CN110532698A - A kind of industrial equipment vibration performance value trend forecasting method based on data model - Google Patents
A kind of industrial equipment vibration performance value trend forecasting method based on data model Download PDFInfo
- Publication number
- CN110532698A CN110532698A CN201910812188.2A CN201910812188A CN110532698A CN 110532698 A CN110532698 A CN 110532698A CN 201910812188 A CN201910812188 A CN 201910812188A CN 110532698 A CN110532698 A CN 110532698A
- Authority
- CN
- China
- Prior art keywords
- model
- data
- industrial equipment
- sequence
- error
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01H—MEASUREMENT OF MECHANICAL VIBRATIONS OR ULTRASONIC, SONIC OR INFRASONIC WAVES
- G01H17/00—Measuring mechanical vibrations or ultrasonic, sonic or infrasonic waves, not provided for in the preceding groups
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
A kind of industrial equipment vibration performance value trend forecasting method based on data model, comprising the following steps: step 1, by vibrating sensor, acquire the characteristic value of the several data type signal of industrial equipment；Step 2, the stationarity of training data is observed；Step 3, autoregressive moving-average model is constituted；Step 4, the number of parameter is determined according to bayesian information criterion function method；Step 5, the randomness of the prediction error sequence of best model is examined；Step 6, model is verified；Step 7, model prediction.The present invention is by the characteristic value of the vibrating sensor acquisition several data type signal of industrial equipment as parameter, and the acquiring way of this parameter is simple, and high-efficient, there is no the states of over-fitting or poor fitting.
Description
Technical field
The invention belongs to industrial equipment states to monitor field, in particular to a kind of industrial equipment vibration based on data model
Characteristic value trend forecasting method.
Background technique
Industrial equipment maintenance is the core content in factory's daily management mission, and the healthy and safe operation of equipment can not only mention
Product yield, optimization productivity effect, and the safety in production that consumption can be reduced, ensure enterprise are risen, factory is helped to be maximized
Economic well-being of workers and staff.
Industrial equipment maintenance mainly have in the mode of factory at present: correction maintenance, preventive maintenance (i.e. periodic maintenance) and in advance
The property surveyed three kinds of modes of maintenance.These three different maintenance modes are also exactly the three phases of equipment management development.For a long time, I
What state's industrial equipment was generally carried out is periodic inspection, the maintenance mechanism of correction maintenance.And facts proved that, up to 60% regular inspection
It repairs and is not necessarily to.Predictive maintenance is the mainstream maintenance mode of industry in the future, i.e., in the feelings for not influencing equipment normal operation
Under condition, the machine performance before failure occurs is monitored by Condition Monitoring Technology and fault diagnosis etc., and predict time of failure
And development trend, specific aim maintenance plan is formulated, most failures is handled in budding state, failure is avoided to deteriorate,
To avoid equipment hang-up, maintenance cost is substantially reduced.Difference of the method for predictive maintenance according to status monitoring means
And it is various to classify, there are commonly vibration monitoring method, pressure monitoring method, oil liquid monitoring method, acoustic emission detection methods etc..Predictive maintenance
System mainly include three early warning, diagnosis and prediction aspects, early warning is to judge whether equipment has by the determination of the amplitude of feature
Abnormal, diagnosis is the reason of determining failure by Time-Frequency Analysis method, and prediction is by predicting becoming in following a period of time
Gesture situation judges the time that equipment needs repairing.Wherein early warning and diagnosis research and application is more, and the research predicted and answers
It is less.
Under normal conditions, the industrial equipment vibration performance Value Data sequence in the following certain time period is obtained, there are two types of sides
Method: first method is using the method for moving average: by passing move and acquire average as predicted value one by the phase to time series
Kind prediction technique, has the simple method of moving average and the method for weighted moving average, is used equally for industrial equipment vibration performance value trend pre-
It surveys；Second method is to predict the industrial equipment vibration performance Value Data of the following certain time period using BP neural network algorithm
Sequence, and then realize alarm etc..However, the above two industrial equipment vibration performance Value Data obtained in the following certain time period
The method of sequence has its difficult point or deficiency.For first method, the method for moving average only uses the data of nearest k phase, every
When secondary moving average calculation, mobile interval is all k, and more stable time series is suitble to be predicted, is believed non-stationary
Number precision of prediction it is bad.For second method, BP neural network algorithm network structure is selected it is different, generally can only be by
Experience is selected.Network structure selection is excessive, inefficient in training, in fact it could happen that over-fitting causes network performance low, holds
Mistake decline；If selection is too small, will cause network again may not restrain, and the structure of network directly affect network approach energy
Power and popularization property, therefore, network structure is difficult to selection restriction BP neural network algorithm in the application of trend prediction.
Summary of the invention
The purpose of the present invention is to provide a kind of industrial equipment vibration performance value trend forecasting method based on data model,
To solve the above problems.
To achieve the above object, the invention adopts the following technical scheme:
A kind of industrial equipment vibration performance value trend forecasting method based on data model, comprising the following steps:
Step 1, by vibrating sensor, the characteristic value of the several data type signal of industrial equipment is acquired, selection is passed through
The historical data and current data of vibration performance value, historical data is as training data, and current data is as test data；
Step 2, the stationarity of training data is observed, if unstable, carries out calculus of differences, obtains second order difference training number
According to；
Step 3, model selects: establishing second order difference training data in autoregression model and moving average model basis
On, autoregression model and moving average model are combined, autoregressive moving-average model is constituted；
Step 4, model training: model parameter is determined firstly the need of parameter Estimation, then according to bayesian information criterion
Function method determines the number of parameter；
Step 5, model testing: the randomness of the prediction error sequence of best model is examined, that is, is examined between prediction error
It is whether independent；
Step 6, model is verified, and the corresponding timestamp of test data is brought into best model and obtains prediction data sequence,
By the accuracy rate for calculating the error validity model of prediction data sequence and test data；
Step 7, model prediction predicts industrial equipment vibration performance in the following certain time period by the data model of foundation
The prediction data sequence of value.
Further, in step 1, the data of acquisition include the spy of acceleration, four seed type signal of speed, displacement and envelope
Value indicative；The historical data for choosing vibration performance value is denoted as { T0, T1..., TmAnd current data be denoted as { Tm+1, Tm+2..., Tn, shape
At one group of data sequence { T with time sequencing0, T1..., Tn}。
Further, in step 2, the calculation formula of calculus of differences are as follows:
T′t=Tt+1-Tt
Wherein T 'tFor t-th of element in the training data sequence after first difference, Tt+1For in training data sequence
The t+1 element, t={ 0,1 ..., m-1 }；First difference training data { T ' is obtained after first difference0, T '1..., T 'm-1}；Two
Secondary difference training data { T "0, T "1..., T "m-2}。
Further, in step 3, the correlation of this different times of expression time sequence is carried out using regression model:
Wherein, T "tIt is observed value of the time series in the t phase, p is the order of autoregression model, indicates T "tOnly with its it
Preceding p phase sequential value is related, etIt is error, indicates the enchancement factor that cannot use specification of a model；Due to second order difference training data
{T"0, T "1..., T "m-2It is original training data { T0, T1..., TmData sequence after second order difference, it is the number of a zero-mean
According to sequence, i.e.,Then:
Meanwhile second order difference training data { T "0, T "1..., T "m-2In T "tThe original of error is predicted according to average early period
It then establishes, predicted value till now can be obtained plus prediction error on predicted value early period, can be obtained by recurrence:
T"t=et-θ1et-1-θ2et-2-…-θqet-q
Wherein T "tIt is observed value of the time series in the t phase, q is the order of moving average model, etIt is that time series exists
The error of t phase；
To sum up, second order difference training data { T "0, T "1..., T "m-2Establish in autoregression model and moving average model base
On plinth, autoregression model and moving average model are combined, just constitute autoregressive moving-average model:
Wherein, p is known as Autoregressive,For autoregressive coefficient；Q is known as sliding average order,
{θ1, θ2..., θqIt is sliding average coefficient.
