CN108764498B - Method and device for acquiring health index of equipment - Google Patents
Method and device for acquiring health index of equipment Download PDFInfo
- Publication number
- CN108764498B CN108764498B CN201810511519.4A CN201810511519A CN108764498B CN 108764498 B CN108764498 B CN 108764498B CN 201810511519 A CN201810511519 A CN 201810511519A CN 108764498 B CN108764498 B CN 108764498B
- Authority
- CN
- China
- Prior art keywords
- data
- sensor
- equipment
- current
- health index
- 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
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
-
- 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
- G06Q10/20—Administration of product repair or maintenance
Landscapes
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Engineering & Computer Science (AREA)
- Strategic Management (AREA)
- Economics (AREA)
- Entrepreneurship & Innovation (AREA)
- Physics & Mathematics (AREA)
- Quality & Reliability (AREA)
- Operations Research (AREA)
- Tourism & Hospitality (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Development Economics (AREA)
- Educational Administration (AREA)
- Game Theory and Decision Science (AREA)
- Medical Treatment And Welfare Office Work (AREA)
- Measuring And Recording Apparatus For Diagnosis (AREA)
Abstract
The invention discloses a method and a device for acquiring a health index of equipment, wherein the method comprises the following steps: acquiring each piece of historical record data of equipment to be evaluated and equipment states corresponding to the record; training a preset naive Bayes classification model according to each record contained in the historical record data and the equipment state corresponding to the record; acquiring current recording data of equipment to be evaluated, and acquiring a corresponding equipment state by using a naive Bayes classification model; acquiring an upper quartile, a lower quartile and a first intermediate value corresponding to each sensor in the current state; calculating the abnormality degree corresponding to the sensor; acquiring a first health index corresponding to a sensor according to a preset corresponding function of the abnormality degree corresponding to the sensor and the first health index of the sensor; and calculating a second health index of the equipment to be evaluated according to the weighted sum of the first health indexes corresponding to all the sensors. By applying the embodiment of the invention, the accuracy of the health index of the equipment to be evaluated can be improved.
Description
Technical Field
The invention relates to a device evaluation method and device, in particular to a method and device for acquiring a health index of a device.
Background
In modern enterprises, equipment is the material technology basis for daily production. The technical state of the equipment and the quality of industrial equipment are closely related to the quantity and quality of products produced by enterprises and other technical and economic indexes, which puts severe requirements on the reliability of mechanical equipment. How to reduce the fault risk and improve the comprehensive efficiency of the equipment is that the equipment is in an ideal state for a long time, is important research content of enterprises, and has important significance for ensuring the rapid development of the enterprises. This is of great importance for the discovery of equipment faults as an important basis for equipment maintenance. The method for discovering the fault of the equipment comprises the following steps: the technical personnel observe whether the equipment generates smoke, sparks and the like according to experience; whether the equipment has noise in the running process or not and whether abnormal temperature, such as burnt smell, is smelled in the running process of the equipment or not; whether some parts of the equipment are overheated or vibrate violently or not in the running process of the equipment; and then according to the diagnosis result, the current health condition of the equipment to be evaluated is comprehensively judged manually. However, in the prior art, the equipment to be evaluated is evaluated manually, and a technician cannot perform the above diagnosis on the equipment to be evaluated in real time, so that the manual evaluation has a technical problem of poor real-time performance.
At present, in order to improve the real-time performance of the health state assessment of the equipment, the current states of all parts of the equipment can be detected by using sensors installed at all parts of the equipment to be assessed, and then the measurement data of the sensors of all the parts are weighted and summed according to the weight of all the parts, so as to obtain the health index of the equipment to be assessed.
However, the device to be evaluated may have a plurality of different working states, which may cause a large difference between device parameters corresponding to the respective working states, and the device parameters corresponding to the respective working states do not continuously change, which may cause a large fluctuation of the health index of the device to be evaluated, and it is difficult to determine whether the device to be evaluated is in a normal state, so that the technical problem in the prior art that the health index of the device to be evaluated is inaccurate exists.
Disclosure of Invention
The invention aims to provide a method and a device for acquiring a health index of equipment so as to improve the accuracy of the health index of the equipment to be evaluated.
