CN110705133B - Predictive maintenance method and predictive maintenance equipment - Google Patents
Predictive maintenance method and predictive maintenance equipment Download PDFInfo
- Publication number
- CN110705133B CN110705133B CN201911077243.4A CN201911077243A CN110705133B CN 110705133 B CN110705133 B CN 110705133B CN 201911077243 A CN201911077243 A CN 201911077243A CN 110705133 B CN110705133 B CN 110705133B
- Authority
- CN
- China
- Prior art keywords
- data
- prediction model
- output value
- value interval
- equipment
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
Abstract
The invention discloses a predictive maintenance method and a predictive maintenance device, wherein the method comprises the following steps: receiving first data sent by data transmission equipment; classifying the first data, wherein the type of the first data comprises normal data, abnormal data and fault data; and establishing a prediction model by using a variance algorithm and the first data. The invention can provide data support for equipment maintenance, realize predictive maintenance of the equipment in practical significance, reduce maintenance cost and down time, and avoid production loss and material waste caused by equipment failure.
Description
Technical Field
The invention relates to the technical field of equipment maintenance, in particular to a predictive maintenance method and predictive maintenance equipment.
Background
Enterprises inevitably have the situation of unplanned shutdown, thereby causing production loss and material waste. Unscheduled shutdowns are generally caused by equipment failures, and therefore, how to maintain industrial equipment is critical. At present, the maintenance of industrial equipment is roughly divided into 3 means, namely, repairable maintenance, preventive maintenance and predictive maintenance, wherein the predictive maintenance belongs to the maintenance in advance, and based on various sensors installed on the equipment, various data such as the running state information and the environmental information of the equipment are utilized, and based on a mathematical statistical model, faults are predicted, so that the maintenance work of the equipment is guided, meanwhile, the running state of the equipment is monitored in real time, and the time when the faults occur is judged more accurately.
At present, in the prior art, the maintenance of the industrial equipment is only in the state maintenance stage, mainly the functional alarm for equipment state identification and monitoring is given, the fault diagnosis and the comprehensive guidance of maintenance management of the equipment cannot be realized, and the predictive maintenance of the industrial equipment cannot be realized in the true sense. And the equipment data analysis and processing process is usually completed in an equipment controller, complex data processing cannot be carried out, and the requirement of predictive maintenance of the equipment in the practical sense cannot be met.
Therefore, a method and apparatus for predictive maintenance are needed to implement predictive maintenance of industrial equipment in a practical sense.
Disclosure of Invention
Therefore, the invention provides a predictive maintenance method based on edge calculation and a predictive maintenance device, so as to solve the problem that the predictive maintenance requirement of the device in the practical sense cannot be met due to the lack of the predictive maintenance method in the prior art.
To achieve the above object, a first aspect of the present invention provides a method for predictive maintenance based on edge calculation, comprising:
receiving first data sent by data transmission equipment;
classifying the first data, wherein the type of the first data comprises normal data, abnormal data and fault data;
and establishing a prediction model by using a variance algorithm and the first data.
Preferably, the variance algorithm uses the formula:
x, Y … … Z is an equipment parameter, Xp and Yp … … Zp are respectively the average value of the equipment parameter of X, Y … … Z in a time length p, Xq and Yq … … Zq are respectively the average value of the equipment parameter of X, Y … … Z in a time length q, U1 and U2 … … Un are preset weights, n is the number of X, Y … … Z, and b is a preset constant;
the above establishing a prediction model by using the variance algorithm and the first data includes:
inputting data in two different preset durations into a variance algorithm formula for normal data, abnormal data and fault data for multiple times respectively to obtain a normal data output value interval, an abnormal data output value interval and a fault data output value interval respectively;
and generating a prediction model according to the normal data output value interval, the abnormal data output value interval and the fault data output value interval.
Preferably, the method for predictive maintenance provided by the present invention further comprises, after establishing the prediction model using the variance algorithm and the first data:
verifying the prediction model according to verification data, wherein the verification data comprises abnormal data and fault data;
and releasing the preset model in response to the verification result of the prediction model meeting the preset condition.
Preferably, the verifying the prediction model according to the verification data includes:
inputting the verification data into a prediction model to obtain a prediction result;
judging whether the types of the prediction result and the verification data are matched, and calculating the prediction accuracy of the prediction model according to the judgment result;
the verification result of the prediction model meets a preset condition, and the verification result comprises the following steps: the prediction accuracy is greater than or equal to a preset threshold value.