Further, in step 4, model parameter is determined: Autoregressive p and sliding average order q；Believed according to Bayes
Criterion function method is ceased to determine the number i.e. order of model of parameter, score evaluation criterion are as follows:
More thanIt is maximum likelihood function estimated value, score is smaller, and model is best；If
Determine p, q upper limit, successively creates autoregressive moving-average model set (M0,0, M0,1..., Mp,q)；Then by second order difference training number
According to { T "0, T "1..., T "m-2Successively bring into set in model, obtain bic score setFind the corresponding model M of set mid-score minimum valuep,qIt is complete as best model
At model training process.
Further, in step 5, randomness is examined by auto-relativity function method, that is, the error sequence that gives a forecast from phase
Functional arrangement is closed, if be a kind of random variation between autocorrelation sequence in autocorrelation function graph, there is no truncation, hangover
Situation then predicts between error that independently best model is upchecked；Hangover refers to autocorrelation sequence with index percent monotone decreasing or shake
Decaying is swung, and truncation refers to that autocorrelation sequence becomes very small from some time point.
Further, in step 6, by test data { Tm+1, Tm+2..., TnCorresponding timestamp is brought into best model
Mp,qObtain prediction data sequence { Pm+1, Pm+2..., Pn, by calculating prediction data sequence { Pm+1, Pm+2..., PnAnd test number
According to { Tm+1, Tm+2..., TnError | Tm+1-Pm+1|, | Tm+2-Pm+2| ..., | Tn-Pn| the accuracy rate of verifying model, it is average exhausted
It is as follows to percentage error:
As best model Mp,qMean absolute percentage error mape within 5% just be applied to actual production.
Further, in step 7, predict that industrial equipment vibrates special in the following certain time period by the data model of foundation
The prediction data sequence of value indicative, the formula of concrete foundation are as follows:
Wherein, p is known as Autoregressive,For autoregressive coefficient；Q is known as sliding average order,
{θ1, θ2..., θqIt is sliding average coefficient.Wherein etIt is independent error term, T "tFor steady, normal state and zero-mean it is following certain
The prediction data sequence of industrial equipment vibration performance value in one period.
Compared with prior art, the present invention has following technical effect:
Data model of the invention can be adapted for non-stationary number by carrying out calculus of differences training to training data
According to the prediction of sequence, and it is not limited solely to stable data sequence, the scope of application is more extensive.
The present invention acquires the characteristic value of the several data type signal of industrial equipment as parameter by vibrating sensor, this
The acquiring way of kind parameter is simple, and high-efficient, there is no the states of over-fitting or poor fitting.
Data model structure is simple in the present invention, be easy to calculate, there is no excessively complicated or simple network structures to cause
Over-fitting or poor fitting state.
Parameter of the invention obtains high-efficient, and parameter is few, can fast and accurately prediction data sequence.
Detailed description of the invention
Fig. 1 is implementation flow chart of the invention；
Fig. 2 is oil transfer pump measuring point installation diagram；
Fig. 3 is the data sequence of history and current oil transfer pump vibration velocity virtual value trend；
Fig. 4-Fig. 6 is the data sequence after calculus of differences twice compared with original data sequence；
Fig. 7 be the following certain time period based on data model in oil transfer pump vibration velocity virtual value trend prediction result with
The line chart of actual result.
Specific embodiment
Below in conjunction with attached drawing, the present invention is further described:
Fig. 1 is please referred to Fig. 7: (1) choosing certain industrial equipment, vibrating sensor is installed in equipment, acceleration can be acquired
The characteristic value of degree, four seed type signal of speed, displacement and envelope, by the historical data and current number of choosing vibration performance value
According to one group of data sequence { T with time sequencing of formation0, T1..., Tn}.Historical data is as training data { T0, T1...,
Tm, current data is as test data { Tm+1, Tm+2..., Tn}。
(2) training data { T is observed0, T1..., TmStationarity, if it is unstable, carry out first-order difference operation, then
Stationarity is observed to decide whether to continue difference, until the unstable tendency of elimination training data sequence.The meter of calculus of differences
Calculate formula are as follows:
T′t=Tt+1-Tt
Wherein T 'tFor t-th of element in the training data sequence after first difference, Tt+1For in training data sequence
The t+1 element, t={ 0,1 ..., m-1 }.First difference training data { T ' is obtained after first difference0, T '1..., T 'm-1}.For
The stationarity for guaranteeing training data sequence generally requires and carries out difference twice to training data sequence.Finally obtain second difference
Divide training data { T "0, T "1..., T "m-2}。
(3) model selects, second order difference training data { T "0, T "1..., T "m-2In T "tVariation by Self-variation shadow
(if being influenced by dry spell observed value before it) is rung, therefore this difference that regression model carrys out expression time sequence can be used
The correlation in period:
Wherein, T "tIt is observed value of the time series in the t phase, p is the order of autoregression model, indicates T "tOnly with its it
Preceding p phase sequential value is related, etIt is error, indicates the enchancement factor that cannot use specification of a model.Due to second order difference training data
{T"0, T "1..., T "m-2It is original training data { T0, T1..., TmData sequence after second order difference, it is the number of a zero-mean
According to sequence, i.e.,Then:
Meanwhile second order difference training data { T "0, T "1..., T "m-2In T "tIt can be according to prediction average early period error
Principle establish, on predicted value early period plus prediction error can obtain predicted value till now, can be obtained by recurrence:
T"t=et-θ1et-1-θ2et-2-…-θqet-q
Wherein T "tIt is observed value of the time series in the t phase, q is the order of moving average model, etIt is that time series exists
The error of t phase.
To sum up, second order difference training data { T "0, T "1..., T "m-2Establish in autoregression model and moving average model base
On plinth, autoregression model and moving average model are combined, just constitute autoregressive moving-average model:
Wherein, p is known as Autoregressive,For autoregressive coefficient；Q is known as sliding average order,
{θ1, θ2..., θqIt is sliding average coefficient.Model had both considered interdependence of the index in time series during prediction,
It is contemplated that the interference of random fluctuation.
(4) model training, it is necessary first to which parameter Estimation determines model parameter: Autoregressive p and sliding average order
q.According to Oscar nurse razor criterion, the number of Model Parameter less, prediction error (generally indicate to predict with residual sum of squares (RSS)
Error) smaller model is better.The number i.e. order of model of parameter, score are determined according to bayesian information criterion function method
Evaluation criterion are as follows:
More thanIt is maximum likelihood function estimated value, score is smaller, and model is best.If
Determine p, q upper limit (generally no greater than 20), successively creates autoregressive moving-average model set (M0,0, M0,1..., Mp,q).Then will
Second order difference training data { T "0, T "1..., T "m-2Successively bring into set in model, obtain bic score setFind the corresponding model M of set mid-score minimum valuep,qIt is complete as best model
At model training process.
(5) best model M is examined in model testingp,qPrediction error sequence { et, et-1..., et-qRandomness, that is, examine
It tests whether independent between prediction error.Randomness can be examined by auto-relativity function method, that is, give a forecast error sequence
Autocorrelation function graph, if be a kind of random variation between autocorrelation sequence in autocorrelation function graph, there is no truncation,
Hangover situation (hangover refer to autocorrelation sequence with index percent monotone decreasing or concussion decaying, and truncation refer to autocorrelation sequence from some
Time point becomes very small), then predict independent between error, best model Mp,qIt upchecks.
(6) model is verified, by test data { Tm+1, Tm+2..., TnCorresponding timestamp is brought into best model Mp,q
To prediction data sequence { Pm+1, Pm+2..., Pn, by calculating prediction data sequence { Pm+1, Pm+2..., PnAnd test data
{Tm+1, Tm+2..., TnError | Tm+1-Pm+1|, | Tm+2-Pm+2| ..., | Tn-Pn| verifying model accuracy rate, average absolute
Percentage error is as follows:
As best model Mp,qMean absolute percentage error mape within 5% just be applied to actual production.
(7) model prediction predicts industrial equipment vibration performance value in the following certain time period by the data model of foundation
Prediction data sequence, the formula of concrete foundation are as follows:
Wherein, p is known as Autoregressive,For autoregressive coefficient；Q is known as sliding average order,
{θ1, θ2..., θqIt is sliding average coefficient.Wherein etIt is independent error term, T "tFor steady, normal state and zero-mean it is following certain
The prediction data sequence of industrial equipment vibration performance value in one period.
Refering to attached drawing 1.Fig. 1 is a kind of entirety of industrial equipment vibration performance value trend forecasting method based on data model
Flow chart.The historical data and current data of vibrating sensor acquisition speed virtual value are installed first on certain industrial equipment, point
It Zuo Wei not training data, test data；Secondly stable data sequence is generated by calculus of differences to training data；Again into
The selection of row model, model training, model testing；Finally model verifying is carried out using test data.
Refering to attached drawing 2.Fig. 2 is to install vibrating sensor in the pump anti-drive end horizontal position of certain oil transportation station workshop oil transfer pump
Acquire the historical data and current data of speed virtual value.
Refering to attached drawing 3.Fig. 3 is using the historical data of sensor acquisition pump anti-drive end horizontal velocity virtual value and currently
Data, respectively as training data, test data.Form data sequence of as shown in Figure 3 two groups with time sequencing.
Refering to attached drawing 4- Fig. 6.Calculus of differences is carried out to collected data sequence, eliminates the unstable trend of data sequence
Property, make the change procedure stationary process of data sequence；The calculation formula of calculus of differences is T 't=Tt+1-Tt, in order to guarantee data
The stationarity of sequence needs to carry out calculus of differences twice to data sequence；Data sequence and original number after calculus of differences twice
It is more as shown in Figure 4-Figure 6 according to sequence.
Table 1 is the autoregressive order that stable data sequence after difference processing is calculated according to modes such as parameter Estimation, grid search
Number p and sliding average order q, sets p, q upper limit as 5, successively creation autoregressive moving-average model set.Then by second difference
Divide training data to be brought into the model in set, obtain bic score set, finds the corresponding model ARMA of score set minimum value
(1,1) it is used as best model, completes model training process.
Table 1
Fig. 7 be the following certain time period based on data model in oil transfer pump vibration velocity virtual value trend prediction result with
The line chart of actual result
Table 2 is specific Error List.
Table 2
Claims (8)
1. a kind of industrial equipment vibration performance value trend forecasting method based on data model, which is characterized in that including following step
It is rapid:
Step 1, by vibrating sensor, the characteristic value of the several data type signal of industrial equipment is acquired, is vibrated by choosing
The historical data and current data of characteristic value, historical data is as training data, and current data is as test data；
Step 2, the stationarity of training data is observed, if unstable, calculus of differences carried out, obtains second order difference training data；
Step 3, model selects: establish second order difference training data on autoregression model and moving average model basis,
Autoregression model and moving average model are combined, autoregressive moving-average model is constituted；
Step 4, model training: determining model parameter firstly the need of parameter Estimation, then according to bayesian information criterion function
Method determines the number of parameter；
Step 5, model testing: examine best model prediction error sequence randomness, that is, examine prediction error between whether
It is independent；
Step 6, model is verified, and the corresponding timestamp of test data is brought into best model and obtains prediction data sequence, is passed through
Calculate the accuracy rate of the error validity model of prediction data sequence and test data；
Step 7, model prediction predicts industrial equipment vibration performance value in the following certain time period by the data model of foundation
Prediction data sequence.
2. a kind of industrial equipment vibration performance value trend forecasting method based on data model according to claim 1,
It is characterized in that, in step 1, the data of acquisition include the characteristic value of acceleration, four seed type signal of speed, displacement and envelope；Choosing
The historical data of vibration performance value is taken to be denoted as { T0, T1..., TmAnd current data be denoted as { Tm+1, Tm+2..., Tn, form one group
Data sequence { T with time sequencing0, T1..., Tn}。
3. a kind of industrial equipment vibration performance value trend forecasting method based on data model according to claim 1,
It is characterized in that, in step 2, the calculation formula of calculus of differences are as follows:
T′t=Tt+1-Tt
Wherein T 'tFor t-th of element in the training data sequence after first difference, Tt+1For the t+1 in training data sequence
A element, t={ 0,1 ..., m-1 }；First difference training data { T ' is obtained after first difference0, T '1..., T 'm-1}；Second difference
Divide training data { T "0, T "1..., T "m-2}。
4. a kind of industrial equipment vibration performance value trend forecasting method based on data model according to claim 1,
It is characterized in that, in step 3, the correlation of this different times of expression time sequence is carried out using regression model:
Wherein, T "tIt is observed value of the time series in the t phase, p is the order of autoregression model, indicates T "tOnly with the p before it
Phase sequential value is related, etIt is error, indicates the enchancement factor that cannot use specification of a model；Due to second order difference training data { T "0,
T"1..., T "m-2It is original training data { T0, T1..., TmData sequence after second order difference, it is the data sequence of a zero-mean
Column, i.e.,Then:
Meanwhile second order difference training data { T "0, T "1..., T "m-2In T "tIt is built according to the principle of prediction average early period error
It is vertical, predicted value till now can be obtained plus prediction error on predicted value early period, can be obtained by recurrence:
T"t=et-θ1et-1-θ2et-2-…-θqet-q
Wherein T "tIt is observed value of the time series in the t phase, q is the order of moving average model, etIt is time series in the t phase
Error；
To sum up, second order difference training data { T "0, T "1..., T "m-2Establish in autoregression model and moving average model basis
On, autoregression model and moving average model are combined, autoregressive moving-average model is just constituted:
-θ1et-1-θ2et-2-…-θqet-q
Wherein, p is known as Autoregressive,For autoregressive coefficient；Q is known as sliding average order, { θ1,
θ2..., θqIt is sliding average coefficient.
5. a kind of industrial equipment vibration performance value trend forecasting method based on data model according to claim 1,
It is characterized in that, in step 4, determines model parameter: Autoregressive p and sliding average order q；According to bayesian information criterion letter
Number methods determine the number i.e. order of model of parameter, score evaluation criterion are as follows:
More thanIt is maximum likelihood function estimated value, score is smaller, and model is best；Set p, q
The upper limit successively creates autoregressive moving-average model set (M0,0, M0,1..., Mp,q)；Then by second order difference training data
{T"0, T "1..., T "m-2Successively bring into set in model, obtain bic score setFind the corresponding model M of set mid-score minimum valuep,qIt is complete as best model
At model training process.
6. a kind of industrial equipment vibration performance value trend forecasting method based on data model according to claim 1,
It being characterized in that, in step 5, randomness is examined by auto-relativity function method, that is, the autocorrelation function graph for the error sequence that gives a forecast,
If the case where being a kind of random variation between autocorrelation sequence in autocorrelation function graph, there is no truncation, hangover situation, then in advance
It surveys between error independently, best model is upchecked；Hangover refers to that autocorrelation sequence is decayed with index percent monotone decreasing or concussion, and
Truncation refers to that autocorrelation sequence becomes very small from some time point.
7. a kind of industrial equipment vibration performance value trend forecasting method based on data model according to claim 1,
It is characterized in that, in step 6, by test data { Tm+1, Tm+2..., TnCorresponding timestamp is brought into best model Mp,qIt obtains pre-
Measured data sequence { Pm+1, Pm+2..., Pn, by calculating prediction data sequence { Pm+1, Pm+2..., PnAnd test data { Tm+1,
Tm+2..., TnError | Tm+1-Pm+1|, | Tm+2-Pm+2| ..., | Tn-Pn| verifying model accuracy rate, average absolute percentage
Ratio error is as follows:
As best model Mp,qMean absolute percentage error mape within 5% just be applied to actual production.
8. a kind of industrial equipment vibration performance value trend forecasting method based on data model according to claim 1,
It is characterized in that, in step 7, the pre- of industrial equipment vibration performance value in the following certain time period is predicted by the data model of foundation
Measured data sequence, the formula of concrete foundation are as follows:
-θ1et-1-θ2et-2-…-θqet-q
Wherein, p is known as Autoregressive,For autoregressive coefficient；Q is known as sliding average order, { θ1,
θ2..., θqIt is sliding average coefficient；Wherein etIt is independent error term, T "tFuture for steady, normal state and zero-mean is a certain
The prediction data sequence of industrial equipment vibration performance value in period.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910812188.2A CN110532698B (en) | 2019-08-30 | 2019-08-30 | Industrial equipment vibration characteristic value trend prediction method based on data model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910812188.2A CN110532698B (en) | 2019-08-30 | 2019-08-30 | Industrial equipment vibration characteristic value trend prediction method based on data model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110532698A true CN110532698A (en) | 2019-12-03 |
| CN110532698B CN110532698B (en) | 2022-11-25 |
Family
ID=68665338
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910812188.2A Active CN110532698B (en) | 2019-08-30 | 2019-08-30 | Industrial equipment vibration characteristic value trend prediction method based on data model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110532698B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113221248A (en) * | 2021-05-21 | 2021-08-06 | 大连海事大学 | Ship system equipment state parameter prediction method based on PF-GARCH model |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20030009311A1 (en) * | 2001-03-23 | 2003-01-09 | Yukihiro Ushiku | Apparatus for predicting life of rotary machine, equipment using the same, method for predicting life and determining repair timing of the same |
| CN106920009A (en) * | 2017-03-03 | 2017-07-04 | 北京北青厚泽数据科技有限公司 | The Forecasting Methodology of hospital services amount |
-
2019
- 2019-08-30 CN CN201910812188.2A patent/CN110532698B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20030009311A1 (en) * | 2001-03-23 | 2003-01-09 | Yukihiro Ushiku | Apparatus for predicting life of rotary machine, equipment using the same, method for predicting life and determining repair timing of the same |
| CN106920009A (en) * | 2017-03-03 | 2017-07-04 | 北京北青厚泽数据科技有限公司 | The Forecasting Methodology of hospital services amount |
Non-Patent Citations (1)
| Title |
|---|
| 吴庚申等: "基于ARMA的汽轮机转子振动故障序列的预测", 《华南理工大学学报(自然科学版)》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113221248A (en) * | 2021-05-21 | 2021-08-06 | 大连海事大学 | Ship system equipment state parameter prediction method based on PF-GARCH model |
| CN113221248B (en) * | 2021-05-21 | 2024-05-03 | 大连海事大学 | Ship system equipment state parameter prediction method based on PF-GARCH model |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110532698B (en) | 2022-11-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP3809220B1 (en) | Method and system for semi-supervised deep anomaly detection for large-scale industrial monitoring systems based on time-series data utilizing digital twin simulation data | |
| CN102062832B (en) | Power system online prewarning method based on microdisturbance signal low-frequency oscillation mode identification | |
| JP5901140B2 (en) | Methods, computer programs, and systems for interpolating sensor data for high system availability. | |
| CN112799898B (en) | Interconnection system fault node positioning method and system based on distributed fault detection | |
| CN109061341A (en) | Kalman filtering transformer fault prediction technique and system neural network based | |
| CN109641602A (en) | Abnormality detecting apparatus, method for detecting abnormality and non-transitory computer-readable medium | |
| CN102937534B (en) | Method for predicting fault of electromechanical device based on combined prediction model | |
| CN1862278A (en) | Method and system for predicting remaining life for motors featuring on-line insulation condition monitor | |
| CN104462808A (en) | Method for fitting safe horizontal displacement and dynamic data of variable sliding window of water level | |
| CN115689393A (en) | Real-time dynamic monitoring system and method for power system based on Internet of things | |
| JP2019159604A (en) | Abnormality detection device, abnormality detection method and abnormality detection program | |
| EA036230B1 (en) | Method and system of remote object monitoring | |
| CN113239132A (en) | Online out-of-tolerance identification method for voltage transformer | |
| CN117270514B (en) | Production process whole-flow fault detection method based on industrial Internet of things | |
| CN108693469A (en) | The method for diagnosing faults and device of GIS device | |
| CN110188399B (en) | Dam safety monitoring single-measuring-point evaluation method based on multiple correlation sequences | |
| CN110532698A (en) | A kind of industrial equipment vibration performance value trend forecasting method based on data model | |
| CN117150244B (en) | Intelligent power distribution cabinet state monitoring method and system based on electrical parameter analysis | |
| CN114338348A (en) | Intelligent alarm method, device, equipment and readable storage medium | |
| CN116663747B (en) | Intelligent early warning method and system based on data center infrastructure | |
| JP2016177676A (en) | Diagnosis device, diagnosis method, diagnosis system and diagnosis program | |
| KR101886210B1 (en) | Measuring instrument reliability evaluation apparatus and operating method thereof | |
| CN111474476A (en) | Motor fault prediction method | |
| KR101137318B1 (en) | System and method for dignosis of semiconduct manufacturing apparatus | |
| KR102110319B1 (en) | System for generating learning data |
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 |