CN113837264A

CN113837264A - Training method, electronic equipment, equipment health diagnosis method and device

Info

Publication number: CN113837264A
Application number: CN202111101678.5A
Authority: CN
Inventors: 韩庆; 吕亦宸; 郭蕊妮; 李丞伦; 吴振廷
Original assignee: Shenzhenshi Yuzhan Precision Technology Co Ltd
Current assignee: Shenzhenshi Yuzhan Precision Technology Co Ltd
Priority date: 2021-09-18
Filing date: 2021-09-18
Publication date: 2021-12-24

Abstract

The application discloses a training method, comprising the following steps: acquiring historical data, wherein the historical data comprises a parameter set and a result set formed by equipment operation; forming a training set and a test set according to the parameter set and the result set; training a first preset model according to the training set; testing the trained first preset model according to the test set; determining that the trained first preset model meets the preset requirement; and determining an estimation model based on the fact that the trained first preset model meets the preset requirement. According to the training method, the estimation model is formed by training the historical data formed by the operation input of the equipment, so that the standard for estimating the health state of the equipment can be obtained through the estimation model, a solution for estimating the model training is provided, the formation of the early warning range is enabled to depend on large data information rather than the working experience of personnel, and the automation and the intellectualization of a factory are facilitated. The application also discloses the electronic equipment, and an equipment health diagnosis method and device.

Description

Training method, electronic equipment, equipment health diagnosis method and device

Technical Field

The application relates to the technical field of equipment diagnosis, in particular to a training method, electronic equipment, and an equipment health diagnosis method and device.

Background

Whether the equipment is operating normally determines the quality of the product. In the prior art, the health state of the equipment is usually diagnosed by manually setting the maximum parameter range of normal operation, when the operation parameter of the equipment exceeds the maximum range, the equipment is indicated to be abnormal, and the equipment gives an alarm to inform maintenance personnel to maintain the equipment.

However, the above-mentioned prior art has at least the following problems: the parameter range of normal operation is set by operators with abundant experience, and because the experience of each operator is inconsistent, a uniform range standard is not easy to form, and the parameter range of normal operation is not easy to set. In addition, the alarm given by the existing equipment usually means that when the operating parameter exceeds the parameter range and causes an abnormality to be found, the equipment has been operated for a period of time, so that a plurality of defective products are generated, even the performance of the equipment is damaged, and the service life of the equipment is shortened.

Disclosure of Invention

In view of the above, it is necessary to provide a training method, an electronic device, a device health diagnosis method and an apparatus, so as to solve the technical problems: the health state is not easy to estimate when the equipment normally operates, target information and an early warning range are not easy to form, and the equipment cannot judge the early warning opportunity or early warning is not timely.

An embodiment of the present application provides a training method for estimating a model, which is used to estimate a health state of a device through the estimation model, and the method includes:

acquiring historical data, wherein the historical data comprises a parameter set and a result set formed by the operation of the equipment, the parameter set comprises at least one of electric information and control information input by the operation of the equipment, and the result set comprises target information for detecting the operation of the equipment through a sensor;

forming a training set and a test set according to the parameter set and the result set, wherein the training set comprises at least one of the electrical information and the control information and a corresponding set of the target information, and the test set comprises at least one of the electrical information and the control information and a corresponding other set of the target information;

training a first preset model according to the training set;

testing the trained first preset model according to the test set;

determining that the trained first preset model meets preset requirements;

and determining an estimation model based on the trained first preset model meeting the preset requirement.

An embodiment of the present application provides an electronic device, including a storage medium including readable instructions for executing the training method described above by a processor.

The application provides a training method for estimating a model and electronic equipment for executing the training method, wherein the estimation model is formed by training historical data formed by at least one of electrical information and control information input during equipment operation, and further the standard (such as an early warning interval) for estimating the health state of the equipment can be obtained through the estimation model, so that a solution for estimating the model training is provided, the formation of an early warning range depends on large data information rather than the working experience of personnel, and the automation and the intellectualization of a factory are favorably realized.

An embodiment of the present application provides an apparatus health diagnosis method for determining a health status of an apparatus, including:

receiving a first parameter set and target information formed according to the operation of the equipment, wherein the first parameter set comprises at least one of electric information and control information input by the operation of the equipment, and the target information is obtained based on the detection of the operation state of the equipment by a sensor;

inputting the first parameter set to an estimation model to form a second parameter set, wherein the estimation model is obtained by training according to historical data of the equipment;

and determining the health state of the equipment according to the second parameter set and the target information.

An embodiment of the present application provides a device health diagnosis apparatus, including a storage medium including readable instructions for executing the device health diagnosis method described above by a processor.

The application provides a method and a device for diagnosing equipment health, at least one of electric information and control information input during equipment operation is input into a trained estimation model during equipment processing, and the electric information and the control information are compared after an early warning interval is directly compared or calculated, so that the health state of the equipment can be determined, whether an early warning instruction needs to be sent or not can be determined, the generation of defective products caused by the fact that the equipment cannot early warn in advance is avoided, the performance of the equipment is prevented from being damaged, and the service life of the equipment is prolonged.

Drawings

Fig. 1 is a schematic flow chart of a training method provided in some embodiments of the present application.

Fig. 2 is a schematic flow chart of step 104 shown in fig. 1.

Fig. 3 is a schematic flow chart of step 106 shown in fig. 1.

FIG. 4 is another detailed flowchart of step 106 shown in FIG. 1

Fig. 5 is a schematic flow chart of a method for diagnosing health of a device according to some embodiments of the present disclosure.

Fig. 6 is a detailed flowchart of step 404 shown in fig. 5.

Fig. 7 is a detailed flowchart of step 406 shown in fig. 5.

FIG. 8 is a schematic flow chart of step 4066 shown in FIG. 7.

Fig. 9 is a hardware architecture diagram of an apparatus health diagnostic device provided in some embodiments of the present application.

FIG. 10 is a functional block diagram of the device health diagnostic system shown in FIG. 9.

Description of the main elements

Equipment health diagnosis device 20

Communicator 22

Processor 24

Memory 26

Device health diagnostic system 30

Receiving module 31

Forming a module 32

Determination module 33

Normalization processing module 34

Building block 35

Detailed Description

Reference will now be made in detail to embodiments of the present application, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to the same or similar elements or elements having the same or similar function throughout. The embodiments described below with reference to the accompanying drawings are illustrative and are only for the purpose of explaining the present application and are not to be construed as limiting the present application.

In the description of the present application, it is to be understood that the terms "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," "counterclockwise," and the like are used in the orientations and positional relationships indicated in the drawings for convenience in describing the present application and for simplicity in description, and are not intended to indicate or imply that the referenced devices or elements must have a particular orientation, be constructed in a particular orientation, and be operated in a particular manner, and are not to be construed as limiting the present application. Furthermore, the terms "first", "second" and "first" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, features defined as "first", "second", may explicitly or implicitly include one or more of the described features. In the description of the present application, it is to be noted that the meaning of "a plurality" is two or more unless specifically defined otherwise.

In the description of the present application, it is to be noted that, unless otherwise explicitly specified or limited, the terms "mounted," "connected," and "connected" are to be construed broadly, e.g., as meaning either a fixed connection, a removable connection, or an integral connection; they may be mechanically coupled, electrically coupled, or in communication with each other, directly coupled, or indirectly coupled through intervening media, in which case they may be interconnected, or in which case they may be in an interconnecting relationship. The specific meaning of the above terms in the present application can be understood by those of ordinary skill in the art as appropriate.

In this application, unless expressly stated or limited otherwise, the first feature "on" or "under" the second feature may comprise direct contact between the first and second features, or may comprise direct contact between the first and second features through another feature in between. Also, the first feature "on," "above" and "over" the second feature includes the first feature being directly above and obliquely above the second feature, or merely indicating that the horizontal thickness of the first feature is higher than that of the second feature. A first feature "under," "below," and "beneath" a second feature includes a first feature that is directly under and obliquely below the second feature, or simply means that the first feature is less horizontally thick than the second feature.

The following disclosure provides many different embodiments or examples for implementing different features of the application. In order to simplify the disclosure of the present application, specific example components and arrangements are described below. Of course, they are merely examples and are not intended to limit the present application. Moreover, the present application may repeat reference numerals and/or letters in the various examples, such repetition is for the purpose of simplicity and clarity and does not in itself dictate a relationship between the various embodiments and/or configurations discussed.

The embodiment of the present application provides a training method for estimating a model, which is used for estimating a health state of a device through the estimation model, and the method includes: acquiring historical data, wherein the historical data comprises a parameter set and a result set, the parameter set is formed by the operation of the equipment and comprises at least one of electric information and control information input by the operation of the equipment, and the result set comprises target information for detecting the operation of the equipment through a sensor; forming a training set and a test set according to the parameter set and the result set, wherein the training set comprises at least one of the electrical information and the control information and a corresponding set of the target information, and the test set comprises at least one of the electrical information and the control information and a corresponding other set of the target information; training a first preset model according to the training set; testing the trained first preset model according to the test set; determining that the trained first preset model meets the preset requirement; determining an estimation model based on the trained first predetermined model meeting the predetermined requirement.

The embodiment of the application also provides an electronic device, which comprises a storage medium, wherein the storage medium comprises readable instructions, and the readable instructions are used for being executed by a processor to execute the training method.

The embodiment of the present application further provides an apparatus health diagnosis method, configured to determine a health state of an apparatus, including: receiving a first parameter set formed according to the operation of the equipment and target information, wherein the first parameter set comprises at least one of electric information and control information input by the operation of the equipment, and the target information is obtained based on the detection of the operation state of the equipment by a sensor; inputting the first parameter set to an estimation model to form a second parameter set, wherein the estimation model is obtained by training according to historical data of the equipment; and determining the health state of the equipment according to the second parameter set and the target information.

The embodiment of the application also provides a device health diagnosis device, which comprises a storage medium, wherein the storage medium comprises readable instructions, and the readable instructions are used for being executed by a processor to execute the device health diagnosis method.

The application provides a method and a device for diagnosing equipment health, wherein at least one of electric information and control information input during equipment operation is input into a trained estimation model when the equipment is processed, and the electric information and the control information are compared after an early warning interval is calculated, so that the health state of the equipment can be determined, whether an early warning instruction needs to be sent or not can be determined, the generation of defective products caused by the fact that the equipment cannot be early warned in advance is avoided, the performance of the equipment is prevented from being damaged, and the service life of the equipment is prolonged.

Embodiments of the present application will be described in detail below with reference to the accompanying drawings.

Referring to fig. 1, fig. 1 is a schematic flow chart of a training method according to some embodiments of the present disclosure. The training method is used for training an estimation model, and the health state of the equipment is estimated through the estimation model. The order of the steps in the flow chart may be changed and some steps may be omitted according to different needs. For convenience of explanation, only portions related to the embodiments of the present application are shown. The training method comprises the following steps.

S102, historical data is obtained, the historical data comprises a parameter set and a result set, the parameter set comprises at least one of electric information and control information input by equipment operation, and the result set comprises target information for detecting equipment operation through a sensor.

Illustratively, historical data formed by the operation of the device is obtained, and the historical data comprises a parameter set and a result set. The parameter set represents at least one of electrical information and control information input when the device operates, the electrical information can be understood as voltage information and current information, the control information can be understood as command information, and the device operates by inputting at least one of the electrical information and the control information. The result set corresponds to at least one of the input electrical information and the control information. It is understood that an element in a parameter set corresponds to an element in a result set, and the result set may be target information for detecting the operation of the device through a sensor in the device, and the target information may reflect the result of whether the device is operating normally.

In some embodiments, the machine may be a Computer Numerically Controlled (CNC) machining machine, such as a CNC tool machining machine. It will be appreciated that in other embodiments where the apparatus includes critical components of the apparatus which directly reflect the health of the apparatus, the historical data obtained may include the parameter set and result set formed by the operation of the critical components, for example, the critical component of the CNC machining apparatus may be a tool spindle on a machine tool, which may be mounted on a servo motor or a motor, and the CNC machining apparatus is dependent in large part on the life of the oil contained in its internal ball bearings, so the higher the temperature of the motor, the shorter the life of the ball bearings. When the motor is used at high temperature for a long time, the lubricating oil in the ball bearing is easy to liquefy and evaporate to dissolve, and further the bearing is blocked, so that if the temperature of the servo motor is too high, the equipment is possibly abnormal, and the service life of the servo motor is reduced. The main shaft is driven by the main shaft servo motor, and the main shaft servo motor are in a direct proportion relation, so that if the health state of the servo motor can be diagnosed, the health state of the CNC cutter machining equipment can be reflected. For example, the electrical information may be information such as current, voltage or power loaded on a machine (e.g., a servo motor), and the control information may be information such as rotational speed, load, feed rate or torque of a tool spindle. The target information may be a temperature of a tool spindle of the CNC tool machining apparatus, or may be a temperature of a servo motor or a tool spindle passing. In some cases, the electrical information represented by the current, voltage or power information loaded on the machine platform can determine the input information of the CNC tool, and the control information of the rotation speed, load, feed rate, torque and the like of the CNC tool processing equipment can be determined through calculation according to the input information, and it can be understood that the parameter set in the historical data can only comprise the electrical information input by the equipment operation. However, in some cases, there is a problem that the input electrical information does not match with the control information, and in order to make the health diagnosis result of the equipment more accurate, it can be understood that the parameter set in the historical data may only include the electrical information input by the operation of the equipment, such as the information of the rotating speed, the load, the feed rate or the torque of the tool spindle. In other scenarios, to ensure accuracy and greater accuracy, the parameter sets in the historical data may include both electrical and control information.

S104, according to the parameter set and the result set, a training set and a test set are formed, wherein the training set comprises at least one of the electrical information and the control information and a set of corresponding target information, and the test set comprises at least one of the electrical information and the control information and another set of corresponding target information.

Illustratively, according to the obtained parameter set and result set, the parameter set and the result set corresponding to the parameter set are divided into a training set and a testing set. The training set is a set including at least one of electrical information and control information and corresponding target information, and the test set is another set including at least one of electrical information and control information and corresponding target information. The training set and the test set may include all parameter sets and result sets, or may include partial parameter sets and result sets corresponding to the partial parameter sets. For example, the training set is a set of the first part of feature data and corresponding temperatures, and the test set is a set of the second part of feature data and corresponding temperatures. For example, the historical data includes 3 sets of data, such as: when the rotation speed is (5,7,9) and the corresponding spindle temperature is (80,85,90), one of the forming methods is: the training set is (5,80) and (7,85), and the test set is (9,90), thereby completing the splitting.

Referring to fig. 2, in some embodiments, there may be abnormal values in the parameter set and the result set, which may reduce the estimation accuracy of the models trained in the training set and the test set. The step of forming the training set and the test set in S104 may further include the following steps S1041 to S1043.

S1041, determining a first quartile and a second quartile of the parameter set or the result set.

Illustratively, the first quartile Q1 and the second quartile Q2 in the parameter set or result set are determined according to a quartile method. When the first quartile and the second quartile are determined, only one of the parameter set and the result set may be determined. For example, after determining the abnormal values in the parameter set, the corresponding result set is also the abnormal values, and the abnormal values in the parameter set and the result set are deleted; or after determining the abnormal values in the result set, the corresponding parameter set is also the abnormal values, and the parameter set and the abnormal values in the result set are deleted.

S1042, forming a four-quantile distance according to the first quartile and the second quartile.

Illustratively, the quartile range IQR is formed based on a difference between the first quartile and the second quartile. It is understood that the quartile range IQR is a difference between the first quartile and the second quartile. In this embodiment, the parameter set or the result set is approximately normally distributed, wherein the first quartile Q1 is a negative number, and the second quartile Q2 is a positive number, and is symmetrically distributed on two sides of the midpoint 0.

S1043, forming a health interval according to the first quartile, the second quartile and the quartile distance, and forming a training set and a test set according to the parameter set and the elements of the result set in the health interval.

Illustratively, the health intervals formed according to the first quartile, the second quartile, and the interquartile range may be (Q1-1.5 IQR, Q3+1.5 IQR). It is understood that in other embodiments, the healthy range may also be (Q1-2 × IQR, Q3+2 × IQR), and the like. Illustratively, one way of processing is: the elements exceeding the healthy interval in the parameter set and the result set are abnormal values, the elements exceeding the healthy interval are deleted, and a training set and a test set are formed according to the remaining parameter set and the result set, so that the abnormal values in the parameter set and the result set can be deleted through the steps S1041-S1043, the estimation accuracy of the estimation model trained by the training set and the test set can be improved, and the over-fitting problem of training can be prevented.

In some embodiments, to improve the generalization ability of the models trained by the training set and the test set, standardized data processing is also required on the result set and the test set before forming the training set and the test set. The step of forming the training set and the test set in S104 may further include the following steps S1044-S1045.

S1044, determining the mean and standard deviation of the parameter set and the result set.

For example, the mean values of the parameter set and the result set are determined, and further, whether the mean value is 0 and the standard deviation is 1 may be determined, and if not, the operation is performed so that the mean value is 0 and the standard deviation is 1.

And S1045, standardizing the parameter set and the result set according to the mean value and the standard deviation.

Illustratively, the parameter set and the result set are Z-Score normalized according to a mean value of 0 and a standard deviation of 1. The characteristic data in the historical data is standardized through a Z-Score standardization process, so that the parameter set and the result set have unified measurement, and the standard of the parameter set and the result set is unified. Therefore, the parameter set and the result set are subjected to Z-Score standardization processing, so that the standards of the parameter set and the result set are unified, the standards of the formed training set and the test set are also unified, and the generalization capability of the models trained by the training set and the test set is favorably improved.

In some embodiments, in order to improve the estimation accuracy of the model trained by the training set and the test set, the model needs to be trained repeatedly, and a plurality of groups of samples of the training set and the test set are also needed. The step of forming the training set and the test set in S104 may further include the following steps S1046 to S1048.

S1046, randomly splitting the parameter set and the result set into N samples, where N is an integer and greater than 1.

For example, the parameter set and the result set corresponding to the parameter set are randomly split into N samples, where N may be 2, 3, 4, 5, and 6 …, N is an integer and is greater than 1, and each sample includes the parameter set and the result set.

S1047, randomly selecting N-1 samples to form a training set.

Illustratively, the model is trained by randomly selecting N-1 samples from N samples as a training set.

And S1048, selecting the remaining 1 sample to form a test set.

Illustratively, the remaining 1 sample is selected as the test set for testing the trained model. For example, N is 10, and 1, 2, 3, 4, 5, 6, 7,8, 9 samples are selected as a training set, and 10 samples are selected as a test set; or selecting 1 st, 2 nd, 3 rd, 4 th, 6 th, 7 th, 8 th, 9 th and 10 th samples as training sets and 5 th samples as test sets; by analogy, 10 training sets and test sets can be obtained, and 10 times of iterative training can be performed on the model. Therefore, the accuracy of the training set and the test set in model training is improved.

It can be understood that when the number of the samples is N, N sets of training sets and test sets can be obtained, N times of iterative training can be performed on the model, and estimation accuracy of the model trained by the training sets and the test sets can be improved by performing iterative training on the model for multiple times and performing cross validation on the trained models.

It will be appreciated that in other embodiments, it is also possible to randomly select N-2, N-3, N-4 … samples as the training set and the remaining 2, 3, 4 … samples as the test set.

And S106, training a first preset model according to the training set.

Illustratively, the first pre-set model is trained by the acquired one or more sets of training sets.

Referring to fig. 3, in some embodiments, the training set may include a first sample and a second sample, wherein the first sample is I₁＝{(x₁，y₁)，(x₂，y₂)，…(x_m，y_m) The second sample is I₂＝{(x₁，y₁)，(x₂，y₂)，…(x_n，y_n) The step of training the first preset model according to the training set in S106 may include the following steps S1061-S1068.

S1061, determining the training parameters and the loss function of the first preset model.

Illustratively, the training parameters of the first predetermined model include a maximum number of iterative operations T, regularization coefficients λ, γ, and a loss function L.

And S1062, inputting a first sample to the first preset model according to the training parameters, and determining a first parameter of a calculation result of the first preset model according to the loss function.

Illustratively, a first sample is input into a first preset model according to the training parameters, and a first parameter f (1) of a calculation result of the first preset model is determined according to the loss function.

S1063, updating the first preset model to be the first model according to the first parameter.

Illustratively, the first pre-set model is updated to form the first model according to the first parameter f (1).

And S1064, determining that the first parameter is not equal to the preset parameter.

For example, it is determined that the first parameter f (1) is not equal to the preset parameter f, and the preset parameter f may be 0. It is understood that the preset parameter f may have other values.

And S1065, inputting a second sample to the first model according to the training parameter based on the fact that the first parameter is not equal to the preset parameter, and determining a second parameter of the calculation result of the first model according to the loss function.

Illustratively, based on the fact that the first parameter f (1) is not equal to the preset parameter f, that is, the first model does not meet the model requirement, the second sample is input into the first model according to the training parameter, and the second parameter f (2) of the calculation result of the first model is determined according to the loss function.

And S1066, updating the first model into a second model according to the second parameter.

Illustratively, the first model is updated to form the second model in accordance with the second parameter f (2).

S1067, determining that the second parameter is equal to the preset parameter.

Illustratively, the second parameter f (2) is determined to be equal to the preset parameter f.

And S1068, determining the second model to be the trained first preset model based on the fact that the second parameter is equal to the preset parameter.

Illustratively, the second model is determined to be the trained first preset model based on the fact that the second parameter f (2) is equal to the preset parameter f, that is, the second model meets the model requirement.

It is to be understood that if the first parameter f (1) is equal to the preset parameter f and the first model meets the model requirement, the first model is determined to be the trained first preset model, and steps S1065-S1068 may be omitted. If the second parameter f (2) is still not equal to the preset parameter f, more samples such as a third sample and a fourth sample are formed, the third sample is continuously input into the second model, and the fourth sample is continuously input into the third model … until the parameter f (n) is equal to the preset parameter f, and the corresponding model is determined to be the trained first preset model.

Thus, through the steps S1061-S1068, the first predetermined model is trained, and the trained first predetermined model has accurate estimation and strong generalization capability.

Referring to fig. 4, in some embodiments, the first predetermined model may also be trained through an XGBoost algorithm process, or may be understood as the first predetermined model is exemplarily an XGBoost model. Illustratively, the step of training the first preset model according to the training set in S106 may further include the following steps S1071 to S1079.

S1071, inputting the training set to the training model to form a first derivative and a second derivative corresponding to each element in the training set.

Illustratively, the training set is input to a training model f_t-1(x_i) Forming each element in the training set based on f_t-1(x_i) First derivative g of_tiAnd second derivative h_ti。

S1072, determining an accumulated value G of the first derivative_tAnd the accumulated value H of the second derivative_t。

Illustratively, the accumulated value G of the first derivatives of all elements in the training set is calculated_tAnd the accumulated value H of the second derivative_t。

S1073, according to the accumulated value G_tAnd an accumulated value H_tAnd determining the split node of the first preset model.

Illustratively, according to the obtained accumulated value G_tAnd tired ofAdded value H_tAnd splitting the first preset model and determining split nodes of the first preset model. It can also be understood that the accumulated value G is used as the basis_tAnd an accumulated value H_tThe condition for splitting the first predetermined model is determined, for example, the splitting may be performed according to whether the electrical information in the parameter set is smaller than a specific value.

And S1074, adjusting the training model according to the split node.

Illustratively, the training model is adjusted to form left and right subtrees according to the split nodes of the first predetermined model. For example, a left sub-tree is a sub-tree in the parameter set in which the electrical information is smaller than a certain value, and a right sub-tree is a sub-tree in the parameter set in which the electrical information is larger than the certain value.

S1075, inputting the training set to the adjusted training model to form a first derivative and a second derivative corresponding to each element of the training set.

Illustratively, the training set is input into the left and right subtrees, each element forming the training set being based on f_t-1(x_i) First derivative g of_tiAnd second derivative h_ti。

S1076, determining the accumulated value G of the first derivative_LAnd an accumulated value G_RAnd the accumulated value H of the second derivative_LAnd an accumulated value H_R。

Illustratively, the accumulated value G of the first derivative of the left subtree is determined_LAnd the accumulated value H of the second derivative_LDetermining the accumulated value G of the first derivative of the right subtree_RAnd the accumulated value H of the second derivative_R。

S1077, according to the accumulated value G_LAnd an accumulated value G_RAnd an accumulated value H_LAnd an accumulated value H_RAnd determining the gain value of the adjusted training model.

Illustratively, according to accumulated value G_LAnd an accumulated value G_RAnd an accumulated value H_LAnd an accumulated value H_RDetermining a gain value score1 of the adjusted training model, wherein score1 satisfies the following conditional expression:

Score1＝max(score，(G_L ²/2(H_L+λ))+(G_R ²/2(H_R+λ))-((G_L+G_R)²/2(H_L+H_R+ λ)) - γ), where λ, γ are regularization coefficients of the adjusted training model.

S1078, determining the gain value to be equal to a preset value.

Illustratively, the gain value score1 is determined to be equal to a preset value, wherein the preset value is score, which may be 0. It is understood that in other embodiments, the preset value score may be more preset values such as 1, 2, etc.

S1079, determining the optimal splitting node of the first preset model based on the gain value equal to the preset value, and adjusting the first preset model according to the optimal splitting node to finish the training of the first preset model.

Illustratively, based on the gain value score1 being equal to the preset value score, an optimal splitting node of the first preset model is determined, the first preset model is adjusted and updated by the optimal splitting node, and the adjusted and updated first preset model is determined as the trained first preset model.

Thus, through the steps S1071 to S1079, the first preset model is trained, and the trained first preset model has accurate estimation and strong generalization capability.

And S108, testing the trained first preset model according to the test set.

Illustratively, the trained first predetermined model is trained according to the test set to test whether the trained first predetermined model meets the requirements.

And S110, determining that the trained first preset model meets the preset requirement.

Illustratively, after the test set testing, it is determined that the trained first predetermined model meets the predetermined requirements. It can be understood that, if the trained first preset model does not meet the preset requirement after the test set test, the first preset model is retrained again by using other training sets and test set samples until the trained first preset model meets the preset requirement.

And S112, determining an estimation model based on the fact that the trained first preset model meets the preset requirement.

For example, if the trained first predetermined model meets the predetermined requirement, the trained first predetermined model is the estimation model. It is understood that, since the number of the samples formed by the training set and the test set is multiple, the number of the determined estimation models may also be multiple, that is, there are multiple trained first predetermined models meeting the predetermined requirement.

In some embodiments, since the number of the training set and the test set is multiple, and the number of the trained first predetermined models meeting the predetermined requirement is multiple, after it is determined that the plurality of trained first predetermined models meet the predetermined requirement, an optimal one of the plurality of trained first predetermined models is further selected as the estimation model, so as to further improve the estimation accuracy of the estimation model. Step S110 may be followed by steps S114-S122 as follows.

And S114, setting random parameters of the trained first preset model based on the fact that the trained first preset model meets preset requirements.

Illustratively, the random parameters of the trained first preset model are set based on the plurality of trained first preset models meeting the preset requirements. It will be appreciated that the random parameter may be a range of values within which the result set that best matches the result may be most quickly estimated by inputting a result set.

And S116, searching the trained first preset model according to the random parameters to form a search result.

Illustratively, a plurality of trained first preset models are searched according to set random parameters to form search results. The search results may be understood as all first predetermined models that are satisfactory within the scope.

And S118, determining that the search result meets the requirement.

Illustratively, whether the plurality of first preset models meet requirements or not is determined according to the plurality of first preset models in the search result, and the condition of error search is avoided.

And S120, determining the optimal parameters of the trained first preset model based on the search result meeting the requirements.

Illustratively, based on the search result conformity requirement, the optimal parameters of the plurality of trained first preset models are determined, and the optimal parameters can be understood as the first preset models which are the fastest and best estimate the result set which best conforms to the result.

And S122, determining an estimation model according to the trained first preset model and the optimal parameters.

Illustratively, the trained first predetermined model is determined as an estimation model according to the optimal parameters and the corresponding trained first predetermined model. Thus, through steps S114-S122, the determined estimation model is the optimal trained first predetermined model, which is beneficial to improving the estimation accuracy of the estimation model.

Some embodiments of the present application also provide an electronic device. The electronic device includes a storage medium comprising readable instructions for execution by a processor of a training method as described above.

The electronic device includes, but is not limited to, a computer, a dedicated device, a data center, a server, and other hardware devices. In one embodiment, the electronic device may be a microcomputer including at least one processor, wherein the processor may include an Application Processor (AP), a modem processor, a Graphics Processing Unit (GPU), an Image Signal Processor (ISP), a controller, a video codec, a Digital Signal Processor (DSP), a baseband processor, and/or a neural-Network Processing Unit (NPU), and/or the like.

According to the training method and the electronic device provided by the embodiment, historical data formed in the operation process of the device is processed, the first preset model is trained and selected, the optimal first preset model is selected as the estimation model, the formed estimation model has a uniform operation range standard, the standard (such as an early warning interval) for estimating the health state of the device can be obtained through the estimation model, a solution for estimation model training is provided, the model generalization capability for estimating the health state of the device is improved, more use scenes can be expanded, the formation of the early warning range depends on big data information rather than the working experience of personnel, and automation and intellectualization of a factory are facilitated. By the training method provided by the embodiment, when the parameter set during the operation of the equipment is input into the estimation model, the corresponding estimation value meeting the operation requirement can be obtained, and whether the equipment normally operates and whether the equipment needs to perform early warning or other operations can be obtained by comparing the estimation value with the actual target information.

Referring to fig. 5, fig. 5 is a flow chart illustrating a method for diagnosing health of a device according to some embodiments of the present application. A device health diagnostic method is used to determine the health status of a device. The order of the steps in the flow chart may be changed and some steps may be omitted according to different needs. For convenience of explanation, only portions related to the embodiments of the present application are shown. The device health diagnosis method includes the following steps.

S402, a first parameter set formed according to the operation of the equipment and target information are received, wherein the first parameter set comprises at least one of electric information and control information input by the operation of the equipment, and the target information is obtained based on the operation state of the equipment detected by a sensor.

In some embodiments, the machine may be a Computer Numerically Controlled (CNC) machining machine, such as a CNC tool machining machine. It will be appreciated that in other embodiments, the critical component of the apparatus is included in the apparatus, and the critical component directly reflects the health of the apparatus, and the acquired historical data may include a set of parameters and a set of results generated by the operation of the critical component, for example, the critical component of the CNC tool machining apparatus may be a tool spindle on a machine tool. For example, the electrical information may be information such as current, voltage or power loaded on the machine, and the control information may be information such as rotational speed, load, feed rate or torque of the tool spindle. The target information may be a spindle temperature of the CNC tool machining apparatus, and then the corresponding sensor is a temperature sensor. In some cases, the electrical information represented by the current, voltage or power information loaded on the machine platform can determine the input information of the CNC tool, and determine the control information of the rotation speed, load, feed rate, torque and the like of the CNC tool processing equipment through calculation according to the input information, and it can be understood that the first parameter set can only comprise the electrical information of the operation input of the equipment. However, in some cases, there is a problem that the input electrical information does not match the control information, and in order to make the device health diagnosis result more accurate, it can be understood that the first parameter set may only include the electrical information input by the device operation, such as the information of the rotation speed, the load, the feed rate or the torque of the tool spindle. In other scenarios, to ensure accuracy and greater accuracy, the first set of parameters may include both electrical and control information. The above-mentioned training method is already described, and will not be described herein.

S404, inputting the first parameter set to an estimation model to form a second parameter set, wherein the estimation model is obtained by training according to historical data of the equipment.

Illustratively, a first parameter is input into an estimation model, forming a second parameter set. The estimation model may be formed by training a first predetermined model by the training method, inputting a first parameter set to the estimation model, and outputting a second parameter set estimated by the estimation model, where the second parameter set may be data such as temperature corresponding to the first parameter set. It should be noted that the second parameter set is a set of estimated values, and is not a value of the actual operation of the device.

Referring to fig. 6, in some embodiments, in order to make the first parameter set and the second parameter set have a uniform metric, the step of forming the second parameter set in S404 further includes the following steps S4042-S4044.

S4042, a mean and a standard deviation of the first parameter set are determined.

Illustratively, the mean value of the first parameter set is determined to be 0 and the standard deviation is determined to be 1.

S4044, standardizing the first parameter set according to the mean and the standard deviation.

Illustratively, according to the mean value of 0 and the standard deviation of 1 of the first parameter set, the first parameter set is subjected to Z-Score normalization processing so that the first parameter set has a uniform metric, and the normalized first parameter set is input to the estimation model to obtain the second parameter set so that the second parameter set also has a uniform metric.

And S406, determining the health state of the equipment according to the second parameter set and the target information.

In some embodiments, the health status of the device may be determined by directly comparing the second set of parameters to the size of the subject information. For example, the second parameter set includes (81,86), the target information is (82,83), and the CNC tool spindle is taken as an example of a key component, and (81,86) may represent estimated temperature values of two tool spindle key points, which may be understood as that if any one of the target information exceeds the two values, the health status of the equipment is determined to be bad, i.e., the health status of the equipment is determined to be bad when the target information is (82, 83).

In other embodiments, the method may include inputting a second set of parameters into a second predetermined model to form the warning interval.

Illustratively, a second parameter set is input into a second predetermined model to form the early warning interval. The second preset model is approximately a normal distribution model, and the second parameter set is input into the normal distribution model to form an early warning interval. It can be understood that one value of the first parameter set is input into the estimation model, and a corresponding value of the second parameter set is output, and the value of the second parameter set is used as a midpoint of the normal distribution model, so that the midpoint forms the early warning interval.

Referring to FIG. 7, in some embodiments, the step of determining the health status of the device may include steps S4062-S4066 as follows.

S4062, inputting a second parameter set to a second preset model, and establishing a distribution model of the second parameter set.

Illustratively, a second parameter set is input to the second predetermined model to establish a distribution model of the second parameter set. It is to be understood that the second predetermined model is substantially a normal distribution model, and the second parameter set is taken as the center of the normal distribution, so that the normal distribution model of the second parameter set is formed.

S4064, determining a standard deviation of the second parameter set according to the distribution model.

Illustratively, the standard deviation σ of the second parameter set is determined according to a normal distribution model of the second parameter set and the principle that 99.7% of the values should lie within the range of standard deviations.

S4066, determining the health state of the equipment according to the target information and the standard deviation.

Illustratively, the health status of the device is determined based on the target information and the standard deviation σ of the second set of parameters. It can be understood that the actual target information of the device during operation is compared with the standard deviation of the second parameter set, and if the target information exceeds the standard deviation of the second parameter set, the health status of the device is bad, and an early warning or other operation is required. If the subject information does not exceed the standard deviation of the second set of parameters, the health status of the device is good and no pre-warning or other operations are required.

Referring to fig. 8, in some embodiments, the standard deviation comprises a first standard deviation. Illustratively, the standard deviation includes a first standard deviation σ₁. Step S4066 may include steps S4081-S4082 as follows.

S4081, forming a first early warning interval according to the first standard deviation.

Illustratively, according to a first standard deviation σ₁And a normal distribution model of a second parameter set, wherein the formed first early warning interval is (-sigma)₁，σ₁)。

S4082, based on the fact that the target information exceeds the first early warning interval, the health state of the equipment is determined to be the first state.

Illustratively, the target-based information exceeds a first warning interval (- σ)₁，σ₁) Determining the health status of the device as a first status, wherein the first status can be understood as the target information exceeding a first warning interval (-sigma)₁，σ₁) The equipment has serious defects and should be shut down for maintenance.

It is to be understood that in other embodiments, the first state may also be understood as the presence of a malfunction in the device, which is pre-warned but not shut down.

In some implementationsIn an example, the standard deviation further includes a second standard deviation, exemplary. The standard deviation includes a second standard deviation σ₂，σ₂<σ₁. Step S4066 may also include steps S4083-S4084 as follows.

S4083, forming a second early warning interval according to the second standard deviation, wherein the second early warning interval is within the first early warning interval.

Illustratively, according to the second standard deviation σ₂And a normal distribution model of a second parameter set, wherein the formed second early warning interval is (-sigma)₂，σ₂) And a second early warning interval (-sigma)₂，σ₂) In the first early warning interval (-sigma)₁，σ₁) And (4) the following steps.

S4084, the health state of the equipment is determined to be a second state in the first early warning interval based on the fact that the target information exceeds the second early warning interval.

Illustratively, the target-based information exceeds a second warning interval (- σ)₂，σ₂) And in the first early warning interval (-sigma)₁，σ₁) And determining the health state of the equipment to be a second state. The second state can be understood as that the equipment is bad, and the equipment gives an alarm or an early warning to prompt an operator to carry out maintenance or inspection.

In some embodiments, the standard deviation further comprises a third standard deviation. Illustratively, the standard deviation includes a third standard deviation σ₃，σ₃<σ₂. Step S4066 may also include steps S4085-S4086 as follows.

S4085, forming a third early warning interval according to the third standard deviation, wherein the third early warning interval is within the second early warning interval.

Illustratively, according to a third standard deviation σ₃And forming a third early warning interval (-sigma) by the normal distribution model of the second parameter set₃，σ₃) And the third early warning interval (-sigma)₃，σ₃) In the second early warning interval (-sigma)₂，σ₂) And (4) the following steps.

S4086, the health state of the equipment is determined to be a third state in the second early warning interval based on the fact that the target information exceeds the third early warning interval.

Illustratively, the health status of the device is determined to be the third status within the second warning interval based on the target information exceeding the third warning interval. The third state can be understood as that the equipment is slightly bad, and the equipment gives out early warning to prompt the operator to check the equipment after production is finished.

It will be appreciated that in other embodiments, the standard deviation may include more deviation values, thereby allowing multiple levels of pre-warning of equipment. When the equipment is bad, the equipment sends out different instruction prompts according to different grades.

The method for diagnosing the health of the equipment provided by the embodiment of the application comprises the steps of obtaining a first parameter set and inputting the first parameter set into an estimation model, and estimating a second parameter set by the estimation model. When target information of the equipment is input, the health state of the equipment can be determined according to the second parameter set or the early warning interval formed by the second parameter set, so that whether an early warning instruction needs to be sent or not can be determined according to the determined health state, the generation of defective products caused by the fact that the equipment cannot early warn in advance is avoided, the performance of the equipment is prevented from being damaged, and the service life of the equipment is prolonged.

Referring to fig. 9, fig. 9 illustrates a device health diagnosis apparatus provided in some embodiments of the present application, which includes a storage medium including readable instructions for being executed by a processor to perform the device health diagnosis method. The device health diagnostic apparatus 20 is used to determine the health status of the device.

Illustratively, device health diagnostic apparatus 20 includes a communicator 22, a processor 24, and a memory 26, with communicator 22 and memory 26 each coupled to processor 24. The memory 26 is illustratively a storage medium in the device health diagnostic apparatus. The steps of the tasks performed by the processor 24 may be the same as or similar to the technical solutions described in the above sections of the training method.

Communicator 22 is configured to receive a first set of parameters derived from device operation, including at least one of electrical information and control information input from device operation, and target information derived based on sensor detection of device operating conditions. The first parameter set represents at least one of electrical information and control information input when the device operates, the electrical information can be understood as voltage information and current information, the control information can be understood as command information, and the device operates by inputting at least one of the electrical information and the control information. The target information may be a temperature generated when the equipment is operated, or the like, which can reflect the result of whether the equipment is normally operated.

The Processor 24 may be a Central Processing Unit (CPU), and may include other general purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field Programmable Gate Arrays (FPGAs) or other Programmable logic devices, discrete Gate or transistor logic devices, discrete hardware components, and the like. The general purpose processor may be a microprocessor or the processor 24 may be any conventional processor or the like, and the processor 24 is the control center of the device health diagnostic apparatus 20 and connects the various parts of the overall device health diagnostic apparatus 20 using various interfaces and lines.

The memory 26 is used for storing various types of data, such as various databases, program codes, and the like, in the device health diagnostic apparatus 20. In some embodiments, Memory 26 may include, but is not limited to, Read-Only Memory (ROM), Random Access Memory (RAM), Programmable Read-Only Memory (PROM), Erasable Programmable Read-Only Memory (EPROM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Compact Disc Read-Only Memory (CD-ROM), or other optical Disc storage, magnetic disk storage, tape storage, or any other medium from which a computer can Read to carry or store data.

Referring to fig. 10 together, fig. 10 shows functional modules of a device health diagnosis system according to some embodiments of the present application. Device health diagnostic system 30 includes one or more computer instructions in the form of a program (i.e., readable instructions) stored in memory 26 and executed by processor 24 to implement the functions provided herein.

Device health diagnostic system 30 may be a digital twin of device health diagnostic apparatus 20, may exist in a computer virtual space, and is a fully equivalent information model of a physical entity (device health diagnostic apparatus 20).

In some embodiments, the device health diagnosis system 30 may be functionally divided into a receiving module 31, a forming module 32 and a determining module 33, but is not limited thereto, and the functionally divided modules are only for illustration and do not represent the actual working form of the device health diagnosis system 30. The functions of the respective modules will be described in detail in the following embodiments.

The receiving module 31 is configured to receive a first parameter set formed by the operation of the device and target information, where the first parameter set includes at least one of electrical information and control information input by the operation of the device, and the target information is obtained based on the sensor detecting the operation state of the device.

The forming module 32 is configured to input the first parameter set to an estimation model, which is obtained by training according to historical data of the device, to form a second parameter set.

For example, the estimation model may be formed by training a first predetermined model by the above-mentioned training method, inputting a first parameter set to the estimation model, and outputting a second parameter set for estimation by the estimation model, where the second parameter set may be data such as temperature corresponding to the first parameter set. It should be noted that the second parameter set is a set of estimated values, and is not a value of the actual operation of the device.

The determination module 33 is configured to determine the health status of the device based on the second parameter set and the target information.

Illustratively, the target information actually generated by the device during operation is compared with the second parameter set (or the warning interval formed according to the second parameter set). Taking the early warning interval formed by the second parameter set as an example, if the target information exceeds the early warning interval, the health state of the equipment is bad, and early warning or other operations are required. And if the target information does not exceed the early warning interval, the health state of the equipment is good, and early warning or other operations are not needed. The early warning interval may be calculated according to the value estimated by the estimation model, or may be compared by directly outputting an early warning interval through the estimation model, but is not limited thereto, or may be compared with the second parameter set directly through the target information to obtain the diagnosis conclusion of the health status of the device.

In some embodiments, device health diagnostic system 30 may also include a normalization processing module 34.

The determination module 33 is also used to determine the mean and standard deviation of the first parameter set.

The normalization module 34 is configured to normalize the first parameter set according to the mean and the standard deviation.

Illustratively, according to the mean value of the first parameter set being 0 and the standard deviation being 1, the first parameter set is subjected to Z-Score normalization processing, so that the first parameter set has a uniform metric, which facilitates the input of the estimation model to form the second parameter set and the formation of the early warning interval by the second parameter set.

In some embodiments, device health diagnostic system 30 may also include a setup module 35.

The establishing module 35 is configured to input a second parameter set to the second preset model, and establish a distribution model of the second parameter set.

For example, in a scenario where the estimation model estimates a specific value, the second preset model may be selected as a normal distribution model, and the second parameter set is used as a midpoint of the normal distribution. In this way, a normal distribution model of the second parameter set is formed.

The determining module 33 is configured to determine a standard deviation of the second parameter set according to the distribution model.

The determination module 33 is further configured to determine the health status of the device based on the target information and the standard deviation.

Illustratively, the health status of the device is determined based on the target information and the standard deviation σ.

In some embodiments, the standard deviation comprises a first standard deviation. Illustratively, the standard deviation includes a first standard deviation σ₁。

The forming module 32 is further configured to form a first warning interval according to the first standard deviation.

Illustratively, according to a first standard deviation σ₁And normal division of the second parameter setThe cloth model forms a first early warning interval of (-sigma)₁，σ₁)。

The determining module 33 is further configured to determine the health status of the device as the first status based on the target information exceeding the first warning interval.

In some embodiments, the standard deviation further comprises a second standard deviation. Illustratively, the standard deviation includes a second standard deviation σ₂，σ₂<σ₁。

The forming module 32 is further configured to form a second warning interval according to the second standard deviation, where the second warning interval is within the first warning interval.

The determining module 33 is further configured to determine that the health status of the device is the second status within the first warning interval based on the target information exceeding the second warning interval.

In some embodiments, the standard deviation further comprises a third standard deviation. Exemplary ofThe standard deviation includes a third standard deviation σ₃，σ₃<σ₂。

The forming module 32 is further configured to form a third early warning interval according to the third standard deviation, where the third early warning interval is within the second early warning interval.

The determining module 33 is further configured to determine that the health status of the device is the third status within the second warning interval based on the target information exceeding the third warning interval.

The device health diagnosing apparatus 20 according to the embodiment of the present application obtains the first parameter set and inputs the first parameter set to an estimation model, so that the estimation model estimates the second parameter set. When target information of the equipment is input, the health state of the equipment can be determined according to the second parameter set or the early warning interval formed by the second parameter set, so that whether an early warning instruction needs to be sent or not can be determined according to the determined health state, the generation of defective products caused by the fact that the equipment cannot early warn in advance is avoided, the performance of the equipment is prevented from being damaged, and the service life of the equipment is prolonged.

It will be evident to those skilled in the art that the present application is not limited to the details of the foregoing illustrative embodiments, and that the present application may be embodied in other specific forms without departing from the spirit or essential attributes thereof. The present embodiments are therefore to be considered in all respects as illustrative and not restrictive, the scope of the application being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein.

Finally, it should be noted that the above embodiments are only used for illustrating the technical solutions of the present application and not for limiting, and although the present application is described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that modifications or equivalent substitutions can be made on the technical solutions of the present application without departing from the spirit and scope of the technical solutions of the present application.

Claims

1. A training method for constructing an estimation model for estimating a health state of a plant, the method comprising:

training a first preset model according to the training set;

testing the trained first preset model according to the test set;

determining that the trained first preset model meets preset requirements;

2. The training method of claim 1, wherein the step of forming a training set and a test set comprises:

determining a first quartile and a second quartile of the parameter set or the result set;

forming a quartile distance according to the first quartile and the second quartile;

and forming a health interval according to the first quartile, the second quartile and the quartile distance, and forming the training set and the test set according to the parameter set and the elements of the result set in the health interval.

3. The training method of claim 1, wherein the step of forming a training set and a test set comprises:

determining a mean and a standard deviation of the parameter set and the result set;

and according to the mean value and the standard deviation, carrying out standardization processing on the parameter set and the result set.

4. The training method of claim 1, wherein the step of forming a training set and a test set comprises:

randomly splitting the parameter set and the result set into N samples, wherein N is an integer and is more than 1;

randomly selecting N-1 samples to form the training set;

the remaining 1 of the samples was selected to form the test set.

5. The training method of claim 1, further comprising:

setting random parameters of the trained first preset model based on the fact that the trained first preset model meets preset requirements;

searching the trained first preset model according to the random parameters to form a search result;

determining that the search result meets the requirements;

determining the optimal parameters of the trained first preset model based on the search result meeting the requirements;

and determining the estimation model according to the trained first preset model and the optimal parameters.

6. The training method of claim 1, wherein the electrical information includes current, voltage, and power at which the device is operating, the control information includes rotational speed, load, feed rate, and torque at which the device is operating, and the target information includes spindle temperature.

7. An electronic device comprising a storage medium comprising readable instructions for execution by a processor of a training method as claimed in any one of claims 1-6.

8. A device health diagnostic method for determining a health status of a device, comprising:

9. The device health diagnostic method of claim 8, wherein said step of forming a second set of parameters comprises:

determining a mean and a standard deviation of the first set of parameters;

and normalizing the first parameter set according to the mean and the standard deviation.

10. The device health diagnostic method of claim 8, wherein said step of determining the health status of said device comprises:

inputting the second parameter set to a second preset model, and establishing a distribution model of the second parameter set;

determining a standard deviation of the second set of parameters from the distribution model;

and determining the health state of the equipment according to the target information and the standard deviation.

11. The device health diagnostic method of claim 10, wherein the standard deviation comprises a first standard deviation; the step of determining the health status of the device comprises:

forming a first early warning interval according to the first standard deviation;

and determining that the health state of the equipment is a first state based on the fact that the target information exceeds the first early warning interval.

12. The device health diagnostic method of claim 11, wherein the standard deviation further comprises a second standard deviation; the step of determining the health status of the device further comprises:

forming a second early warning interval according to the second standard deviation, wherein the second early warning interval is within the first early warning interval;

and determining that the health state of the equipment is the second state in the first early warning interval based on the fact that the target information exceeds the second early warning interval.

13. The device health diagnostic method of claim 11, wherein the electrical information includes current, voltage and power at which the device is operating, the control information includes rotational speed, load, feed rate and torque at which the device is operating, and the target information includes spindle temperature.

14. A device health diagnostic apparatus comprising a storage medium including readable instructions for execution by a processor of the device health diagnostic method of any one of claims 8-13.