Preferably, the method for predictive maintenance provided by the present invention further comprises: responding to the verification result of the prediction model not meeting the preset condition, adjusting the prediction model, and verifying the adjusted prediction model until the adjusted prediction model meets the preset condition;
the adjusting the prediction model includes:
keeping the normal data output value interval, the abnormal data output value interval and the fault data output value interval unchanged, and adjusting the weights U1 and U2 … … Un and the constant b in the variance algorithm formula.
As a second aspect of the present invention, there is provided a predictive maintenance apparatus comprising: the device comprises a receiving module, a classifying module and an establishing module, wherein the receiving module is used for receiving first data sent by data transmission equipment;
the classification module is used for classifying the first data, wherein the type of the first data comprises normal data, abnormal data and fault data;
the establishing module is used for establishing a prediction model by using a variance algorithm and the first data.
Preferably, the variance algorithm uses the formula:
wherein X, Y … … Z is equipment parameter, Xp and Yp … … Zp are X, Y … respectively… Z, Xq and Yq … … Zq are respectively the average values of X, Y … … Z equipment parameters in the duration q, U1 and U2 … … Un are preset weights, n is the number of X, Y … … Z, and b is a preset constant;
the establishing module is used for: inputting data in two different preset durations into a variance algorithm formula for multiple times according to normal data, abnormal data and fault data respectively to obtain a normal data output value interval, an abnormal data output value interval and a fault data output value interval respectively; and generating a prediction model according to the normal data output value interval, the abnormal data output value interval and the fault data output value interval.
Preferably, the predictive maintenance device provided by the invention further comprises a verification module and a release module, wherein the verification module is used for verifying the prediction model according to verification data, and the verification data comprises abnormal data and fault data;
the issuing module is used for responding to the fact that the verification result of the prediction model meets the preset condition and issuing the preset model.
Preferably, the verification module is used for inputting verification data into the prediction model to obtain a prediction result; judging whether the types of the prediction result and the verification data are matched, and calculating the prediction accuracy of the prediction model according to the judgment result; the verification result of the prediction model meets a preset condition, and the verification result comprises the following steps: the prediction accuracy is greater than or equal to a preset threshold value.
Preferably, the predictive maintenance device provided by the present invention further includes an adjusting module, where the adjusting module is configured to adjust the prediction model in response to a verification result of the prediction model not meeting a preset condition, and verify the adjusted prediction model until the adjusted prediction model meets the preset condition; wherein adjusting the predictive model comprises: keeping the normal data output value interval, the abnormal data output value interval and the fault data output value interval unchanged, and adjusting the weights U1 and U2 … … Un and the constant b in the variance algorithm formula.
The present invention provides a method of predictive maintenance, the method comprising: receiving first data sent by data transmission equipment; classifying the first data, wherein the type of the first data comprises normal data, abnormal data and fault data; and establishing a prediction model by using a variance algorithm and the first data. The method adopts the first data to carry out mathematical modeling, the data source is reliable and has certain reference value, the operation state of the equipment can be effectively distinguished according to the data for the first data classification, the accuracy of the prediction model can be provided by adopting the variance algorithm to carry out the mathematical modeling, the data support can be provided for equipment maintenance by utilizing the variance algorithm and the first data to establish the prediction model, the predictive maintenance of the equipment in practical significance is realized, the maintenance cost is reduced, the downtime is reduced, and the production loss and the material waste caused by the equipment failure are avoided.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the principles of the invention and not to limit the invention.
FIG. 1 is a schematic diagram of a system architecture according to an embodiment of the present invention;
FIG. 2 is a flow chart illustrating a method for predictive maintenance according to an embodiment of the present invention;
FIG. 3 is a schematic flow chart of establishing a prediction model according to an embodiment of the present invention;
FIG. 4 is a schematic flow chart illustrating a verification of a predictive model according to an embodiment of the present invention;
FIG. 5a is a schematic structural diagram of a predictive maintenance device according to an embodiment of the present invention;
FIG. 5b is a second schematic structural diagram of a predictive maintenance apparatus according to an embodiment of the present invention;
fig. 5c is a third schematic structural diagram of a predictive maintenance apparatus according to an embodiment of the present invention.
Detailed Description
The following describes in detail embodiments of the present invention with reference to the drawings. It should be understood that the detailed description and specific examples, while indicating the preferred embodiment of the invention, are given by way of illustration and explanation only, not limitation.
The invention provides a predictive maintenance method based on edge calculation, which is applied to a system shown in FIG. 1, and the system comprises the following steps: various industrial devices (device 1, device 2 … …, device n), PLCs (Programmable Logic controllers), industrial controllers, smart meters, predictive maintenance devices, and cloud servers.
The PLC, the industrial controller and the intelligent instrument are used for monitoring equipment parameters of various industrial equipment in a production process, the data transmission equipment is used for acquiring the equipment parameters of the various industrial equipment from the PLC, the industrial controller and the intelligent instrument and sending the equipment parameters to the predictive maintenance equipment through a network (the network can be but is not limited to a 5G network), the predictive maintenance equipment is used for establishing a prediction model according to the equipment parameters and deploying the prediction model to the cloud server, and the cloud server is used for carrying out application such as prediction analysis, management and early warning of the various industrial equipment according to the prediction model.
As shown in fig. 2, the present invention provides a method for predictive maintenance, which includes the following steps:
step S201, receiving first data sent by the data transmission device.
The first data may be a device parameter that can provide a reference value for monitoring the operating state of the device, for example, the first data may be production device data, plant environment data, video monitoring data, and the like. Specifically, the production equipment data may include normal production data and abnormal production data, and the plant environment data may include temperature, humidity, pressure, and other data.
In the invention, the data transmission equipment can acquire the first data through various sensors and instruments such as a PLC, an industrial controller, an intelligent instrument and the like. In addition, the data transmission equipment acquires the first data through various sensors and instruments and meters, and can also monitor the equipment state, for example, if certain equipment does not send equipment data to the data transmission equipment within a certain preset fixed time period (which can be set by the data transmission equipment according to actual production requirements), the data transmission equipment can give an alarm, and the real-time performance and the normality of data acquisition are ensured.
The data transmission device may transmit the first data in real time through the 5G network. The 5G network can ensure the characteristics of high quality transmission rate and low time delay real-time linear performance, and the data transmission equipment transmits the first data through the 5G network, so that continuous monitoring on various sensors and instruments can be realized, the real-time monitoring requirement in the production process is met, the first data can be acquired and transmitted in real time with high precision and low time delay, and the real-time requirement of scenes such as high-speed operation, precision production, motion control and the like of a production line on data acquisition is met. It should be noted that the present invention is not limited to be applied to a 5G network.
Step S102, classifying the first data, wherein the type of the first data comprises normal data, abnormal data and fault data.
In the embodiment of the invention, the first data are classified according to the monitored operation states of the industrial equipment and the device, so that the first data are divided into normal data, abnormal data and fault data.
Specifically, after the predictive maintenance device receives the first data sent by the data transmission device, the operation states of various industrial devices and apparatuses can be deeply analyzed based on the big data. If the equipment can normally produce in a certain time period, which indicates that the equipment can normally operate, dividing the first data of the equipment in the time period into normal data. If the equipment can produce in a certain time period but is not enough to meet the normal production requirement, the equipment is abnormal in operation, and the first data of the equipment in the time period are divided into abnormal data. If the equipment cannot be produced in a certain time period and cannot meet the production requirement, and the equipment is failed, dividing the first data of the equipment in the time period into failure data. In addition, if the equipment fails and then sends out alarm information, the first data of the equipment in the time period for sending out the alarm information is also divided into failure data.
In an embodiment of the present invention, after classifying the first data, the predictive maintenance device may first store the first data in the device data storage table in real time, where the device data storage table stores contents such as device code, device type, device manufacturer and address, device status information, environment information, and time. And judging whether the data in the equipment data storage table is fault data, if so, storing the fault data into the equipment fault data storage table, wherein the equipment fault data storage table can store the contents of equipment codes, equipment types, equipment manufacturers and addresses, equipment state information, environment information, time, fault times, fault reasons and the like. And further judging whether the data in the equipment data storage table is abnormal data or not, if so, storing the abnormal data into the equipment abnormal data storage table, wherein the equipment abnormal data storage table stores the contents such as equipment codes, equipment types, equipment manufacturers and addresses, equipment state abnormal information, environment abnormal information, time and the like.
After the first data is stored in a classified manner, only normal data is reserved in the device data storage table, and abnormal data and fault data are stored in the device abnormal data storage table and the device fault data storage table respectively. The first data are classified, and different types of data are stored respectively, so that the complexity of the subsequent data processing process is reduced.
Step S103, a prediction model is established by using a variance algorithm and the first data.
In the embodiment of the invention, the first data can be input into the variance algorithm formula, the types of the first data comprise normal data, abnormal data and fault data, the three data can correspond to different calculation results, and a prediction model can be established by integrating three different settlement results, so that the type of the input data can be judged according to the output result of the prediction model when the prediction model is officially applied.
The first data come from an actual production scene, so that the reference value is high. A linear model is adopted in a general prediction model, the accuracy of the prediction model can be improved by adopting a variance algorithm to establish the prediction model, and the prediction model can be optimized according to actual conditions.