The invention solves the technical problems through the following technical scheme:
the embodiment of the invention provides a method for acquiring a health index of equipment, which comprises the following steps:
acquiring each piece of historical record data of equipment to be evaluated and an equipment state corresponding to the record, wherein each piece of historical record data comprises at least one data item, each data item corresponds to data acquired by a sensor of one type, and the equipment state is a parameter for representing whether the equipment to be evaluated works normally or not;
training a preset naive Bayes classification model according to each record contained in the historical record data and the equipment state corresponding to the record;
dividing the historical record data into sub-historical record data according to the state corresponding to each piece of historical record data; for each piece of sub-history data, dividing data items contained in the history data into at least one data item set according to the sensor category corresponding to the data item contained in each record of the sub-history data; for each data item set, sequencing the data item sets by utilizing a quartile model, and acquiring an upper quartile corresponding to the data item set, a lower quartile corresponding to the data item set, and a first intermediate value acquired according to a difference value between the upper quartile and the lower quartile;
acquiring current recorded data of the equipment to be evaluated, and acquiring an equipment state corresponding to the current recorded data by using a trained naive Bayes classification model; acquiring an upper quartile, a lower quartile and a first intermediate value corresponding to each sensor in the current state according to the current state of the equipment and the type of the sensor corresponding to the current recorded data;
acquiring an upper quartile corresponding to each sensor or a distance from a lower quartile corresponding to each sensor to the current recorded data, and calculating the abnormality degree corresponding to each sensor according to the ratio of the distance to the first intermediate value;
acquiring a first health index corresponding to the sensor according to a preset corresponding function of the degree of abnormality corresponding to the sensor and the first health index of the sensor, wherein the corresponding function comprises: piecewise linear functions or polynomial curve functions;
and calculating a second health index of the equipment to be evaluated according to the weighted sum of the first health indexes corresponding to all the sensors.
Optionally, in a specific implementation manner of the embodiment of the present invention, before training a preset naive bayesian classification model according to each record included in the history data and a device state corresponding to the record, the method further includes:
and performing data cleaning operation on each piece of history record data, wherein the data cleaning operation comprises the following steps: and deleting the current historical record data when the current historical record data has a missing value.
Optionally, in a specific implementation manner of the embodiment of the present invention, before training a preset naive bayesian classification model according to each record included in the history data and a device state corresponding to the record, the method further includes:
and screening out data items with main relevance in the current piece of history data and data items without main relevance with other data items aiming at each piece of history data, wherein the data items with main relevance are data items of which the change can cause the change of other data items in the current piece of history data.
Optionally, in a specific implementation manner of the embodiment of the present invention, the obtaining a distance from the upper quartile or the lower quartile to the currently recorded data, and calculating the abnormality degree corresponding to the sensor according to a ratio of the distance to the first intermediate value includes:
by means of the formula (I) and (II),
calculating the degree of abnormality corresponding to the sensor, wherein,
Eithe abnormality degree corresponding to the sensor; siThe current record data is obtained; UQiAn upper quartile corresponding to the set of data items; iqriIs the first intermediate value; LQiAnd the lower quartile corresponding to the data item set.
Optionally, in a specific implementation manner of the embodiment of the present invention, the calculating a second health index of the device to be evaluated according to a weighted sum of the first health indexes corresponding to all the sensors includes:
by means of the formula (I) and (II),
calculating a second health index corresponding to the device to be evaluated, wherein,
sigma is a summation function; wiThe weight of the corresponding abnormality degree of each sensor;
a corresponding degree of abnormality for each sensor;
is a function of the operation of taking the product.
Optionally, in a specific implementation manner of the embodiment of the present invention, the method further includes:
under the current state, calculating a second intermediate value according to the product of a preset coefficient and the first intermediate value; calculating an upper bound corresponding to a sensor corresponding to the data item set according to the sum of the second intermediate value and the upper quartile; calculating a lower bound corresponding to the sensor according to the difference between the lower quartile and the second intermediate value;
for each sensor, judging whether the measurement value of the sensor exceeds a data range corresponding to an upper bound corresponding to the sensor and a lower bound corresponding to the sensor;
and if so, outputting the measured value of the sensor as an abnormal parameter.
The embodiment of the invention also provides a device for acquiring the health index of the equipment, which comprises: a first acquisition module, a training module, a second acquisition module, a third acquisition module, a first calculation module, a fourth acquisition module, and a second calculation module,
the first acquisition module is used for acquiring each piece of historical record data of the equipment to be evaluated and the equipment state corresponding to the record, wherein each piece of historical record data comprises at least one data item, each data item corresponds to data acquired by a sensor of one category, and the equipment state is a parameter for representing whether the equipment to be evaluated works normally or not;
the training module is used for training a preset naive Bayesian classification model according to each record contained in the historical record data and the equipment state corresponding to the record;
the second acquisition module is used for dividing the historical record data into sub-historical record data according to the state corresponding to each piece of historical record data; for each piece of sub-history data, dividing data items contained in the history data into at least one data item set according to the sensor category corresponding to the data item contained in each record of the sub-history data; for each data item set, sequencing the data item sets by utilizing a quartile model, and acquiring an upper quartile corresponding to the data item set, a lower quartile corresponding to the data item set, and a first intermediate value acquired according to a difference value between the upper quartile and the lower quartile;
the third obtaining module is used for obtaining the current recording data of the equipment to be evaluated and obtaining the equipment state corresponding to the current recording data by using a trained naive Bayesian classification model; acquiring an upper quartile, a lower quartile and a first intermediate value corresponding to each sensor in the current state according to the current state of the equipment and the type of the sensor corresponding to the current recorded data;
the first calculation module is used for acquiring the upper quartile corresponding to the sensor or the distance from the lower quartile corresponding to the sensor to the current recorded data aiming at each sensor, and calculating the abnormality degree corresponding to the sensor according to the ratio of the distance to the first intermediate value;
the fourth obtaining module is configured to obtain a first health index corresponding to the sensor according to a preset corresponding function between the degree of abnormality corresponding to the sensor and the first health index of the sensor, where the corresponding function includes: piecewise linear functions or polynomial curve functions;
and the second calculation module is used for calculating a second health index of the equipment to be evaluated according to the weighted sum of the first health indexes corresponding to all the sensors.
Optionally, in a specific implementation manner of the embodiment of the present invention, the apparatus further includes: a cleaning module, configured to perform a data cleaning operation on each piece of history data, where the data cleaning operation includes: and deleting the current historical record data when the current historical record data has a missing value.
Optionally, in a specific implementation manner of the embodiment of the present invention, the apparatus further includes: and the screening module is used for screening out data items with main relevance in the current historical record data and data items without main relevance with other data items aiming at each piece of historical record data, wherein the data items with main relevance are data items, and the change of the data items can cause the change of other data items in the current historical record data.
Optionally, in a specific implementation manner of the embodiment of the present invention, the first calculating module is further configured to:
by means of the formula (I) and (II),
calculating the degree of abnormality corresponding to the sensor, wherein,
Eithe abnormality degree corresponding to the sensor; siThe current record data is obtained; UQiFor the data item set pairUpper quartile of response; iqriIs the first intermediate value; LQiAnd the lower quartile corresponding to the data item set.
Compared with the prior art, the invention has the following advantages:
by applying the embodiment of the invention, the Bayesian classification model is trained through the historical record data of the equipment to be evaluated, then the trained Bayesian classification model is utilized to identify the working state corresponding to the current record data of the equipment, and then the health index of the equipment to be evaluated is calculated according to the current working state of the equipment to be evaluated, so that the accuracy of the health index of the equipment to be evaluated is improved.
Drawings
Fig. 1 is a schematic flow chart of a method for obtaining a health index of a device according to an embodiment of the present invention;
FIG. 2 is a schematic diagram illustrating a degree of abnormality associated with the sensor as a function of a first health index of the sensor, in accordance with an embodiment of the present invention;
fig. 3 is a schematic structural diagram of an apparatus for acquiring a health index of a device according to an embodiment of the present invention.
Detailed Description
The following examples are given for the detailed implementation and specific operation of the present invention, but the scope of the present invention is not limited to the following examples.
In order to solve technical problems in the prior art, embodiments of the present invention provide a method and an apparatus for obtaining a health index of a device, and first, a method for obtaining a health index of a device provided in an embodiment of the present invention is described below.
Fig. 1 is a schematic flow chart of a method for obtaining a health index of a device according to an embodiment of the present invention, as shown in fig. 1, the method includes:
s101: acquiring each piece of historical record data of equipment to be evaluated and equipment state corresponding to the record, wherein each piece of historical record data comprises at least one data item, each data item corresponds to data acquired by one type of sensor, and the equipment state is a parameter for representing whether the equipment to be evaluated works normally or not.
For example, the obtained historical data of the device to be evaluated may be as shown in table 1.
Table 1 is an acquired history data table of the device to be evaluated.
TABLE 1
| Time of day | Sensor 1 | Sensor 2 | Sensor 3 | Sensor 4 | Sensor 5 | Device status |
| 160053 | 32 | 66.6 | 0.18518 | 0.45 | 60.3 | 3 |
| 162346 | 46 | 64.1 | 0.22281 | 0.43 | 57.1 | 4 |
| 164418 | 18 | 64.6 | 0.29025 | 0.49 | 67.5 | 1 |
| 165518 | 24 | 64.6 | 0.49 | 67.5 | 1 | |
| 166618 | 17 | 0.29025 | 0.49 | 67.5 | 1 |
As shown in table 1, the measurement data of each sensor in the same row corresponding to 160053 is a piece of history data, and the state "3" is the device state corresponding to the piece of history data.
It is understood that, in the piece of history data with the time of 160053, the measurement data "32" corresponding to the sensor 1 may be the data item-1, and the measurement data "66.6" corresponding to the sensor 2 may be the data item-2. Each data item originates from a sensor.
It should be noted that, in the same history data, there may be a plurality of data items corresponding to the same kind of sensors, and of course, these same kind of sensors may be sensors installed on different parts of the device to be evaluated.
In practical applications, the sensor may be a temperature sensor, a vibration sensor, a rotation speed sensor, a voltage sensor, a current sensor, a power sensor, and the like.
S102: and training a preset naive Bayes classification model according to each record contained in the historical record data and the equipment state corresponding to the record.
Illustratively, each piece of history record data obtained in step S101 and shown in table 1, and the device state corresponding to the record are input into a preset naive bayesian classification model, and the trained naive bayesian classification model is trained to obtain a trained naive bayesian classification model.
It is understood that the training method of the naive bayes classification model is prior art and will not be described herein.
S103: dividing the historical record data into sub-historical record data according to the state corresponding to each piece of historical record data; for each piece of sub-history data, dividing data items contained in the history data into at least one data item set according to the sensor category corresponding to the data item contained in each record of the sub-history data; and for each data item set, sequencing the data item sets by utilizing a quartile model, and acquiring an upper quartile corresponding to the data item set, a lower quartile corresponding to the data item set and a first intermediate value acquired according to a difference value between the upper quartile and the lower quartile.
For example, taking the data in table 1 as an example, the set of history data with the device status of "3" may be used as the sub-history data-1; the set of history data having a device status of "4" may be regarded as one sub-history data-2.
Then, for each sub-history data, taking sub-history data-1 as an example, the data item corresponding to sensor 1 may be: "32", "46", "18", "24" as the data item set-1, can be the data item corresponding to sensor 2: "66.6", "64.1", "64.6" as data item set-2.
Then, for each data item set, taking the data item set-1 as an example, the data items "32", "46", "18", and "24" are sorted from large to small, and the sorted sequence-1, "46", "32", "24", and "18" is obtained. The upper quartile 46 × 0.75+32 × 0.25 ═ 42.5 of the sequence-1 is obtained by using the quartile model, the lower quartile 24 × 0.25+18 × 0.75 ═ 19.5 of the sequence-1 is obtained by using the quartile model, and then the first intermediate value 42.5-19.5 ═ 23 corresponding to the data item set-1 is calculated.
Further, an upper quartile, a lower quartile, and a first intermediate value corresponding to each device state may be obtained.
S104: acquiring current recorded data of the equipment to be evaluated, and acquiring an equipment state corresponding to the current recorded data by using a trained naive Bayes classification model; and acquiring an upper quartile, a lower quartile and a first intermediate value corresponding to each sensor in the current state according to the current state of the equipment and the type of the sensor corresponding to the current recorded data.
Exemplarily, current record data of the device to be evaluated is obtained, each record of the current record data is classified by using the trained naive bayesian classification model in the step S102, and a device state corresponding to the current record data is obtained.
If the device status corresponding to the currently recorded data-1 of one of the pieces is "3". The upper quartile, the lower quartile, and the first intermediate value corresponding to the device state being "3" are acquired from the step S103.
It is to be understood that the form of the currently recorded data may be as shown in table 1, and currently, the value of the data item in the currently recorded data may be different from the value of the data item shown in table 1.
S105: and acquiring the upper quartile corresponding to the sensor or the distance from the lower quartile corresponding to the sensor to the current recorded data aiming at each sensor, and calculating the abnormality degree corresponding to the sensor according to the ratio of the distance to the first intermediate value.
Specifically, the obtaining a distance from the upper quartile or the lower quartile to the current recording data, and calculating the abnormality degree corresponding to the sensor according to a ratio of the distance to the first intermediate value includes:
by means of the formula (I) and (II),
calculating the degree of abnormality corresponding to the sensor, wherein,
Eithe abnormality degree corresponding to the sensor; siThe current record data is obtained; UQiAn upper quartile corresponding to the set of data items; iqriIs the first intermediate value; LQiAnd the lower quartile corresponding to the data item set.
For example, the degree of abnormality corresponding to each data item corresponding to the current recording data can be calculated by substituting the upper quartile, the lower quartile and the first intermediate value corresponding to the current recording data acquired in step S105 into the above formula.
S106: acquiring a first health index corresponding to the sensor according to a preset corresponding function of the degree of abnormality corresponding to the sensor and the first health index of the sensor, wherein the corresponding function comprises: piecewise linear functions or polynomial curve functions.
For example, fig. 2 is a schematic diagram of a corresponding function of the degree of abnormality corresponding to the sensor and the first health index of the sensor according to an embodiment of the present invention, and as shown in fig. 2, the preset corresponding function of the degree of abnormality corresponding to the sensor and the first health index of the sensor may be,
Hia first health index for the sensor; eiAnd the corresponding abnormality degree of the sensor.
In practical application, a piecewise linear function or a polynomial curve function can be adopted to calculate the first health index of the sensor according to the abnormality degree corresponding to the sensor.
Similarly, the first health index of the sensor corresponding to each data item of the currently recorded data is also calculated according to the above method, and is not described herein again.
FIG. 2 is a schematic diagram illustrating a degree of abnormality associated with the sensor as a function of a first health index of the sensor, in accordance with an embodiment of the present invention;
s107: and calculating a second health index of the equipment to be evaluated according to the weighted sum of the first health indexes corresponding to all the sensors.
In particular, a formula may be used,
calculating a second health index corresponding to the device to be evaluated, wherein,
sigma is a summation function; wiThe weight of the corresponding abnormality degree of each sensor;
a corresponding degree of abnormality for each sensor;
is a function of the operation of taking the product.
Illustratively, each data obtained by the calculation in step S106Substituting the first health index of the sensor corresponding to the item into a formula,
a second health index for the device under evaluation is calculated.
In practical application, the weight corresponding to each first health index may be preset, for example, a larger weight may be set for a higher failure occurrence rate, a higher failure loss caused after a failure, or a higher maintenance cost after a failure, or a smaller weight may be set for the higher weight.
By applying the embodiment of the invention shown in the figure 1, the Bayesian classification model is trained through the historical record data of the equipment to be evaluated, then the trained Bayesian classification model is utilized to identify the working state corresponding to the current record data of the equipment, and then the health index of the equipment to be evaluated is calculated according to the current working state of the equipment to be evaluated, so that the accuracy of the health index of the equipment to be evaluated is improved.
In a specific implementation manner of the embodiment of the present invention, on the basis of the embodiment shown in fig. 1 of the present invention, before the step S102, the method further includes:
s108 (not shown): and performing data cleaning operation on each piece of history record data, wherein the data cleaning operation comprises the following steps: and deleting the current historical record data when the current historical record data has a missing value.
Illustratively, as shown in table 1, the history data of time "165518" lacks the measurement result corresponding to sensor 3, the history data of time "166618" lacks the measurement result corresponding to sensor 2, and the history data of time "165518" and time "166618" is deleted.
In practical application, the data cleaning operation may further be to delete data in the history data, the time length of which from the current time exceeds a set time length, or delete data repeatedly reported by the sensor at the same time.
By applying the embodiment of the invention, the influence of incomplete data on the calculation of the health index of the equipment can be avoided.
In a specific implementation manner of the embodiment of the present invention, on the basis of the embodiment shown in fig. 1 of the present invention, before the step S102, the method further includes:
s109 (not shown in the figure): and screening out data items with main relevance in the current piece of history data and data items without main relevance with other data items aiming at each piece of history data, wherein the data items with main relevance are data items of which the change can cause the change of other data items in the current piece of history data.
Illustratively, taking voltage data and current data as an example, if the sensor 1 measures the input voltage of the heating wire of the device, the sensor 2 measures the input current of the heating wire. In practical applications, for a heating wire, when a certain voltage is input, the input current is determined, and therefore, the voltage is a data item of main correlation. The current is a voltage, and the data item corresponding to the sensor 2 is deleted.
By applying the embodiment of the invention, the number of data items can be reduced, the calculation amount is further reduced, and the calculation efficiency is improved.
In a specific implementation manner of the embodiment of the present invention, on the basis of the embodiment shown in fig. 1 of the present invention, the method further includes:
s1010 (not shown): under the current state, calculating a second intermediate value according to the product of a preset coefficient and the first intermediate value; calculating an upper bound corresponding to a sensor corresponding to the data item set according to the sum of the second intermediate value and the upper quartile; and calculating a lower bound corresponding to the sensor according to the difference between the lower quartile and the second intermediate value.
Illustratively, taking sensor 1 as an example, the quartile is recorded as UQ1Lower quartile is LQ1Difference value IQR1＝UQ1-LQ1。
The corresponding upper limit of the sensor 1 can be setThe method comprises the following steps: UQ1+k*IQR1(ii) a The corresponding lower bound of sensor 1 may be set to LQ1-k*IQR1Where k is a predetermined coefficient, which can be set empirically.
Other sensors for the device to be evaluated
S1011 (not shown): for each sensor, judging whether the measurement value of the sensor exceeds a data range corresponding to an upper bound corresponding to the sensor and a lower bound corresponding to the sensor; if yes, go to step S1012.
For each sensor, judging whether the measured value of the sensor exceeds a data range corresponding to the upper limit and the lower limit corresponding to the sensor; can be as follows:
taking sensor 1 as an example, it is determined whether the measurement data of sensor 1 is less than LQ1-k*IQR1(ii) a Or whether it is greater than UQ1+k*IQR1If yes, go to step S1012.
S1012 (not shown): and outputting the measured value of the sensor as an abnormal parameter.
For example, the measured value measured by the sensor whose determination result is that it is the corresponding sensor is output as the abnormal parameter.
In practical applications, the abnormal parameter may be displayed on a display device, such as a display or an indicator light.
Corresponding to the embodiment shown in fig. 1 of the present invention, the embodiment of the present invention further provides a device for obtaining a health index of a device.
Fig. 3 is a schematic structural diagram of an apparatus for acquiring a health index of a device according to an embodiment of the present invention, as shown in fig. 3, the apparatus includes: a first acquisition module 301, a training module 302, a second acquisition module 303, a third acquisition module 304, a first calculation module 305, a fourth acquisition module 306, and a second calculation module 307, wherein,
the first obtaining module 301 is configured to obtain each piece of history data of a device to be evaluated and a device state corresponding to the record, where each piece of history data includes at least one data item, each data item corresponds to data acquired by a type of sensor, and the device state is a parameter that represents whether the device to be evaluated normally operates;
the training module 302 is configured to train a preset naive bayesian classification model according to each record included in the historical record data and the device state corresponding to the record;
the second obtaining module 303 is configured to divide the history data into sub-history data according to a state corresponding to each piece of history data; for each piece of sub-history data, dividing data items contained in the history data into at least one data item set according to the sensor category corresponding to the data item contained in each record of the sub-history data; for each data item set, sequencing the data item sets by utilizing a quartile model, and acquiring an upper quartile corresponding to the data item set, a lower quartile corresponding to the data item set, and a first intermediate value acquired according to a difference value between the upper quartile and the lower quartile;
the third obtaining module 304 is configured to obtain current recorded data of the device to be evaluated, and obtain a device state corresponding to the current recorded data by using a trained naive bayes classification model; acquiring an upper quartile, a lower quartile and a first intermediate value corresponding to each sensor in the current state according to the current state of the equipment and the type of the sensor corresponding to the current recorded data;
the first calculating module 305 is configured to obtain, for each sensor, an upper quartile corresponding to the sensor or a distance from a lower quartile corresponding to the sensor to the current recording data, and calculate an abnormality degree corresponding to the sensor according to a ratio of the distance to the first intermediate value;
the fourth obtaining module 306 is configured to obtain the first health index corresponding to the sensor according to a preset corresponding function between the degree of abnormality corresponding to the sensor and the first health index of the sensor, where the corresponding function includes: piecewise linear functions or polynomial curve functions;
the second calculating module 307 is configured to calculate a second health index of the device to be evaluated according to a weighted sum of the first health indexes corresponding to all the sensors.
By applying the embodiment of the invention shown in fig. 3, the Bayesian classification model is trained through the historical record data of the equipment to be evaluated, then the trained Bayesian classification model is utilized to identify the working state corresponding to the current record data of the equipment, and then the health index of the equipment to be evaluated is calculated according to the current working state of the equipment to be evaluated, so that the accuracy of the health index of the equipment to be evaluated is improved.
In a specific implementation manner of the embodiment of the present invention, on the basis of the embodiment shown in fig. 3 of the present invention, the apparatus further includes: a cleansing module 308 (not shown in the figures) configured to perform a data cleansing operation on each piece of history data, where the data cleansing operation includes: and deleting the current historical record data when the current historical record data has a missing value.
In a specific implementation manner of the embodiment of the present invention, on the basis of the embodiment shown in fig. 3 of the present invention, the apparatus further includes: a screening module 309 (not shown in the figure) for screening out, for each piece of history data, a data item having a main relevance in the current piece of history data and a data item having no main relevance to other data items, where the data item having a main relevance is a data item whose change may cause a change in other data items in the current piece of history data.
In a specific implementation manner of the embodiment of the present invention, the first calculating module 305 is further configured to:
by means of the formula (I) and (II),
calculating the degree of abnormality corresponding to the sensor, wherein,
Eithe abnormality degree corresponding to the sensor; siThe current record data is obtained; UQiAn upper quartile corresponding to the set of data items; iqriIs the first intermediate value; LQiAnd the lower quartile corresponding to the data item set.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents and improvements made within the spirit and principle of the present invention are intended to be included within the scope of the present invention.
Claims (10)
1. A method for obtaining a health index of a device, the method comprising:
acquiring each piece of historical record data of equipment to be evaluated and an equipment state corresponding to the record, wherein each piece of historical record data comprises at least one data item, each data item corresponds to data acquired by a sensor of one type, and the equipment state is a parameter for representing whether the equipment to be evaluated works normally or not;
training a preset naive Bayes classification model according to each record contained in the historical record data and the equipment state corresponding to the record;
dividing the historical record data into sub-historical record data according to the state corresponding to each piece of historical record data; for each piece of sub-history data, dividing data items contained in the history data into at least one data item set according to the sensor category corresponding to the data item contained in each record of the sub-history data; for each data item set, sequencing the data item sets by utilizing a quartile model, and acquiring an upper quartile corresponding to the data item set, a lower quartile corresponding to the data item set, and a first intermediate value acquired according to a difference value between the upper quartile and the lower quartile;
acquiring current recorded data of the equipment to be evaluated, and acquiring an equipment state corresponding to the current recorded data by using a trained naive Bayes classification model; acquiring an upper quartile, a lower quartile and a first intermediate value corresponding to each sensor in the current state according to the current state of the equipment and the type of the sensor corresponding to the current recorded data;
acquiring an upper quartile corresponding to each sensor or a distance from a lower quartile corresponding to each sensor to the current recorded data, and calculating the abnormality degree corresponding to each sensor according to the ratio of the distance to the first intermediate value;
acquiring a first health index corresponding to the sensor according to a preset corresponding function of the degree of abnormality corresponding to the sensor and the first health index of the sensor, wherein the corresponding function comprises: piecewise linear functions or polynomial curve functions;
and calculating a second health index of the equipment to be evaluated according to the weighted sum of the first health indexes corresponding to all the sensors.
2. The method for obtaining the health index of the equipment according to claim 1, wherein before training a preset naive bayes classification model according to each record contained in the history data and the equipment state corresponding to the record, the method further comprises:
and performing data cleaning operation on each piece of history record data, wherein the data cleaning operation comprises the following steps: and deleting the current historical record data when the current historical record data has a missing value.
3. The method for obtaining the health index of the equipment according to claim 1, wherein before training a preset naive bayes classification model according to each record contained in the history data and the equipment state corresponding to the record, the method further comprises:
and screening out data items with main relevance in the current historical record data and data items without main relevance with other data items aiming at each piece of historical record data, wherein the data items with main relevance are data items of which the change causes the change of other data items in the current historical record data.
4. The method for obtaining the health index of the equipment according to claim 1, wherein the step of obtaining the distance from the upper quartile or the lower quartile to the currently recorded data and calculating the degree of abnormality corresponding to the sensor according to the ratio of the distance to the first intermediate value comprises:
by means of the formula (I) and (II),
calculating the degree of abnormality corresponding to the sensor, wherein,
Eithe abnormality degree corresponding to the sensor; siThe current record data is obtained; UQiAn upper quartile corresponding to the set of data items; iqriIs the first intermediate value; LQiAnd the lower quartile corresponding to the data item set.
5. The method for obtaining the health index of the equipment according to claim 1, wherein the calculating the second health index of the equipment to be evaluated according to the weighted sum of the first health indexes corresponding to all the sensors comprises:
by means of the formula (I) and (II),
calculating a second health index corresponding to the device to be evaluated, wherein,
sigma is a summation function; wiThe weight of the corresponding abnormality degree of each sensor;
a corresponding degree of abnormality for each sensor;
is a function of the operation of taking the product.
6. The method for obtaining the health index of the equipment according to claim 1, wherein the method further comprises:
under the current state, calculating a second intermediate value according to the product of a preset coefficient and the first intermediate value; calculating an upper bound corresponding to a sensor corresponding to the data item set according to the sum of the second intermediate value and the upper quartile; calculating a lower bound corresponding to the sensor according to the difference between the lower quartile and the second intermediate value;
for each sensor, judging whether the measurement value of the sensor exceeds a data range corresponding to an upper bound corresponding to the sensor and a lower bound corresponding to the sensor;
and if so, outputting the measured value of the sensor as an abnormal parameter.
7. An apparatus for obtaining a health index of a device, the apparatus comprising: a first acquisition module, a training module, a second acquisition module, a third acquisition module, a first calculation module, a fourth acquisition module, and a second calculation module,
the first acquisition module is used for acquiring each piece of historical record data of the equipment to be evaluated and the equipment state corresponding to the record, wherein each piece of historical record data comprises at least one data item, each data item corresponds to data acquired by a sensor of one category, and the equipment state is a parameter for representing whether the equipment to be evaluated works normally or not;
the training module is used for training a preset naive Bayesian classification model according to each record contained in the historical record data and the equipment state corresponding to the record;
the second acquisition module is used for dividing the historical record data into sub-historical record data according to the state corresponding to each piece of historical record data; for each piece of sub-history data, dividing data items contained in the history data into at least one data item set according to the sensor category corresponding to the data item contained in each record of the sub-history data; for each data item set, sequencing the data item sets by utilizing a quartile model, and acquiring an upper quartile corresponding to the data item set, a lower quartile corresponding to the data item set, and a first intermediate value acquired according to a difference value between the upper quartile and the lower quartile;
the third obtaining module is used for obtaining the current recording data of the equipment to be evaluated and obtaining the equipment state corresponding to the current recording data by using a trained naive Bayesian classification model; acquiring an upper quartile, a lower quartile and a first intermediate value corresponding to each sensor in the current state according to the current state of the equipment and the type of the sensor corresponding to the current recorded data;
the first calculation module is used for acquiring the upper quartile corresponding to the sensor or the distance from the lower quartile corresponding to the sensor to the current recorded data aiming at each sensor, and calculating the abnormality degree corresponding to the sensor according to the ratio of the distance to the first intermediate value;
the fourth obtaining module is configured to obtain a first health index corresponding to the sensor according to a preset corresponding function between the degree of abnormality corresponding to the sensor and the first health index of the sensor, where the corresponding function includes: piecewise linear functions or polynomial curve functions;
and the second calculation module is used for calculating a second health index of the equipment to be evaluated according to the weighted sum of the first health indexes corresponding to all the sensors.
8. The apparatus for obtaining health index of a device according to claim 7, further comprising: a cleaning module, configured to perform a data cleaning operation on each piece of history data, where the data cleaning operation includes: and deleting the current historical record data when the current historical record data has a missing value.
