CN116429240A - Rotating equipment vibration signal prediction system based on deep Boltzmann machine - Google Patents
Rotating equipment vibration signal prediction system based on deep Boltzmann machine Download PDFInfo
- Publication number
- CN116429240A CN116429240A CN202111680810.2A CN202111680810A CN116429240A CN 116429240 A CN116429240 A CN 116429240A CN 202111680810 A CN202111680810 A CN 202111680810A CN 116429240 A CN116429240 A CN 116429240A
- Authority
- CN
- China
- Prior art keywords
- prediction
- boltzmann machine
- data
- prediction system
- rotating equipment
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000006073 displacement reaction Methods 0.000 claims abstract description 22
- 238000004458 analytical method Methods 0.000 claims description 3
- 238000005457 optimization Methods 0.000 claims description 3
- 238000001514 detection method Methods 0.000 claims description 2
- 230000005540 biological transmission Effects 0.000 claims 1
- 230000008054 signal transmission Effects 0.000 claims 1
- 238000000034 method Methods 0.000 abstract description 11
- 238000005553 drilling Methods 0.000 abstract description 8
- 239000003208 petroleum Substances 0.000 abstract description 4
- 238000009826 distribution Methods 0.000 description 14
- 238000012360 testing method Methods 0.000 description 9
- 238000010586 diagram Methods 0.000 description 5
- 238000012544 monitoring process Methods 0.000 description 5
- 238000012423 maintenance Methods 0.000 description 4
- 238000012549 training Methods 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- 238000006243 chemical reaction Methods 0.000 description 3
- 238000004140 cleaning Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 230000006378 damage Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- NAWXUBYGYWOOIX-SFHVURJKSA-N (2s)-2-[[4-[2-(2,4-diaminoquinazolin-6-yl)ethyl]benzoyl]amino]-4-methylidenepentanedioic acid Chemical compound C1=CC2=NC(N)=NC(N)=C2C=C1CCC1=CC=C(C(=O)N[C@@H](CC(=C)C(O)=O)C(O)=O)C=C1 NAWXUBYGYWOOIX-SFHVURJKSA-N 0.000 description 1
- 238000007476 Maximum Likelihood Methods 0.000 description 1
- 230000018199 S phase Effects 0.000 description 1
- 208000027418 Wounds and injury Diseases 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000013499 data model Methods 0.000 description 1
- 230000004069 differentiation Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000009776 industrial production Methods 0.000 description 1
- 208000014674 injury Diseases 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Images
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
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01B—MEASURING LENGTH, THICKNESS OR SIMILAR LINEAR DIMENSIONS; MEASURING ANGLES; MEASURING AREAS; MEASURING IRREGULARITIES OF SURFACES OR CONTOURS
- G01B21/00—Measuring arrangements or details thereof, where the measuring technique is not covered by the other groups of this subclass, unspecified or not relevant
- G01B21/02—Measuring arrangements or details thereof, where the measuring technique is not covered by the other groups of this subclass, unspecified or not relevant for measuring length, width, or thickness
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M99/00—Subject matter not provided for in other groups of this subclass
- G01M99/005—Testing of complete machines, e.g. washing-machines or mobile phones
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/21—Design, administration or maintenance of databases
- G06F16/215—Improving data quality; Data cleansing, e.g. de-duplication, removing invalid entries or correcting typographical errors
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Databases & Information Systems (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Quality & Reliability (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The invention discloses a rotating equipment vibration signal prediction system based on a depth Boltzmann machine, which belongs to the technical field of petroleum exploitation equipment and particularly comprises a displacement sensor, a data acquisition instrument and an online prediction system. The invention further provides a rotating equipment vibration signal prediction method based on the deep Boltzmann machine. The technical scheme device of the invention can be internally provided with the top drive key component, can be externally arranged on the shell, is small in size and easy to install, detach and operate on site, can accurately learn and fit the data rule of the tested sample, directly predicts the future trend of the data, achieves the purpose of predicting the future state of other key rotating components such as the top drive motor, timely maintains according to the prediction information, reduces the failure occurrence rate of rotating equipment such as the top drive, avoids casualties and economic losses, and improves the drilling efficiency. Has universal applicability to all critical components of rotary equipment.
Description
Technical Field
The invention relates to the technical field of petroleum exploitation equipment, in particular to a rotating equipment vibration signal prediction system based on a deep Boltzmann machine.
Background
The top driving well drilling device (hereinafter referred to as a top drive) is an important rotation device for petroleum exploitation, and is a power device which is placed at the top of a drilling tool in petroleum well drilling and directly drives the drilling tool to rotate instead of a turntable for well drilling construction. The power is typically a hydraulic motor or an electric motor. The electric top drive has the forms of direct current, alternating current frequency conversion and the like. The underground complex situation can be effectively reduced and prevented, and inverted reaming can be performed when slight collapse occurs; the drill string and circulating mud may be rotated continuously as necessary during tripping. If the running state of the device is good, the drilling operation is greatly influenced, if the top drive state cannot be estimated in time, the best maintenance time of the device can be measured, serious field accidents can be caused, and damage can be caused to personnel and economy. At present, the research on the fault prediction method of the top drive is weak, the method still basically depends on experience prediction of field personnel, the dependence on personnel is high, the subjectivity is high, the technical support is lacked, and the prediction accuracy of the fault is low.
The influence factors of the top drive running state are numerous, and the accuracy of prediction also depends on the data information of the top drive in a long time domain, so that the mass property of data and the analysis and the processing of diversity characteristics are required to be realized by means of a leading big data technology. The state prediction is carried out on the top drive rotating equipment based on the big data model depth Boltzmann machine, the characteristic information of data can be obtained from multi-layer machine learning, the prediction of the expected working state of the top drive can be realized more informationized and accurately, the health and long-period operation of the top drive can be ensured more reliably, the maintenance resources can be optimized more reasonably, and the optimal economic benefit is created for industrial production.
Disclosure of Invention
The invention aims to provide a rotary equipment vibration signal prediction system based on a deep Boltzmann machine, which predicts a top drive state by selecting proper influencing factors and makes correct judgment on an expected working state of the top drive state, so that a set of reasonable and scientific maintenance strategy is formulated in time, the problem of post-maintenance is overcome, the failure occurrence rate of rotary equipment such as top drive is reduced, casualties and economic losses are avoided, and the drilling efficiency is improved.
The invention provides a rotating equipment vibration signal prediction system based on a depth Boltzmann machine, which specifically comprises a displacement sensor 1, a data acquisition instrument 2 and an online prediction system 3. The displacement sensors 1 are arranged in the horizontal and vertical directions at the driving end and the non-driving end bearings of the top drive motor, and the number of the displacement sensors is not less than 4. The data acquisition instrument 2 can transmit the acquired signals of the displacement sensor 1 to the online prediction system 3 by adopting the prior art. And the online detection system 3 adopts a prediction model based on a depth Boltzmann machine to analyze and predict fault signals of the rotating equipment and displays faults. The data acquisition instrument 2 is a data acquisition instrument S8000.
Specifically, 4 measuring points are arranged in the horizontal direction of the driving end, the vertical direction of the driving end, the horizontal direction of the non-driving end and the vertical direction of the non-driving end of the motor of the rotary equipment. 4 measuring points in the front horizontal direction, the front vertical direction, the rear horizontal direction and the rear vertical direction of the low-speed teeth of the gear box; the front end of the high-speed tooth is in the horizontal direction, the front end of the high-speed tooth is in the vertical direction, the rear end of the high-speed tooth is in the horizontal direction, and the rear end of the high-speed tooth is in the vertical direction, and 4 measuring points are measured in the front end horizontal direction and the rear end vertical direction. The displacement sensor 1 is arranged at a measuring point.
Preferably, the displacement sensor is a wireless displacement sensor.
Further, the displacement sensor is installed on the casing in an internal or external mode.
The data acquisition instrument 2 stores the acquired data in a system database, and finds out a data set to be analyzed according to vibration index data provided by the database. And cleaning the monitoring data collected by the vibration sensors arranged at each measuring point of each device according to a specified data cleaning rule, rearranging the obtained data according to time sequence after cleaning, storing the data into a uniform format, and tabulating the data according to the measuring points of the device, wherein the obtained data is modeling data required by prediction.
The online prediction system 3 mainly provides a rotation equipment vibration signal prediction method based on a depth Boltzmann machine. The online prediction system 3 comprises a primary prediction model, a model optimization module, a fault prediction model module and a fault display module. The data collection device 2 collects the formed data set to be divided into a training data set and a prediction data set. The online prediction system 3 comprises a primary prediction model forming a primary training model through a training data set; and entering the optimized parameters into a primary prediction model module to obtain a fault prediction model module. The fault prediction model module may iteratively predict fault data in a single step or in multiple steps using the prediction dataset. And the fault display module displays the fault data on a human-computer interface.
The online prediction system 3 comprises an online prediction system for collecting real-time data of a sensor and extracting the rotating speed and the total vibration value at fixed time.
The prediction model based on the depth Boltzmann machine is formed by stacking a plurality of RBMs to form a multi-layer network, and the output of the former layer in the network model is the input of the latter layer.
The same monitoring signal is predicted by using the constructed Gaussian depth Boltzmann machine prediction model, so that the method can automatically learn in a large number of data sets, can realize the representation of the mass and the diversity of characteristic information in the automatic learning process, accurately captures the data distribution rule, and can well reflect the trend of the vibration signal by the result obtained by the generated prediction model.
The G-DBM model can directly learn input signals in different modes, and all characteristic information of the signals is learned through layer-by-layer characteristic transformation in the model learning process, so that a fault recognition result is obtained.
Specifically, the primary prediction model flow:
(1) Assuming N data samples are collected, the observed value of the time sequence is y i (i＝1，2，..，N y ) Given a series of iterative sequences of the system's phase space (the phase space iterative sequences represent the evolution of the orbit in the system) or a set of observed numerical sequences, a nonlinear map describing the original dynamics system is constructed, which can then be used as a primary predictive model.
y i (i+h)＝[y(i)，y(i-τ)，...，y(n-(m-1)τ)] T (1)
In the formula, tau is normalized embedded time delay width;
the reconstruction condition is that m is more than or equal to 2d 1 +1，d 1 Is the associated dimension of the power system;
h is the predicted number of steps.
(2) To build the prediction model, a multi-input, single-output nonlinear system model f is constructed: r is R q R may be used as a single-step prediction or as a multi-step prediction. I.e.
Where f is a prediction function, corresponding here to a deep boltzmann machine prediction model, τ is the delay, typically τ=1.
(3) Iterative prediction is carried out on the data, and ideal output is as follows:
through iterative algorithm, can obtain
Is a predicted result a of (a).
Specifically, the model optimization module flow is as follows:
the DBM model stacks a plurality of RBMs to form a multi-layer network, and the output of the former layer in the network model is the input of the latter layer.
Definition of visible and hidden layer nodes { v, h 1 ，h 2 ，h 3 The energy expression between } is:
wherein ψ= { W 1 ，W 2 ，W 3 ，B，A 1 ，A 2 ，A 3 Is a parametric model, W 1 ，W 2 ，W 3 Representing symmetrical connections of visible layer to hidden layer node and hidden layer to visible layer node, respectively, B, A 1 ，A 2 ，A 3 The threshold values of visible layer nodes and hidden layer nodes, respectively.
The probability distribution for a visible element vector V state of 1 is:
the conditional probability distribution of the hidden units j, m, l and the visible unit i when the state is 1 is:
performing maximum likelihood learning on the DBM model, taking logarithm of the formula (8), W 1 The partial differentiation is performed to obtain a parameter update rule as shown in formula (10), and an approximate expected value is obtained according to the formula.
Specifically, entering the optimized parameters into an initially built model to form the fault prediction model:
specifically, the fault point can be predicted in a single step:
in single-step prediction, only one future prediction value is output at a time. When a predicted value is output, the actual value at the next moment and other historical actual values form new test input data, the new test input data is input into a network, and then the next prediction is performed.
Single-step iteration prediction is carried out on the data, and ideal output is as follows:
the predicted result can be obtained through a single-step iterative algorithm:
where f is a prediction function, corresponding here to a boltzmann machine prediction model of gaussian depth, τ is the delay, typically τ=1.
After reconstruction of the data, the data is divided into N-m input samples, and it can be determined that the G-DBM prediction model has m input nodes, one output node. When the prediction test is finally performed, the first input sample is y1, y2, …, ym-1, the target output is ym, the actual output is am, namely the second input sample of the network predicted data is y2, y3, …, ym, the target output is ym+1, the actual output is am+1, and the last input sample is y N-m+1 ，y N-m+2 ，..，y N-1 The target output is y N The actual output is a N I.e. predicting the data value at the next instant as a N 。
Multi-step prediction using a G-DBM prediction model can be divided into Multi-step iterative prediction and Multi-step direct prediction of Multiple Input and Multiple Output (MIMO).
Further, a multi-step prediction iteration strategy can also be used for prediction:
in multi-step iterative prediction, similar to single-step prediction, only one future predicted value is output in each iterative prediction, but when one predicted value is output, the predicted value and other historical true values are combined together to form new test input data, the new test input data is input into a network, and then the next-step prediction is performed until the step length of the multi-step prediction is reached.
Performing multi-step iterative prediction on the data, wherein ideal output is as follows:
through an iterative algorithm, a multi-step prediction result can be obtained:
where 7 is the prediction function, corresponding here to the G-DBM prediction model, τ is the delay, typically τ=1.
After reconstruction of the data, the data is again divided into N-m input samples, and it is also determined that the G-DBM prediction model has m input nodes, one output node. However, when the prediction test is finally performed, the first input sample is test data y1, y2, …, ym-1, the target output is ym, and the actual output is am, namely the predicted data value at the next moment; then updating the next input sample with the predicted value to obtain a second input sample of y2, y3, …, a m The target output is y m+1 The actual output is a m+1 And so on, the last input sample is y N-m+1 ，y N-m+2 ，…，a N-1 The target output is y N The actual output is a N I.e. predicting the data value at the next instant as a N 。
The embodiment of the invention has the beneficial effects that:
the invention provides a rotating equipment vibration signal prediction system based on a depth Boltzmann machine, which can be internally arranged in a top drive key component, can be externally arranged in a shell, is small in size and easy to install, detach and operate on site. Has universal applicability to all critical components of rotary equipment.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings that are needed in the embodiments will be briefly described below, it being understood that the following drawings only illustrate some embodiments of the present invention and therefore should not be considered as limiting the scope, and other related drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a diagram of a displacement sensor derrick layout of the present invention;
FIG. 2 is a graph of displacement sensor site locations in accordance with the present invention;
FIG. 3 is a block diagram of a rotating equipment vibration signal prediction system based on a depth Boltzmann machine of the present invention;
FIG. 4 is a graph of the results of a single step prediction of the present invention;
FIG. 5 is a Gaussian depth Boltzmann machine model diagram of the multi-step prediction iteration strategy of the present invention, wherein (a) the delay window is shifted 24 times the prediction contrast, (b) the delay window is shifted 24 times the prediction error distribution, (c) the delay window is shifted 96 times the prediction contrast, and (d) the delay window is shifted 96 times the prediction error distribution;
FIG. 6 is a Gaussian depth Boltzmann machine model diagram of the multi-step predictive direct strategy of the invention;
FIG. 7 is a graph of 24-step predictions under two strategies of the multi-step prediction of the present invention;
FIG. 8 is a graph of a 96-step prediction error distribution under two strategies of the multi-step prediction of the present invention, wherein (a) the first strategy multi-step prediction iterates the strategy prediction results, (b) the first strategy delay time window moves 24 error distributions, (c) the second strategy delay time window moves 96 multi-step prediction direct strategy prediction results, (d) the second strategy delay time window moves 96 error distributions;
FIG. 9 is a 96-step prediction error distribution under two strategies of multi-step prediction, wherein (a) the strategy error distribution is iterated; (b) direct policy error distribution.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments of the present invention. The components of the embodiments of the present invention generally described and illustrated in the figures herein can be arranged and designed in a wide variety of different configurations.
Referring to fig. 1-2, a first embodiment of the present invention is shown.
The invention provides a rotating equipment vibration signal prediction system based on a depth Boltzmann machine, which mainly comprises a displacement sensor 1, a data acquisition instrument 2 and an online prediction system 3. 4 measuring points are arranged in the horizontal and vertical directions at the driving end and the non-driving end bearing of the top drive motor, a displacement sensor is arranged (the displacement sensor can be internally arranged or externally arranged in a shell form), the top drive runs in a no-load mode, an online prediction system collects real-time data of the sensor, rotation speed and vibration values are extracted at fixed time, and data operation software can be based on any ground operation system or an independent computer.
The method of the online prediction system is specifically shown in fig. 2, and the measuring points are 4 measuring points in the horizontal direction of the driving end, the vertical direction of the driving end, the horizontal direction of the non-driving end and the vertical direction of the non-driving end of the motor 4 of the rotating equipment. 4 measuring points in the front horizontal direction, the front vertical direction, the rear horizontal direction and the rear vertical direction of the low-speed teeth of the gear box 5; the front end of the high-speed tooth is in the horizontal direction, the front end of the high-speed tooth is in the vertical direction, the rear end of the high-speed tooth is in the horizontal direction, and the rear end of the high-speed tooth is in the vertical direction, and 4 measuring points are measured in the front end horizontal direction and the rear end vertical direction. A total of 12 measuring points, namely the numbers 1 to 6 in the box in fig. 2, 6 points are provided with measuring points in the x-axis and y-axis directions, respectively, at which the displacement sensor 1 is arranged.
The rotational speed and total vibration values in the table are extracted at regular intervals (1 time every 5 minutes, which can be set).
The data set to be analyzed is found out according to the vibration index data (shown in table 1) provided by the database.
And reading the SQL database to obtain vibration monitoring data of the top drive, reading vibration values corresponding to measuring points distributed on the motor and the low-speed high-speed gearbox according to a vibration channel corresponding table of the top drive, and storing all read data conversion formats.
Table 1 vibration index data provided by the database
The monitoring data of each sensor are extracted, the first 50000 data are selected as training sets, and 50001 to 50100 data segments are selected as test sets. 100 data points are predicted. After the data is ready, the data is input into a G-DBM prediction model to perform prediction. And primarily selecting the number of neurons of each layer in the G-DBM predictive model of the top drive rotating equipment according to a large number of experiments, and primarily constructing a proper predictive model.
The single-step prediction predicts only one future value at a time, predicts by a single-step Gaussian depth Boltzmann machine model, and iteratively predicts 24 times in a time-delay window to obtain 24 single-step predicted values within two hours, as shown in fig. 4, where (a) in fig. 5 is a comparison graph of the monitored value obtained by single-step prediction every five minutes and the predicted value, and (b) in fig. 5 is a prediction error distribution of 24 points within two hours.
The delay time window is moved 96 times, the prediction result of the single step prediction for 96 steps, namely eight hours is shown in fig. 7, the (c) in fig. 5 is a comparison graph of the monitored value and the predicted value obtained by the single step prediction every five minutes, and the (d) in fig. 5 is the prediction error distribution of 96 points within eight hours.
From 24 predicted values and 96 predicted values obtained by single-step prediction, the single-step prediction has great application value in evaluating the adaptability and the robustness of a predicted model, and even if an undesirable predicted value is output in one step of prediction, the single-step prediction does not cause chain reaction and cause the error of the subsequent prediction, because the true value is used for the next step of test. Through the model research of single-step prediction of the G-DBM, a very good prediction result can be obtained, and the next vibration value of the vibration data of the top drive rotating equipment can be predicted very accurately.
In order to better prove the effectiveness of the method, the example operation analysis is carried out, and the ideal future data and the predicted actual data are obtained through the collected data and the operation. In order to verify the accuracy and the effectiveness of the method, the prediction effect is more intuitively evaluated, and three prediction error evaluation indexes are introduced:
sum of squares error:
average absolute relative error:
correlation coefficient:
wherein x is r (i) Is ideal for transportationThe ith sample data, x p (i) Is the predicted i-th sample data.
The smaller the sum of squares of the errors and the average absolute relative error, the better the prediction effect of the model. The indices for the 24 values and 96 values for the single step prediction are shown in table 2. The closer the correlation coefficient is to 1, the more similar the predicted data sequence is to the historical data sequence.
TABLE 2 error index for single step prediction
By comparison, it is evident that the single-step prediction results are very accurate.
The multi-step prediction may predict a plurality of future values of the top drive rotational apparatus monitoring data, and predict a trend of vibration of the apparatus over a period of time in the future. According to the method, a multi-step iteration strategy and a direct strategy G-DBM prediction model are respectively constructed, schematic diagrams are respectively shown in fig. 6 and 7, vibration data of top drive rotating equipment are predicted, the vibration trend of the top drive rotating equipment for two hours in the future is predicted, the vibration data for two hours can be calculated to be 24 according to the data acquired every 5 minutes, and a multi-step G-DBM prediction result with a prediction step length of 24 is obtained as shown in fig. 8.
The multi-step prediction iteration strategy and multi-step prediction direct strategy predict 96 point error distributions as shown in fig. 9. The prediction results under the two strategies of multi-step prediction can be seen that when predicting future values of step sizes within one to two hours, the iteration strategy of multi-step prediction can obtain more accurate results, and more accurate prediction values can be provided for running vibration values of the top drive rotating equipment within one or two hours in the future, but because of continuous superposition of errors, the larger the prediction step sizes, the more the prediction values are, and the inaccurate prediction is caused. The direct strategy of multi-step prediction has the advantages that although the prediction error is larger than the error of the iterative strategy when the step length is smaller than 24 steps, the prediction model under the strategy well avoids the iteration of the error, and the situation that the predicted value diverges due to the increase of the prediction step length is avoided, so that the model under the strategy can provide a more reliable theoretical basis for the running state of the top drive rotating equipment for more than 8 hours in the future.
TABLE 3 error index for multi-step prediction iteration strategy
TABLE 4 error index for multi-step prediction direct strategy
Fault display
Through the prediction of the signal vibration value of the top drive rotating equipment, a future running data set of the top drive rotating equipment can be obtained, whether the future running state of the top drive rotating equipment can be predicted to be faulty or not can be predicted by observing whether the data set has a value or trend exceeding a normal vibration threshold value, and safety loss and personnel injury caused by major equipment faults can be avoided by performing the intervention in time through the prediction.
It is to be understood that the above-described embodiments of the present invention are merely illustrative of or explanation of the principles of the present invention and are in no way limiting of the invention. Accordingly, any modification, equivalent replacement, improvement, etc. made without departing from the spirit and scope of the present invention should be included in the scope of the present invention. Furthermore, the appended claims are intended to cover all such changes and modifications that fall within the scope and boundary of the appended claims, or equivalents of such scope and boundary.
Claims (6)
1. The utility model provides a rotary equipment vibration signal prediction system based on degree of depth Boltzmann machine, its characterized in that includes displacement sensor (1), data acquisition appearance (2), online prediction system (3), displacement sensor (1) are at the drive end of top drive motor, the level of non-drive end bearing department, the vertical direction is arranged, its quantity is not less than 4, data acquisition appearance (2) are with displacement sensor (1) signal transmission who gathers to online prediction system (3), online detection system (3) adopts the prediction model based on degree of depth Boltzmann machine to carry out analysis prediction to rotary equipment fault signal to the fault is shown.
2. The rotating equipment vibration signal prediction system based on the depth boltzmann machine according to claim 1 is characterized in that the displacement sensor (1) is arranged at a rotating equipment motor and a gear transmission system in an internal mounting mode or an external mounting mode.
3. The rotating equipment vibration signal prediction system based on the depth boltzmann machine according to claim 1, wherein the online prediction system (3) comprises a primary prediction model, a model optimization module, a fault prediction model module and a fault display module.
4. The rotating equipment vibration signal prediction system based on the depth boltzmann machine according to claim 1, wherein the data acquisition instrument (2) is a data acquisition instrument S8000.
5. The rotating equipment vibration signal prediction system based on the depth boltzmann machine according to claim 1, wherein the displacement sensor is a wireless displacement sensor.
6. The rotating equipment vibration signal prediction system based on the depth boltzmann machine according to claim 1, wherein the displacement sensor is installed in the casing in an internal or external mode.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111680810.2A CN116429240A (en) | 2021-12-31 | 2021-12-31 | Rotating equipment vibration signal prediction system based on deep Boltzmann machine |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111680810.2A CN116429240A (en) | 2021-12-31 | 2021-12-31 | Rotating equipment vibration signal prediction system based on deep Boltzmann machine |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116429240A true CN116429240A (en) | 2023-07-14 |
Family
ID=87093072
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111680810.2A Pending CN116429240A (en) | 2021-12-31 | 2021-12-31 | Rotating equipment vibration signal prediction system based on deep Boltzmann machine |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116429240A (en) |
-
2021
- 2021-12-31 CN CN202111680810.2A patent/CN116429240A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111596604B (en) | Intelligent fault diagnosis and self-healing control system and method for engineering equipment based on digital twinning | |
| CN109508827B (en) | Drilling accident early warning method based on time recursion neural network | |
| CN108008332A (en) | A kind of new energy Remote testing device method for diagnosing faults based on data mining | |
| CN105300692B (en) | A kind of bearing failure diagnosis and Forecasting Methodology based on expanded Kalman filtration algorithm | |
| CN105550943A (en) | Method for identifying abnormity of state parameters of wind turbine generator based on fuzzy comprehensive evaluation | |
| CN109858140B (en) | Fault diagnosis method for water chilling unit based on information entropy discrete Bayesian network | |
| CN102252843B (en) | Assessment method for rolling bearing performance variation | |
| CN113339204B (en) | Wind driven generator fault identification method based on hybrid neural network | |
| CN101799674A (en) | Method for analyzing service state of numerical control equipment | |
| CN116358871B (en) | Rolling bearing weak signal composite fault diagnosis method based on graph rolling network | |
| CN114519923A (en) | Intelligent diagnosis and early warning method and system for power plant | |
| CN116665421A (en) | Early warning processing method and device for mechanical equipment and computer readable storage medium | |
| CN115234220A (en) | Method and device for identifying underground stick-slip vibration in real time by using intelligent drill bit | |
| CN117273440A (en) | Engineering construction Internet of things monitoring and managing system and method based on deep learning | |
| CN117846894A (en) | Data processing method, early warning device, equipment and medium of wind turbine generator | |
| CN114462820A (en) | Bearing state monitoring and health management system performance testing and optimizing method and system | |
| CN117930815A (en) | Wind turbine generator remote fault diagnosis method and system based on cloud platform | |
| CN117471346A (en) | Method and system for determining remaining life and health status of retired battery module | |
| CN116862076A (en) | Drainage pipe network flow prediction method, device and storage medium | |
| CN116429240A (en) | Rotating equipment vibration signal prediction system based on deep Boltzmann machine | |
| CN116821828A (en) | Multi-dimensional time sequence prediction method based on industrial data | |
| CN116680607A (en) | On-line monitoring method for motor state of drilling pump based on LSTM neural network | |
| CN114320773B (en) | Wind turbine generator system fault early warning method based on power curve analysis and neural network | |
| CN115469643A (en) | Nuclear power station rotating machinery health management method, system and medium | |
| CN111946258B (en) | GRU-based sliding orientation intelligent control 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 |