The predictive maintenance means that the fault is predicted based on a mathematical statistical model by using data such as state information, environmental information and the like of equipment operation. After an accurate prediction model is established by using the variance algorithm and the first data, the prediction model can be deployed into an actual production system to operate, and the method is suitable for any scenes and requirements of predictive maintenance application. If the predictive maintenance finds a fault hidden trouble according to the prediction model, an alarm or a repair command can be automatically triggered, the predictive maintenance can be used for optimizing production operation, and 20% -30% of efficiency gain can be brought.
It can be seen from the above steps S101-S103 that, in the present invention, first data sent by a data transmission device is received, the first data is classified, wherein the type of the first data includes normal data, abnormal data and fault data, and a prediction model is established by using a variance algorithm and the first data. The method adopts the first data to carry out mathematical modeling, the data source is reliable and has certain reference value, the operation state of the equipment can be effectively distinguished according to the data for the first data classification, the accuracy of the prediction model can be provided by adopting the variance algorithm to carry out the mathematical modeling, the data support can be provided for equipment maintenance by utilizing the variance algorithm and the first data to establish the prediction model, the predictive maintenance of the equipment in practical significance is realized, the maintenance cost is reduced, the downtime is reduced, and the production loss and the material waste caused by the equipment failure are avoided.
It should be noted that, for different actual production scenarios, the first data sent by the data transmission device is generally raw device data that is not preprocessed. Moreover, data standards or proprietary communication protocols adopted by various industrial devices and apparatuses may not be uniform, so that the problems that the original device data may have non-uniform standards and protocols cannot be adapted are caused. Further, directly analyzing and processing the first data may waste computing resources and affect the accuracy of the computing result.
In the invention, after receiving the first data sent by the data transmission equipment, the data protocol conversion can be carried out on the first data. Specifically, OPC UA (OLE for Process Control, Unified Architecture) technology may be used to convert data standards or proprietary communication protocols adopted by various industrial devices and apparatuses into standard OPC UA communication protocols. OPC UA can enable the transmission of raw data and preprocessed information from the manufacturing level to the production Planning or ERP (Enterprise Resource Planning) level. Through OPC UA, carry out standardized message to the different message structure's of heterogeneous field bus and ethernet bus data and disassemble, satisfy the equipment data standard unification, the agreement adaptation of various industrial equipment and device and data interconnection intercommunication.
In the invention, after the data protocol conversion is carried out on the first data, the first data can be preprocessed. Specifically, data representing the same type of equipment parameters (such as time, temperature or humidity) in the first data is converted into data with a unified measurement unit, null data and data which do not conform to actual conditions are filtered, and the data of the same type are merged or reconstructed. The preprocessing of the first data can reduce the complexity of calculation, save calculation resources and avoid influencing the accuracy of the calculation result.
Further, the variance algorithm of the present invention may employ the formula:
wherein X, Y … … Z is an equipment parameter, Xp and Yp … … Zp are respectively the average values of the X, Y … … Z equipment parameters in the duration p, Xq and Yq … … Zq are respectively the average values of the X, Y … … Z equipment parameters in the duration q, U1 and U2 … … Un are preset weights, n is the number of X, Y … … Z, and b is a preset constant.
As shown in fig. 3, the establishing of the prediction model by using the variance algorithm and the first data in the present invention may include the following steps:
step S301, inputting the data in two different preset time lengths into a variance algorithm formula for multiple times respectively aiming at the normal data, the abnormal data and the fault data to respectively obtain a normal data output value interval, an abnormal data output value interval and a fault data output value interval.
It should be noted that, in the present invention, the device parameters X, Y … … Z input into the variance algorithm formula for two different preset time periods include, but are not limited to, temperature, humidity, pressure, size, type, density, etc
Specifically, as an embodiment, for the normal data, the data with the first preset time length of 24 hours and the data with the second preset time length of 48 hours are input into the variance algorithm formula. Specifically, first, the average value of the 24-hour data of the device parameter independent variable X in the 24-hour data is calculated: x 24 ＝(X 1 +X 2 +X 3 +……X 24 ) 24, mean value of 24 hours data of the plant parameter independent variable Y: y is 24 ＝(Y 1 +Y 2 +Y 3 +……Y 24 ) Average value Z of 24-hour data of the/24 … … plant parameter independent variable Z 24 ＝(Z 1 +Z 2 +Z 3 +……Z 24 ) 24; similarly, X in the data with the calculation time of 48 hours 48 、Y 48 ……Z 48 (ii) a Finally, X is 24 、Y 24 ……Z 24 And X 48 、Y 48 ……Z 48 And the number n of the equipment parameters X, Y … … Z, and inputting the n into the variance algorithm formula to obtain an output value F (U, X) of the variance algorithm formula 24 ，Y 24 ……Z 24 ，X 48 ，Y 48 ……Z 48 ) The output value is a primary output value of normal data.