9. The apparatus for obtaining health index of a device according to claim 7, further comprising: and the screening module is used for screening out data items with main relevance in the current historical record data and data items without main relevance with other data items aiming at each piece of historical record data, wherein the data items with main relevance are data items of which the change causes the change of other data items in the current historical record data.
10. The apparatus for obtaining the health index of a device according to claim 7, wherein the first calculating module is further configured to:
by means of the formula (I) and (II),
calculating the degree of abnormality corresponding to the sensor, wherein,
Eithe abnormality degree corresponding to the sensor; siThe current record data is obtained; UQiAn upper quartile corresponding to the set of data items; iqriIs the first intermediate value; LQiAnd the lower quartile corresponding to the data item set.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810511519.4A CN108764498B (en) | 2018-05-25 | 2018-05-25 | Method and device for acquiring health index of equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810511519.4A CN108764498B (en) | 2018-05-25 | 2018-05-25 | Method and device for acquiring health index of equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108764498A CN108764498A (en) | 2018-11-06 |
| CN108764498B true CN108764498B (en) | 2020-12-29 |
Family
ID=64005948
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810511519.4A Active CN108764498B (en) | 2018-05-25 | 2018-05-25 | Method and device for acquiring health index of equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108764498B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111949941B (en) * | 2020-07-03 | 2023-03-03 | 广州明珞汽车装备有限公司 | Equipment fault detection method, system, device and storage medium |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7945427B2 (en) * | 2008-04-18 | 2011-05-17 | The Boeing Company | Methods and systems for providing unanticipated demand predictions for maintenance |
| CN103454516B (en) * | 2013-05-06 | 2015-10-28 | 国家电网公司 | Intelligent transformer substation secondary equipment health state diagnostic method |
| CN103631681B (en) * | 2013-12-10 | 2016-04-20 | 国家电网公司 | A kind of method of online reparation abnormal data of wind power plant |
-
2018
- 2018-05-25 CN CN201810511519.4A patent/CN108764498B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN108764498A (en) | 2018-11-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111461555B (en) | Production line quality monitoring method, device and system | |
| US11156997B2 (en) | Systems and methods for artificial intelligence-based maintenance of an air conditioning system | |
| EP1982301B1 (en) | Method of condition monitoring | |
| Ghasemi et al. | Optimal condition based maintenance with imperfect information and the proportional hazards model | |
| CN104573850A (en) | Method for evaluating state of thermal power plant equipment | |
| US11003178B2 (en) | Facility diagnosis device, facility diagnosis method, and facility diagnosis program | |
| DE102017003165A1 (en) | A machine learning device and a machine learning method for learning the error prediction of a main shaft or a motor that drives the main shaft, and an error prediction device and an error prediction system comprising a machine learning device | |
| KR20200017506A (en) | Device for diagnosing abnormalities in processes and how to diagnose abnormalities | |
| CN111664083A (en) | Nuclear power main pump fault diagnosis method based on Bayesian network | |
| CN109213034B (en) | Equipment health degree monitoring method and device, computer equipment and readable storage medium | |
| CN108921305B (en) | Component life period monitoring method | |
| Qin et al. | Remaining useful life prediction for rotating machinery based on optimal degradation indicator | |
| JP2009086896A (en) | Failure prediction system and failure prediction method for computer | |
| CN108764498B (en) | Method and device for acquiring health index of equipment | |
| JP5565357B2 (en) | Equipment diagnostic device, equipment diagnostic method, equipment diagnostic program, and computer-readable recording medium recording the same | |
| Amora | On the validity assessment of formative measurement models in PLS-SEM | |
| CN115829321B (en) | Intelligent processing risk disposal system for food enterprises | |
| CN108182306B (en) | Method for determining degradation failure threshold of abrasive particle characteristic parameters of vehicle power transmission device | |
| Salgueiro et al. | On-line oil monitoring and diagnosis | |
| Assaf | Prognostics and health management for multi-component systems | |
| Tochev et al. | System condition monitoring through Bayesian change point detection using pump vibrations | |
| JP7281394B2 (en) | Abnormal diagnosis device and program | |
| US11965859B1 (en) | System and method for empirical estimation of life remaining in industrial equipment | |
| CN116627093B (en) | Nitrile glove processing control method, system, equipment and storage medium | |
| Zhou et al. | Early Warning Strategy of Sparse Failures for Highly Reliable Products Based on the Bayesian Method |
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 | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20200817 Address after: Room 203, building 3, Zone C, intelligent technology park, 3963 Susong Road, Hefei Economic and Technological Development Zone, 230000 Anhui Province Applicant after: Anhui Lingyun IOT Technology Co.,Ltd. Address before: 236000 Anhui city of Hefei Province Economic and Technological Development Zone Cuiwei Road No. 6 Haiheng building room 6012 Applicant before: HRG INTERNATIONAL INSTITUTE FOR RESEARCH & INNOVATION |
|
| TA01 | Transfer of patent application right | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |