health prediction index threshold corresponding to an equipment model of each power equipment 130 is stored. For example, if the power equipment 130 in FIG. 1 has two equipment models in total, the database 114 will store two health prediction index thresholds corresponding to the two equipment models.

The prediction model training module 112 is used to establish and train a health index prediction model for each equipment model. Each health index prediction model is based on the basic data and sensory data of the power equipment 130 with a corresponding equipment model in the database 114. Measurement data, health status prediction data, abnormal event data, life data and maintenance data are established and trained through machine learning. In this example, the machine learning algorithms used by the cloud server 110 include artificial neural network algorithm, decision tree algorithm, K-means algorithm, and support vector machine algorithm. At least one of vector machine algorithm, linear regression algorithm and logistic regression algorithm.

The data collection module 116 is used to collect basic data, sensing data, health status prediction data, and abnormal event data of each power equipment 130 through the communication module 120, and obtain a corresponding life span of each power equipment 130 based on the basic data. data. In addition, the data collection module 116 is further configured to provide a web-based user interface 116-1 to collect the maintenance data of each power equipment 130 and the health prediction index thresholds through the web-based user interface 116-1, and All collected data, the health prediction index thresholds and the obtained life span data are stored in the database 114 .

Next, the establishment method of each health index prediction model will be further explained. Figure 3 shows the establishment process of each health index prediction model. Please refer to Figure 3 and Figure 1 at the same time. Before starting to establish the health index prediction model, the database 114 needs to pre-store the basic data, sensing data, health status prediction data, and abnormality of at least part of the power equipment 130 corresponding to at least one equipment model. Event data, life data and maintenance data, and the pre-stored data must reach a certain amount. In addition, part of the pre-stored data can be given by the user, such as health status prediction data. The cloud server 110 establishes the required health index prediction model based on the existing data pre-stored in its database 114 .

When starting to build a health index prediction model, the prediction model training module 112 will obtain the basic data, sensing data, health status prediction data, abnormal event data, and the power equipment 130 with a corresponding equipment model from the database 114. The life data and maintenance data are used to establish a failure rate prediction model through machine learning (as shown in step S310). Next, the prediction model training module 112 brings the obtained basic data, sensing data, health state prediction data, abnormal event data, life data and maintenance data of each power equipment 130 into the above-mentioned fault occurrence rate prediction model to obtain the aforementioned A predicted fault occurrence rate of each power equipment 130 (as shown in step S312).

Next, the prediction model training module 112 uses machine learning based on the obtained basic data, sensing data, health state prediction data, abnormal event data, life data, maintenance data and fault occurrence rate prediction values of the power equipment 130 to establish a remaining life prediction model (as shown in step S314). Then, the prediction model training module 112 brings the obtained basic data, sensing data, health state prediction data, abnormal event data, life data, maintenance data and failure rate prediction value of each power equipment 130 into the remaining life prediction. model to obtain a remaining life prediction value of each of the aforementioned power equipment 130 (as shown in step S316). Next, the prediction model training module 112 obtains basic data, sensing data, health state prediction data, abnormal event data, life data, maintenance data, fault occurrence rate prediction values and remaining life prediction values of the power equipment 130 A health index prediction model is established through machine learning (as shown in step S318).

Next, the training method of each health index prediction model will be further explained. After establishing the required health index prediction models, the cloud server 110 will transmit the established health index prediction models to the corresponding power equipment 130 through the communication module 120 so that the power equipment 130 can generate new health status predictions. data. Then, the cloud server 110 will collect the basic data, sensing data, health status prediction data, abnormal event data and maintenance data of the power equipment 130 corresponding to the health index prediction models through the data collection module 116, and based on the aforementioned The basic data of each power equipment 130 is obtained to obtain corresponding life data, and the newly collected data and the newly obtained life data are stored in the database 114 . In this way, the updated database 114 stores original pre-stored data, newly collected data, and newly obtained life data. The cloud server 110 trains the aforementioned health index prediction models based on the original pre-stored data and newly stored data in the database 114 .

Figure 4 shows the training process of each health index prediction model. Please refer to Figure 4 and Figure 1 at the same time. When starting to train one of the health index prediction models, the prediction model training module 112 will obtain the basic data of the power equipment 130 with the corresponding equipment model from the database 114 after the storage content is updated. , sensing data, health status prediction data, abnormal event data, life data and maintenance data, and use machine learning to train the corresponding failure rate prediction model (as shown in step S410). Next, the prediction model training module 112 brings the obtained basic data, sensing data, health status prediction data, abnormal event data, life data and maintenance data of each power equipment 130 into the trained fault incidence prediction model to Obtain a fault occurrence rate prediction value of each of the aforementioned power equipment 130 (as shown in step S412).

Next, the prediction model training module 112 uses machine learning based on the obtained basic data, sensing data, health state prediction data, abnormal event data, life data, maintenance data and fault occurrence rate prediction values of the power equipment 130 to train the corresponding remaining life prediction model (as shown in step S414). Then, the prediction model training module 112 brings the obtained basic data, sensing data, health state prediction data, abnormal event data, life data, maintenance data and fault occurrence rate prediction values of each power equipment 130 into the remaining training data. The life prediction model is used to obtain a remaining life prediction value of each of the aforementioned power equipment 130 (as shown in step S416). Next, the prediction model training module 112 obtains basic data, sensing data, health state prediction data, abnormal event data, life data, maintenance data, fault occurrence rate prediction values and remaining life prediction values of the power equipment 130 The corresponding health index prediction model is trained through machine learning (as shown in step S418).

After training the required health index prediction models, the cloud server 110 will transmit the trained health index prediction models to the corresponding power equipment 130 through the communication module 120 for use by the power equipment 130 . Of course, the aforementioned training process can be executed again regularly or irregularly to retrain these health index prediction models. As the number of training times increases, the accuracy of health status prediction data will also increase.

Please refer to FIG. 1 again. In this example, when the data collection module 116 determines that any newly obtained health state prediction data shows that the health state of the corresponding power equipment 130 is poor, the data collection module 116 will send a warning through a warning. The message push module 118 pushes a warning message to the corresponding user's smart handheld device. The smart handheld device is, for example, a mobile phone, a notebook computer or a tablet computer. In addition, in this example, the data collection module 116 further receives an alarm action and at least one judgment condition set by the user through the network user interface 116-1, thereby generating an alarm execution script, and transmits the alarm through the communication module 120. The alarm execution script is sent to at least one corresponding power device 130 .

Although in the foregoing embodiments, the cloud server 110 has the warning message push module 118, and each power device 130 has the alarm module 140, this is not intended to limit the present invention. Those of ordinary skill in the art should know that the warning message push module The module 118 and the alarm module 140 can be used according to actual design requirements. In addition, although in the foregoing embodiments, it is taken as an example that all the power equipment 130 are uninterruptible power systems, this is not intended to limit the present invention. Those of ordinary skill in the art should know that the above-mentioned power equipment 130 can also be power supplies. The distribution units are either all automatic power switching switches, or are composed of a mixture of at least two of the above three types of power equipment. Of course, if the power equipment 130 is implemented by a power distribution unit or an automatic power switch, its basic data, sensing data, health state prediction data, abnormal event data, life data, maintenance data and corresponding health prediction index thresholds The content needs to be adjusted accordingly, and the factors considered in the corresponding health index prediction model, failure incidence prediction model and remaining life prediction model also need to be adjusted accordingly.

In addition, through the above description, a person with ordinary knowledge in the art should be able to summarize some basic operation steps of the self-health state prediction method of the power equipment of the present invention, which are illustrated in FIG. 5 . Figure 5 is a flowchart of a method for predicting the self-health state of electric power equipment according to an embodiment of the present invention. The power equipment includes a plurality of sensors and a communication module, and the method includes the following steps: first, obtain a health index prediction model and a health prediction index threshold of the power equipment from a cloud server through the communication module (such as Shown in step S510), the health index prediction model is established and trained by the cloud server through machine learning. Then, a plurality of sensing data are obtained through the sensors (as shown in step S512). Next, the sensing data and the basic data of the power equipment are brought into the health index prediction model for calculation to obtain a health prediction index (as shown in step S514). Then, a health state prediction data of the power equipment is obtained according to the comparison result between the health prediction index and the health prediction index threshold, so as to perform a subsequent process (as shown in step S516). Of course, the aforementioned steps S510 and S512 can be reversed.

To sum up, since the power equipment of the present invention can perform the self-health state prediction function and can perform corresponding follow-up processing when the health state prediction data shows that its health state is not good, the maintenance personnel can perform the self-health state prediction function when the health state of the power device is not good. When possible, take preventive measures first. In addition, since the cloud server of the present invention can continuously retrain the health index prediction model required for power equipment, the accuracy of health status prediction data will also continue to improve.

The above-described embodiments only illustrate the technical ideas and characteristics of the present invention. Their purpose is to enable those skilled in the art to understand the content of the present invention and implement it accordingly. They cannot be used to limit the patent scope of the present invention, that is, Any equivalent changes or modifications made in accordance with the spirit disclosed in the present invention shall still be covered by the patent scope of the present invention.

Claims (31)

1. A power device having a self-health status prediction function, the power device comprising:

a plurality of sensors for acquiring a plurality of sensing data;

a communication module;

the control unit is used for obtaining a health index prediction model and a health index threshold value of the power equipment from a cloud server through the communication module, carrying the sensing data and basic data of the power equipment into the health index prediction model to operate so as to obtain a health prediction index, and obtaining health state prediction data of the power equipment according to a comparison result of the health prediction index and the health index threshold value so as to execute subsequent processing, wherein the health index prediction model is established and trained by the cloud server through a machine learning mode; and

a storage unit for storing the basic data of the power equipment, the health index prediction model and the health index threshold value,

the cloud server is used for establishing and training a health index prediction model for each device model of a plurality of power devices, and the steps executed when establishing each health index prediction model comprise:

obtaining basic data, sensing data, health state prediction data, abnormal event data, service life data and maintenance data of the electric power equipment with a corresponding equipment model from a database, and establishing a fault occurrence rate prediction model according to the basic data, the sensing data, the health state prediction data, the abnormal event data, the service life data and the maintenance data through a machine learning mode;

the obtained basic data, sensing data, health state prediction data, abnormal event data, life data and maintenance data of each power equipment are brought into the fault occurrence rate prediction model so as to obtain a fault occurrence rate prediction value of each power equipment;

establishing a residual life prediction model through a machine learning mode according to the acquired basic data, sensing data, health state prediction data, abnormal event data, life data, maintenance data and failure occurrence rate prediction values of the power equipment;

bringing the obtained basic data, sensing data, health state prediction data, abnormal event data, life data, maintenance data and failure occurrence rate prediction value of each power equipment into the residual life prediction model to obtain a residual life prediction value of each power equipment; and

a health index prediction model is established through a machine learning mode according to the obtained basic data, sensing data, health state prediction data, abnormal event data, life data, maintenance data, fault occurrence rate prediction value and residual life prediction value of the power equipment.

2. The power device with a self-health status prediction function according to claim 1, wherein the control unit includes:

the health index prediction module is used for carrying the sensing data and the basic data of the power equipment into the health index prediction model to perform operation so as to obtain the health index; and

and the health state analysis module is used for comparing the health prediction index with the health prediction index threshold value so as to obtain the health state prediction data.

3. The power device with a self-health status prediction function according to claim 1, wherein the basic data of the power device includes at least one of a device model number, a manufacturing date, a rated power, a rated voltage, and a rated current.

4. The power device with self-health prediction according to claim 1, wherein the sensed data includes at least one of an input voltage, an input current, an output voltage, an output current, a battery voltage, a battery charging current, a battery discharging current, an ambient temperature, and an ambient humidity.

5. The power device with self-health status prediction function according to claim 1, wherein the subsequent processing includes transmitting the health status prediction data to the cloud server via the communication module.

6. The power device with self-health prediction according to claim 1, wherein the subsequent processing includes notifying a mail server via the communication module to issue an alert letter when the health prediction data indicates that the health of the power device is poor.

7. The power device with self-health status prediction according to claim 1, wherein the subsequent processing includes notifying a short message sender to send an alarm short message through the communication module when the health status prediction data indicates that the health status of the power device is not good.

8. The power device with self-health status prediction function according to claim 1, further comprising an alarm module, wherein the control unit obtains an alarm execution script from the cloud server through the communication module, the alarm execution script has an alarm action set by a user and at least one judgment condition, and when the control unit judges that the conditions are satisfied, the control unit controls the alarm module to execute the alarm action.

9. The power device with self-health status prediction function according to claim 8, wherein the alarm module comprises at least one of a display device and an audible alert device.

10. The power device with self-health status prediction function according to claim 8, wherein the control unit further controls the alarm module to issue an alarm when the health status prediction data indicates that the health status of the power device is poor.

11. The power device with self-health prediction according to claim 1, wherein the machine learning algorithm used by the cloud server comprises at least one of a neural network-like algorithm, a decision tree algorithm, a K-average algorithm, a support vector machine algorithm, a linear regression algorithm, and a logistic regression algorithm.

12. The power device with self-health status prediction function according to claim 1, wherein the power device comprises an uninterruptible power system, a power distribution unit or an automatic power switch.

13. The power device with self-health status prediction function as in claim 12, wherein the uninterruptible power system comprises an online uninterruptible power system, an offline uninterruptible power system, or an online interactive uninterruptible power system.

14. A method for predicting a self-health status of an electrical device, wherein the electrical device comprises a plurality of sensors and a communication module, the method comprising:

acquiring a health index prediction model and a health index threshold value of the power equipment from a cloud server through the communication module, wherein the health index prediction model is established and trained by the cloud server through a machine learning mode;

acquiring a plurality of sensing data through the sensors;

carrying the sensing data and the basic data of the power equipment into the health index prediction model to perform operation so as to obtain a health prediction index; and

obtaining health state prediction data of the power equipment according to the comparison result of the health prediction index and the health prediction index threshold value so as to execute a subsequent process,

the cloud server is used for establishing and training a health index prediction model for each device model of a plurality of power devices, and the steps executed when establishing each health index prediction model comprise:

obtaining basic data, sensing data, health state prediction data, abnormal event data, service life data and maintenance data of the electric power equipment with a corresponding equipment model from a database, and establishing a fault occurrence rate prediction model according to the basic data, the sensing data, the health state prediction data, the abnormal event data, the service life data and the maintenance data through a machine learning mode;

the obtained basic data, sensing data, health state prediction data, abnormal event data, life data and maintenance data of each power equipment are brought into the fault occurrence rate prediction model so as to obtain a fault occurrence rate prediction value of each power equipment;

establishing a residual life prediction model through a machine learning mode according to the acquired basic data, sensing data, health state prediction data, abnormal event data, life data, maintenance data and failure occurrence rate prediction values of the power equipment;

bringing the obtained basic data, sensing data, health state prediction data, abnormal event data, life data, maintenance data and failure occurrence rate prediction value of each power equipment into the residual life prediction model to obtain a residual life prediction value of each power equipment; and

a health index prediction model is established through a machine learning mode according to the obtained basic data, sensing data, health state prediction data, abnormal event data, life data, maintenance data, fault occurrence rate prediction value and residual life prediction value of the power equipment.

15. The method for predicting the self-health of an electrical device as recited in claim 14, wherein the primary data of the electrical device includes at least one of a device model number, a manufacturing date, a rated power, a rated voltage, and a rated current.

16. The method of claim 14, wherein the sensed data includes at least one of an input voltage, an input current, an output voltage, an output current, a battery voltage, a battery charging current, a battery discharging current, an ambient temperature, and an ambient humidity.

17. The method of claim 14, wherein the post-processing includes transmitting the health status prediction data to the cloud server via the communication module.

18. The method of claim 14, wherein when the health status prediction data indicates that the health status of the power device is poor, the subsequent processing includes notifying a mail server via the communication module to issue an alert letter.

19. The method of claim 14, wherein when the health status prediction data indicates that the health status of the power device is poor, the subsequent processing includes notifying a short message sender to send an alarm short message via the communication module.

20. The method of claim 14, wherein the follow-up processing includes issuing an alarm through an alarm module of the power device when the health prediction data indicates that the health of the power device is poor.

21. The method of claim 14, wherein the machine learning algorithm used by the cloud server comprises at least one of a neural network-like algorithm, a decision tree algorithm, a K-average algorithm, a support vector machine algorithm, a linear regression algorithm, and a logistic regression algorithm.

22. The method of claim 14, wherein the power device comprises an uninterruptible power system, a power distribution unit, or an automatic power switch.

23. The method of claim 22, wherein the uninterruptible power system comprises an online uninterruptible power system, an offline uninterruptible power system, or an online interactive uninterruptible power system.

24. The utility model provides a high in clouds server, is applicable to a plurality of power equipment, this high in clouds server includes:

a communication module;

the database is used for storing basic data, sensing data, health state prediction data, abnormal event data, service life data and maintenance data of each power device and storing a health prediction index threshold value corresponding to a device model of each power device;

the system comprises a prediction model training module, a machine learning module and a machine learning module, wherein the prediction model training module is used for establishing and training a health index prediction model for each equipment model, and each health index prediction model is established and trained in a machine learning mode according to basic data, sensing data, health state prediction data, abnormal event data, service life data and maintenance data of the power equipment with a corresponding equipment model in the database; and

a data collection module for collecting basic data, sensing data, health state prediction data and abnormal event data of each power equipment through the communication module, obtaining corresponding service life data according to the basic data of each power equipment, providing a network user interface for collecting maintenance data and the health prediction index thresholds of each power equipment through the network user interface, storing all collected data, the health prediction index thresholds and the obtained service life data into the database,

wherein the predictive model training module performs the steps of, in building each health index predictive model:

obtaining basic data, sensing data, health state prediction data, abnormal event data, service life data and maintenance data of the electric power equipment with a corresponding equipment model from the database, and establishing a fault occurrence rate prediction model according to the basic data, the sensing data, the health state prediction data, the abnormal event data, the service life data and the maintenance data through a machine learning mode;

the obtained basic data, sensing data, health state prediction data, abnormal event data, life data and maintenance data of each power equipment are brought into the fault occurrence rate prediction model so as to obtain a fault occurrence rate prediction value of each power equipment;

establishing a residual life prediction model through a machine learning mode according to the acquired basic data, sensing data, health state prediction data, abnormal event data, life data, maintenance data and failure occurrence rate prediction values of the power equipment;

bringing the obtained basic data, sensing data, health state prediction data, abnormal event data, life data, maintenance data and failure occurrence rate prediction value of each power equipment into the residual life prediction model to obtain a residual life prediction value of each power equipment; and

a health index prediction model is established through a machine learning mode according to the obtained basic data, sensing data, health state prediction data, abnormal event data, life data, maintenance data, fault occurrence rate prediction value and residual life prediction value of the power equipment.

25. The cloud server of claim 24, wherein the predictive model training module, after the health index predictive models are built, trains each health index predictive model, the predictive model training module performing the steps of:

obtaining basic data, sensing data, health state prediction data, abnormal event data, service life data and maintenance data of the power equipment with the corresponding equipment model from the database with updated storage content, and training the failure occurrence rate prediction model in a machine learning mode according to the basic data, the sensing data, the health state prediction data, the abnormal event data, the service life data and the maintenance data;

the obtained basic data, sensing data, health state prediction data, abnormal event data, life data and maintenance data of each power equipment are brought into the trained fault occurrence rate prediction model so as to obtain a fault occurrence rate prediction value of each power equipment;

training the residual life prediction model through a machine learning mode according to the acquired basic data, sensing data, health state prediction data, abnormal event data, life data, maintenance data and failure occurrence rate prediction values of the power equipment;

bringing the obtained basic data, sensing data, health state prediction data, abnormal event data, life data, maintenance data and failure occurrence rate prediction values of each power device into the trained residual life prediction model to obtain a residual life prediction value of each power device; and

the health index prediction model is trained through a machine learning mode according to the obtained basic data, sensing data, health state prediction data, abnormal event data, life data, maintenance data, fault occurrence rate prediction value and residual life prediction value of the power equipment.

26. The cloud server of claim 24, wherein the base data for each power device includes at least one of a device model number, a date of manufacture, a rated power, a rated voltage, and a rated current.

27. The cloud server of claim 24, wherein the sensed data for each power device includes at least one of an input voltage, an input current, an output voltage, an output current, a battery voltage, a battery charging current, a battery discharging current, an ambient temperature, an ambient humidity.

28. The cloud server of claim 24, further comprising an alert message pushing module, wherein when the data collection module determines that any one of the predicted health status data indicates that the corresponding power device is not good, the data collection module pushes an alert message to an intelligent handheld device of a corresponding user through the alert message pushing module.

29. The cloud server of claim 24, wherein the data collection module further receives an alarm action and at least one judgment condition set by a user through the network user interface, so as to generate an alarm execution script, and transmits the alarm execution script to at least one corresponding power device through the communication module.

30. The cloud server of claim 24, wherein the machine learning algorithm used by the cloud server comprises at least one of a neural network-like algorithm, a decision tree algorithm, a K-average algorithm, a support vector machine algorithm, a linear regression algorithm, and a logistic regression algorithm.

31. The cloud server of claim 24, wherein the power devices are uninterruptible power systems, power distribution units, or automatic power switches.