For normal data, data in two different preset durations are selected for multiple times to input the variance algorithm formula for multiple times, for example, data with durations of 2 hours and 12 hours can be input for the second time, data with durations of 16 hours and 32 hours can be input for the third time, and the like, output values of the normal data are calculated respectively, and finally output values of the normal data are integrated, so that an output value range of the normal data is obtained.
The input method of the abnormal data and the fault data is the same as that of the normal data, and is not described herein.
It should be noted that, because the data size is extremely large, there may be an intersection between the output value range of the normal data, the output value range of the abnormal data, and the output value range of the fault data, and it is necessary to manually perform fine adjustment on the three output value ranges, so that there is no intersection between every two of the three output value ranges, and then a normal data output value interval, an abnormal data output value interval, and a fault data output value interval can be obtained.
Step S302, a prediction model is generated according to the normal data output value interval, the abnormal data output value interval and the fault data output value interval.
Specifically, the normal data output value interval, the abnormal data output value interval, and the fault data output value interval are obtained from the first data subjected to the data protocol conversion, the preprocessing, and the classification, and then the variance algorithm formula, the normal data output value interval, the abnormal data output value interval, and the fault data output value interval may constitute a prediction model. When the prediction model is used, data is input into a variance algorithm formula, and the specific position of the output value in a normal data output value interval, an abnormal data output value interval or a fault data output value interval is judged, so that whether the data is normal data, abnormal data or fault data can be judged, the normal degree, the abnormal degree and the fault degree of the data are further judged, and the purpose of predicting equipment faults is achieved.
Further, the method may further include the following steps after the predictive model is established by using the variance algorithm and the first data: verifying the predictive model based on verification data, the verification data including anomaly data and fault data. And responding to the fact that the verification result of the prediction model meets a preset condition, and releasing the preset model.
It should be noted that the source of the verification data is not particularly limited, for example, the abnormal data and the failure data in the first data may be used to verify the prediction model to reduce the computation flow and save the computation resources, and further, the abnormal data and the failure data that are not used in the building of the prediction model in the first data may be used to verify the prediction model more accurately.
In the embodiment of the invention, the abnormal data and the fault data are respectively input into a variance algorithm formula to obtain two groups of output results. And comparing the output result with the abnormal data output value interval and the fault data output value interval to obtain a verification result. And setting a condition for the verification result of the prediction model according to the actual production condition in advance, and if the verification result meets the preset condition, indicating that the prediction model can achieve the expected predictive maintenance effect, releasing and deploying the prediction model to a cloud server or a predictive maintenance platform for operation, continuously receiving data acquired by a device layer, and inputting the prediction model for application such as predictive analysis, management, early warning and the like.
Further, as shown in fig. 4, the verifying the prediction model according to the verification data of the present invention may include the following steps:
step S401, inputting the verification data into the prediction model to obtain a prediction result.
Specifically, when the variance algorithm formula is used for calculation, an output value of the variance formula can be obtained through each independent calculation. And inputting data of two different acquisition periods into the variance algorithm formula for multiple times aiming at the abnormal data and the fault data to respectively obtain a series of abnormal data output values and fault data output values.
Step S402, judging whether the prediction result is matched with the type of the verification data, and calculating the prediction accuracy of the prediction model according to the judgment result.
Specifically, it is determined whether a series of abnormal data output values and fault data output values obtained in step S401 fall in the above-mentioned abnormal data output value interval and fault data output value interval, and the prediction accuracy of the prediction model is calculated according to the number of times that the abnormal data output values fall in the abnormal data output value interval and the total verification number of times, and the number of times that the fault data output values fall in the fault data output value interval. In calculating the prediction accuracy of the prediction model. The corresponding accuracy rates can be respectively calculated according to the times of the abnormal data output values falling in the abnormal data output value interval and the times of the fault data output values falling in the fault data output value interval, and the overall accuracy rate can also be calculated according to the total times of the abnormal data output values falling in the abnormal data output value interval and the fault data output values falling in the fault data output value interval. Specifically, taking the case of calculating the overall accuracy according to the total times as an example, if the prediction model is verified 50 times for abnormal data, wherein 40 times of abnormal data output values all fall within the abnormal data output value interval, and 50 times for fault data, wherein 45 times of fault data output values all fall within the fault data output value interval, the prediction accuracy of the prediction model may be calculated by: (40+45)/(50+50) × 100% ═ 85%, in which case the accuracy of the predictive model may be 85%.
Further, the verification result of the prediction model in the invention meets the preset conditions, including: the prediction accuracy is greater than or equal to a preset threshold value.
It should be noted that the same threshold value may be set for the abnormal data and the failure data, or two different threshold values may be set. Specifically, for example, in the case where the overall accuracy is calculated according to the total number of times, a threshold value may be preset to be 90%, and since the accuracy of the prediction model is 85%, the verification result of the prediction model does not satisfy the preset condition.
Further, the present invention may further include the following steps after verifying the prediction model based on the verification data: and responding to the verification result of the prediction model not meeting the preset condition, adjusting the prediction model, and verifying the adjusted prediction model until the adjusted prediction model meets the preset condition.
Specifically, if the verification result of the prediction model does not satisfy the preset condition, it indicates that the prediction model cannot accurately detect the verification data.
If the abnormal data and the fault data which are not used when the prediction model is established are used for verifying the prediction model, the reason that the prediction model cannot accurately detect the verification data may be that the data has problems in terms of protocols and formats, and the preset weights U1, U2 … … Un and the constant b are not suitable for verifying equipment parameters of equipment corresponding to the data. Therefore, an iterative process is required.
Specifically, as an implementation manner, first data sent by the data transmission device may be received, data protocol conversion and preprocessing are performed on the first data, the first data is classified into normal data, abnormal data and fault data, and the prediction model is adjusted.
Further, the adjusting the prediction model according to the present invention may include: keeping the normal data output value interval, the abnormal data output value interval and the fault data output value interval unchanged, and adjusting the weights U1 and U2 … … Un and the constant b in the variance algorithm formula.
Specifically, the data in two different preset durations are input into a variance algorithm formula respectively aiming at normal data, abnormal data and fault data, the normal data output value interval, the abnormal data output value interval and the fault data output value interval obtained in model building are kept unchanged, and corresponding weights U1, U2 … … Un and a constant b are adjusted according to the actual source condition of the equipment for inputting the data, so that the output values still fall in the corresponding output value intervals when the data in the two different preset durations are input.
And adjusting the prediction model for multiple times and verifying the adjusted prediction model according to verification data until the adjusted prediction model meets the preset condition, otherwise, continuously adjusting and verifying the prediction model. The method for verifying the predictive model based on the verification data is described in detail above and will not be described herein.
Based on the same technical concept, the embodiment of the present invention further provides a predictive maintenance device, as shown in fig. 5a, the platform may include: a receiving module 501, a classification module 502 and a building module 503.
The receiving module 501 is configured to receive first data sent by a data transmission device.
The classification module 502 is configured to classify the first data, wherein the type of the first data includes normal data, abnormal data, and failure data.
The building block 503 is configured to build a prediction model using a variance algorithm and the first data.
Preferably, the variance algorithm uses the formula:
x, Y … … Z is an equipment parameter, Xp and Yp … … Zp are respectively the average value of the equipment parameter of X, Y … … Z in the time length p, Xq and Yq … … Zq are respectively the average value of the equipment parameter of X, Y … … Z in the time length q, U1 and U2 … … Un are preset weights, n is the number of X, Y … … Z, and b is a preset constant.
The establishing module 503 is configured to input data within two different preset durations into a variance algorithm formula for multiple times, respectively aiming at the normal data, the abnormal data and the fault data, so as to obtain a normal data output value interval, an abnormal data output value interval and a fault data output value interval, respectively; and generating a prediction model according to the normal data output value interval, the abnormal data output value interval and the fault data output value interval.
Preferably, as shown in fig. 5b, the platform may further include: a verification module 504 and an issuance module 505.
The verification module 504 is configured to verify the prediction model based on verification data, which includes abnormal data and fault data.
The issuing module 505 is configured to issue the preset model in response to a verification result of the prediction model meeting a preset condition.
Preferably, as shown in fig. 5b, the verification module 504 is configured to input verification data into the prediction model to obtain a prediction result; judging whether the types of the prediction result and the verification data are matched, and calculating the prediction accuracy of the prediction model according to the judgment result; the verification result of the prediction model meets a preset condition, and the verification result comprises the following steps: the prediction accuracy is greater than or equal to a preset threshold value.
Preferably, as shown in fig. 5c, the platform may further include: an adjustment module 506.
The adjusting module 506 is configured to adjust the prediction model in response to that the verification result of the prediction model does not satisfy the preset condition, and verify the adjusted prediction model until the adjusted prediction model satisfies the preset condition.
Wherein adjusting the predictive model comprises: keeping the normal data output value interval, the abnormal data output value interval and the fault data output value interval unchanged, and adjusting the weights U1 and U2 … … Un and the constant b in the variance algorithm formula.
The predictive maintenance equipment provided by the invention can ensure data security, prevent the data from being invaded by potential attackers, ensure information security, utilize the calculation and storage functions of users and edge network equipment in the network, bear the control, management and service functions in part of core nodes, can improve the traditional mobile broadband service capability and deal with emerging machine services, can meet wide use cases and business models in a 5G network, and enable an operator to flexibly provide personalized network services for the users at low cost according to the requirements of a third party and the network condition.
It will be understood that the above embodiments are merely exemplary embodiments taken to illustrate the principles of the present invention, which is not limited thereto. It will be apparent to those skilled in the art that various changes and modifications can be made therein without departing from the spirit and scope of the invention, and these changes and modifications are to be considered as the scope of the invention.
Claims (8)
1. A method of predictive maintenance, the method comprising:
receiving first data sent by data transmission equipment;
classifying the first data, wherein the type of the first data comprises normal data, abnormal data and fault data;
establishing a prediction model by using a variance algorithm and the first data;
the variance algorithm adopts a formula as follows: f (U, Xp, Xq, Yp, Yq … … Zp, Zq) =
+ b, wherein X, Y … … Z is the equipment parameter, Xp and Yp … … Zp are the average values of X, Y … … Z equipment parameters in the time length p, and Xq and Yq … … Zq are divided intoAverage values of equipment parameters which are X, Y … … Z in the time length q, U1 and U2 … … Un are preset weights, n is the number of X, Y … … Z, and b is a preset constant;
the building a prediction model using a variance algorithm and the first data comprises:
inputting data in two different preset durations into the variance algorithm formula for multiple times respectively aiming at normal data, abnormal data and fault data to respectively obtain a normal data output value interval, an abnormal data output value interval and a fault data output value interval;
and generating the prediction model according to the normal data output value interval, the abnormal data output value interval and the fault data output value interval.
2. The method of predictive maintenance according to claim 1, further comprising, after said building a predictive model using a variance algorithm and said first data:
verifying the prediction model according to verification data, wherein the verification data comprises abnormal data and fault data;
and responding to the fact that the verification result of the prediction model meets the preset condition, and releasing the prediction model.
3. The method of predictive maintenance according to claim 2, wherein said validating said predictive model based on validation data comprises:
inputting the verification data into the prediction model to obtain a prediction result;
judging whether the types of the prediction result and the verification data are matched, and calculating the prediction accuracy of the prediction model according to the judgment result;
the verification result of the prediction model meets a preset condition, and the verification result comprises the following steps: the prediction accuracy is greater than or equal to a preset threshold.
4. The method of predictive maintenance according to claim 2, further comprising: responding to the verification result of the prediction model not meeting the preset condition, adjusting the prediction model, and verifying the adjusted prediction model until the adjusted prediction model meets the preset condition;
the adjusting the prediction model includes:
keeping the normal data output value interval, the abnormal data output value interval and the fault data output value interval unchanged, and adjusting the weights U1 and U2 … … Un and the constant b in the variance algorithm formula.
5. A predictive maintenance device, the device comprising a receiving module, a classification module, and an establishment module, the receiving module configured to receive first data transmitted by a data transmission device;
the classification module is used for classifying the first data, wherein the type of the first data comprises normal data, abnormal data and fault data;
the establishing module is used for establishing a prediction model by utilizing a variance algorithm and the first data;
the variance algorithm adopts a formula as follows: f (U, Xp, Xq, Yp, Yq … … Zp, Zq) =
+ b, wherein X, Y … … Z is an equipment parameter, Xp and Yp … … Zp are respectively the average values of the equipment parameter of X, Y … … Z in the time length p, Xq and Yq … … Zq are respectively the average values of the equipment parameter of X, Y … … Z in the time length q, U1 and U2 … … Un are preset weights, n is the number of X, Y … … Z, and b is a preset constant;
the establishing module is used for: inputting data in two different preset durations into the variance algorithm formula for multiple times respectively aiming at normal data, abnormal data and fault data to respectively obtain a normal data output value interval, an abnormal data output value interval and a fault data output value interval; and generating the prediction model according to the normal data output value interval, the abnormal data output value interval and the fault data output value interval.
6. The predictive maintenance device of claim 5, further comprising a validation module and a publication module, the validation module configured to validate the predictive model based on validation data, the validation data including anomaly data and fault data;
the issuing module is used for responding to the fact that the verification result of the prediction model meets the preset condition and issuing the prediction model.
7. The predictive maintenance device of claim 6, wherein the validation module is configured to input the validation data into the predictive model to obtain a predicted result; judging whether the types of the prediction result and the verification data are matched, and calculating the prediction accuracy of the prediction model according to the judgment result; the verification result of the prediction model meets a preset condition, and the verification result comprises the following steps: the prediction accuracy is greater than or equal to a preset threshold.
8. The predictive maintenance device of claim 6, further comprising an adjustment module configured to, in response to a result of the verification of the predictive model not satisfying a preset condition, adjust the predictive model and verify the adjusted predictive model until the adjusted predictive model satisfies the preset condition; wherein adjusting the predictive model comprises: keeping the normal data output value interval, the abnormal data output value interval and the fault data output value interval unchanged, and adjusting the weights U1 and U2 … … Un and the constant b in the variance algorithm formula.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911077243.4A CN110705133B (en) | 2019-11-06 | 2019-11-06 | Predictive maintenance method and predictive maintenance equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911077243.4A CN110705133B (en) | 2019-11-06 | 2019-11-06 | Predictive maintenance method and predictive maintenance equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110705133A CN110705133A (en) | 2020-01-17 |
| CN110705133B true CN110705133B (en) | 2022-09-23 |
Family
ID=69205427
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911077243.4A Active CN110705133B (en) | 2019-11-06 | 2019-11-06 | Predictive maintenance method and predictive maintenance equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110705133B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108664402A (en) * | 2018-05-14 | 2018-10-16 | 北京航空航天大学 | A kind of failure prediction method based on software network feature learning |
| CN109507989A (en) * | 2018-12-24 | 2019-03-22 | 上海应用技术大学 | A kind of method of unmanned plane sensor fault diagnosis |
| CN110320892A (en) * | 2019-07-15 | 2019-10-11 | 重庆邮电大学 | The sewage disposal device fault diagnosis system and method returned based on Lasso |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160245279A1 (en) * | 2015-02-23 | 2016-08-25 | Biplab Pal | Real time machine learning based predictive and preventive maintenance of vacuum pump |
-
2019
- 2019-11-06 CN CN201911077243.4A patent/CN110705133B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108664402A (en) * | 2018-05-14 | 2018-10-16 | 北京航空航天大学 | A kind of failure prediction method based on software network feature learning |
| CN109507989A (en) * | 2018-12-24 | 2019-03-22 | 上海应用技术大学 | A kind of method of unmanned plane sensor fault diagnosis |
| CN110320892A (en) * | 2019-07-15 | 2019-10-11 | 重庆邮电大学 | The sewage disposal device fault diagnosis system and method returned based on Lasso |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110705133A (en) | 2020-01-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CA2564724C (en) | Automatic remote monitoring and diagnostics system and communication method for communicating between a programmable logic controller and a central unit | |
| US10909137B2 (en) | Streaming data for analytics in process control systems | |
| CN107390636B (en) | Operation monitoring method, maintenance method and system for numerical control equipment | |
| US11734213B2 (en) | Integration of multiple communication physical layers and protocols in a process control input/output device | |
| CN105122162B (en) | Control system and method for the remote control of nextport hardware component NextPort | |
| JP2013012229A (en) | Shared-use data processing for process control systems | |
| US20140277604A1 (en) | Distributed big data in a process control system | |
| CN102375453A (en) | Seamless integration of process control devices in a process control environment | |
| WO2021119140A1 (en) | Centralized knowledge repository and data mining system | |
| US20220156704A1 (en) | Method and system for transmitting data from manufacturing data originators to target computing systems using defined message schemas | |
| CN112995327A (en) | Cloud platform system, factory end, industrial cloud platform and service docking system | |
| KR20210072473A (en) | Data collection gateway having AI module | |
| KR102418965B1 (en) | Apparatus for Interfacing, Semiconductor Manufacturing System and Management Method | |
| KR20220122922A (en) | Data Logger System based on Artificial Intelligence | |
| Al Sunny et al. | Remote monitoring and online testing of machine tools for fault diagnosis and maintenance using MTComm in a cyber-physical manufacturing cloud | |
| CN110705133B (en) | Predictive maintenance method and predictive maintenance equipment | |
| Dang et al. | On the road to Industry 4.0: a fieldbus architecture to acquire specific smart instrumentation data in existing industrial plant for predictive maintenance | |
| Chiorean et al. | Modelling an OPC UA client application for predictive maintenance support | |
| CN113848834B (en) | Workshop equipment access system and method based on edge cloud cooperation | |
| US11595409B2 (en) | Method for monitoring an industrial network | |
| CN115118578A (en) | SCADA system based on WEB | |
| US20200241488A1 (en) | Controller Optimization for a Control System of a Technical Plant | |
| Carvalho et al. | A framework for dependability evaluation of PROFIBUS networks | |
| US11237550B2 (en) | Ultrasonic flow meter prognostics with near real-time condition based uncertainty analysis | |
| CN113759849A (en) | Intelligent operation and maintenance service support system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |