CN115326397A - Method for establishing crankshaft bearing wear degree prediction model and prediction method and related device - Google Patents
Method for establishing crankshaft bearing wear degree prediction model and prediction method and related device Download PDFInfo
- Publication number
- CN115326397A CN115326397A CN202210903822.5A CN202210903822A CN115326397A CN 115326397 A CN115326397 A CN 115326397A CN 202210903822 A CN202210903822 A CN 202210903822A CN 115326397 A CN115326397 A CN 115326397A
- Authority
- CN
- China
- Prior art keywords
- vibration signal
- crankshaft bearing
- wear degree
- bearing wear
- decomposition
- 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
- 238000000034 method Methods 0.000 title claims abstract description 74
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 66
- 230000015654 memory Effects 0.000 claims abstract description 42
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 41
- 238000013528 artificial neural network Methods 0.000 claims abstract description 33
- 238000002922 simulated annealing Methods 0.000 claims abstract description 32
- 238000000605 extraction Methods 0.000 claims abstract description 25
- 230000001133 acceleration Effects 0.000 claims abstract description 19
- 238000012545 processing Methods 0.000 claims abstract description 15
- 230000008569 process Effects 0.000 claims description 7
- 230000006403 short-term memory Effects 0.000 abstract description 5
- 230000007787 long-term memory Effects 0.000 abstract description 4
- 230000006870 function Effects 0.000 description 22
- 238000005299 abrasion Methods 0.000 description 8
- 239000000243 solution Substances 0.000 description 8
- 238000012360 testing method Methods 0.000 description 8
- 239000011159 matrix material Substances 0.000 description 7
- 238000012549 training Methods 0.000 description 6
- 230000009471 action Effects 0.000 description 4
- 238000003745 diagnosis Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 239000013598 vector Substances 0.000 description 3
- 238000000137 annealing Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 230000000295 complement effect Effects 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 238000012843 least square support vector machine Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000000644 propagated effect Effects 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 230000002776 aggregation Effects 0.000 description 1
- 238000004220 aggregation Methods 0.000 description 1
- 230000001174 ascending effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000009795 derivation Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000002347 injection Methods 0.000 description 1
- 239000007924 injection Substances 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 101150112154 outJ gene Proteins 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M13/00—Testing of machine parts
- G01M13/04—Bearings
- G01M13/045—Acoustic or vibration analysis
Landscapes
- Physics & Mathematics (AREA)
- Acoustics & Sound (AREA)
- General Physics & Mathematics (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The invention relates to a method for establishing a crankshaft bearing wear degree prediction model, a prediction method and a related device, and the method for establishing the crankshaft bearing wear degree prediction model comprises the following steps: acquiring a diesel engine vibration signal x (t) acquired by an industrial acceleration sensor; carrying out wavelet threshold denoising on the obtained vibration signal x (t), and selecting a soft threshold method with a threshold as an unbiased likelihood estimation threshold for processing to obtain a processed vibration signal x' (t); decomposing by using a set empirical mode decomposition optimized and improved by a simulated annealing algorithm, and performing feature extraction on the processed vibration signal x' (t) by using a singular value decomposition method and storing the feature extraction into a feature set; and (3) taking the feature set as the input of the long-short term memory artificial neural network model, and optimizing the long-short term memory artificial neural network model by using a simulated annealing algorithm to generate a crankshaft bearing wear degree prediction model. The invention uses the simulated annealing algorithm to search the optimal parameters of the long-term and short-term memory artificial neural network, and can effectively improve the prediction accuracy of the system.
Description
Technical Field
The invention relates to a diesel engine, in particular to a model and a method for predicting the wear degree of a crankshaft bearing and a related device.
Background
The diesel engine is a power source of a plurality of vehicles, potential safety hazards are brought to personnel on the vehicles due to faults of the diesel engine, the traditional mechanical fault diagnosis method needs a large amount of manpower and material resources due to complex structure, machines need to be disassembled, even parts are damaged, the fault diagnosis method based on vibration signals can avoid the problems, disassembly-free detection is achieved, and a large amount of cost is saved. Therefore, fault diagnosis methods based on vibration signals have become a hot spot of research by many scholars in recent years.
Chen Y and the like analyze and process the vibration signals by Fast Fourier Transform (FFT), but the characteristics of the method have the characteristic of easy aggregation, can not be well identified and need manual intervention; the method adopts wavelet packet-AR spectrum to extract the characteristics of the bearing faults of the speed changer, but has the defect of difficulty in selecting wavelet base; cheng J et al, which decompose the vibration signal using Empirical Mode Decomposition (EMD), but EMD has modal aliasing; xia, W and the like use self-adaptive wavelets and Ensemble Empirical Mode Decomposition (EEMD) to decompose signals and analyze the size of an oil injection advance angle by using a fault tree, but because actual operation is limited, the influence of added white noise cannot be well eliminated, reconstruction errors can be caused, although a Complementary Ensemble Empirical Mode Decomposition (CEEMD) can well eliminate the influence of the white noise on the signals but pseudo components, zhengyude and the like can be generated, the problems can be relieved by using an improved Ensemble Empirical Mode Decomposition (CPEMD), but the problems of more parameters and poor stability can be generated, the ultra-short-period prediction can be performed on wind power by using the improved Ensemble Empirical Mode Decomposition and Least Square Support Vector Machine (LSSVM), but the fitting precision of the algorithm is in direct proportion to the number of sampling points, and more data is needed.
Disclosure of Invention
In view of the above, the present invention provides a model and a method for predicting wear of a crankshaft bearing, and a related device.
In order to solve the above problems, the technical scheme provided by the invention is as follows:
in a first aspect, the present invention provides a method for establishing a model for predicting wear of a crankshaft bearing, comprising the steps of:
s11: acquiring a diesel engine vibration signal x (t) acquired by an industrial acceleration sensor;
s12: carrying out wavelet threshold denoising on the obtained vibration signal x (t), and selecting a soft threshold method with a threshold as an unbiased likelihood estimation threshold to process to obtain a processed vibration signal x' (t);
s13: decomposing by using the ensemble empirical mode decomposition optimized and improved by the simulated annealing algorithm, and performing feature extraction on the processed vibration signal x' (t) by using a singular value decomposition method and storing the extracted feature in a feature set;
s14: and taking the feature set as the input of the long-short term memory artificial neural network model, and optimizing the long-short term memory artificial neural network model by using a simulated annealing algorithm to generate a crankshaft bearing wear degree prediction model.
Further, the step S13 includes the steps of:
s131: processing the vibration signal x' (t) by improved set empirical mode decomposition;
s132: optimizing the improved set empirical mode decomposition pair processed vibration signal x' (t) by a simulated annealing algorithm;
s133: performing feature extraction on the processed vibration signal x' (t) by using a singular value decomposition method;
s134: and storing the features obtained by decomposition into a feature set.
In a second aspect, the present invention provides a method for predicting wear degree of a crankshaft bearing, comprising the steps of:
s21: acquiring a diesel engine vibration signal x (t) acquired by an industrial acceleration sensor;
s22: inputting the diesel engine vibration signal x (t) into a crankshaft bearing wear degree prediction model, and outputting a prediction result of the crankshaft bearing wear degree, wherein the crankshaft bearing wear degree prediction model is established according to the method for establishing the crankshaft bearing wear degree prediction model in any one of claims 1-2.
In a third aspect, the present invention provides an apparatus for modeling wear of a crankshaft bearing, the apparatus comprising:
the acquisition unit is used for acquiring a diesel engine vibration signal x (t) acquired by the industrial acceleration sensor;
the processing unit is used for carrying out wavelet threshold denoising on the obtained vibration signal x (t), and selecting a soft threshold method with the threshold as an unbiased likelihood estimation threshold for processing to obtain a processed vibration signal x' (t);
and the extraction unit is used for decomposing the set empirical mode decomposition optimized and improved by using the simulated annealing algorithm, and extracting and storing the characteristics of the processed vibration signal x' (t) into a characteristic set by using a singular value decomposition method.
And the generating unit is used for taking the vibration signal characteristics with the concentrated characteristics as the input of the long-short term memory artificial neural network model, optimizing the long-short term memory artificial neural network model by using a simulated annealing algorithm and generating a crankshaft bearing wear degree prediction model.
Further, the extraction unit includes:
a first decomposition unit for improved collective empirical mode decomposition processing of the vibration signal x' (t);
the second decomposition unit is used for simulating an annealing algorithm to optimize the improved set empirical mode decomposition to the processed vibration signal x' (t);
an extraction subunit, configured to perform feature extraction on the processed vibration signal x' (t) by using a singular value decomposition method;
and the storage unit is used for storing the features obtained by decomposition into the feature set.
In a fourth aspect, the present invention provides a crankshaft bearing wear degree prediction device, including:
the vibration signal acquisition unit is used for acquiring a diesel engine vibration signal x (t) acquired by the industrial acceleration sensor;
a prediction result output unit for inputting the diesel engine vibration signal x (t) collected by the industrial acceleration sensor into a crankshaft bearing wear degree prediction model, which is established according to the method for establishing a crankshaft bearing wear degree prediction model of any one of claims 1 to 2, and outputting a crankshaft bearing wear degree prediction result.
In a fifth aspect, the present invention provides a terminal device, including:
a processor and a memory;
the memory is used for storing program codes and transmitting the program codes to the processor;
the processor is used for calling the instructions in the memory to execute the method for establishing the crankshaft bearing wear degree prediction model according to any one of claims 1-2.
In a sixth aspect, the present invention provides a terminal device, including:
a processor and a memory;
the memory is used for storing program codes and transmitting the program codes to the processor;
the processor is used for calling instructions in the memory to execute the crankshaft bearing wear degree prediction method in claim 3.
Therefore, the invention has the following advantages and beneficial effects:
the method adopts the combination of singular value decomposition and ensemble empirical mode decomposition optimized and improved by a simulated annealing algorithm to extract the characteristics of the signals, and uses the simulated annealing algorithm to optimize the long-term and short-term memory artificial neural network to predict the faults. The method comprises the steps of decomposing a vibration signal of a crankshaft bearing of the diesel engine by using various decomposition methods, analyzing various indexes, selecting an optimal decomposition method, optimizing and decomposing by using a simulated annealing algorithm, extracting the characteristics of data by using a singular value decomposition method, and taking the data as the characteristic set of a neural network module. The long-short term memory artificial neural network is suitable for multi-classification problems and is suitable for predicting the wear degree of a crankshaft bearing. In addition, the optimal parameters of the long-term and short-term memory artificial neural network are searched by utilizing a simulated annealing algorithm, and the prediction accuracy of the system can be effectively improved.
The improved set empirical mode decomposition SA-CPEMD optimized by the simulated annealing algorithm can solve the problems of excessive CPEMD parameters and poor stability;
the CPEMD can reduce the problem of false components caused by excessive noise added by EEMD and CEEMD;
the mode confusion phenomenon of the EMD, CPEMD can better eliminate the phenomenon, so that the reconstruction error is smaller and the completeness is better.
2) SVD singular value decomposition has good stability, and SA-CPEMD-SVD can well extract the characteristics of the vibration signal.
3) The LSTM long-term and short-term artificial neural network can process input with any length, is also suitable for multi-output problems, and is suitable for predicting the wear degree of a crankshaft bearing.
4) The long-short-term artificial neural network optimized by the SA-LSTM simulated annealing algorithm improves the prediction accuracy of the original long-short-term artificial neural network model.
Drawings
The invention is described in further detail below with reference to the following figures and examples:
FIG. 1 is a flow chart of a method of establishing a crankshaft bearing wear prediction model;
FIG. 2 is a diagram of the structure of the LSTM;
FIG. 3 is a flow chart for optimizing a long-short term memory artificial neural network using a simulated annealing algorithm:
FIG. 4 is a flowchart of a method for predicting wear of a crankshaft bearing;
FIG. 5 is a block diagram of an apparatus for modeling wear of a crankshaft bearing;
fig. 6 is a block diagram of the crankshaft bearing wear degree prediction device.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Example 1
As shown in fig. 1, a method for establishing a wear degree prediction model of a crankshaft bearing of the present embodiment, taking a diesel engine as an example, includes the following steps:
s11: placing an industrial acceleration sensor at a specified test point of a diesel engine to be researched, acquiring a collected diesel engine vibration signal x (t) collected by the industrial acceleration sensor, and taking the collected signal as an original signal of a system; in the embodiment, an industrial acceleration sensor is placed at the left and right positions of a 3-cylinder joint of an oil bottom and a cylinder body, and is attracted to the surface of a test part by additionally arranging a strong magnetic seat.
S12: due to the fact that noise interference exists outside, wavelet threshold denoising is conducted on the obtained vibration signal x (t), and a soft threshold method with the threshold as an unbiased likelihood estimation threshold is selected for processing to obtain a processed vibration signal x' (t).
S13: decomposing by using a set empirical mode decomposition optimized and improved by a simulated annealing algorithm, and performing feature extraction on the processed vibration signal x' (t) by using a singular value decomposition method and storing the feature extraction into a feature set;
s14: and taking the feature set as the input of the long-short term memory artificial neural network model, optimizing the long-short term memory artificial neural network model by using a simulated annealing algorithm, and generating a long-short term memory artificial neural network (SA-LSTM) model optimized based on the simulated annealing algorithm, wherein the long-short term memory artificial neural network (SA-LSTM) model optimized based on the simulated annealing algorithm is the crankshaft bearing wear degree prediction model.
Wherein the step S13 includes the steps of:
s131: processing the vibration signal x' (t) by improved set empirical mode decomposition;
with a first vibration signal x' 1 (t) for example, vibration signal x 'is first analyzed using a Complementary Ensemble Empirical Mode Decomposition (CEEMD)' 1 (t), decomposing, wherein t is the sequence length, and the method comprises the following steps:
white noise with random equal length, given standard deviation and opposite signAnd added to the vibration signal: to obtainWherein the content of the first and second substances,is the u-th 1 Secondly, adding the white noise to the signal; decomposing a signal after positive white noise added for the first time by using an EMD algorithmFor example:i=1,h 0 =r i-1 (t),j=1。
obtaining the ith intrinsic mode function, comprising the following steps:
(a) Finding the intermediate variable transition action h j-1 (t) local extrema;
(b) To h is paired j-1 (t) performing cubic spline function interpolation on the maximum value point and the minimum value point respectively to form an upper envelope line and a lower envelope line;
(c) Calculate the mean m of the upper and lower envelopes j-1 (t);
(d) Let h j (t)=h j-1 (t)-m j-1 (t);
(e) If h j (t) is the eigenmode function, then imf i (t)=h j (t); otherwise, j = j +1, go to step (a).
Let r be i (t)=r i-1 (t)-imf i (t)。
If r is i (t) electrodeIf the number of the value points is still more than 2, i = i +1, and the step of continuously finding the intrinsic mode function is carried out; otherwise, the decomposition is finished and finally r is obtained i (t) is the residual component. r is n The (t) residual component is typically denoted by r, e.g., x = imf1+ imf2+ \8230, + imfn + r. The algorithm finally yields: obtaining a preset number of intrinsic mode function components and remainder.
By analogy, the following steps are carried outJ (j) decomposed by empirical mode decomposition 1 The intrinsic mode function components are respectivelyAndthe remainder components are respectivelyAndand summing the intrinsic mode function components and the remainder and calculating the average value of the intrinsic mode function components and the remainder to obtain:
to the j th final result 1 An intrinsic mode function; r is the decomposed remainder component. k is the number of added positive white noises, where the number of added negative white noises is equal to the number of added positive white noises.
Sequential detectionTo a permutation entropy value ofFirst eigenmode function of bar signalConsider, as an example, a time series of length NSpatial reconstruction is carried out on the data to obtain the following time sequence:
…
…
wherein: m is the embedding dimension and λ is the time delay.
If present, isThen press j * Is arranged in terms of the magnitude of the value of (i.e. when j is * w1 <j * w2 In time, there are:
therefore, any one vector IMF (i) * ) A set of symbol sequences is obtained: s (g) = [ J ] 1 ,J 2 ,…,J m ]Total m! The probability P of occurrence of k symbol sequences is correspondingly shared by different permutations 1 ,P 2 ,…,P k Wherein:wherein N is g K is the total number of the types (k is less than or equal to m) after repeated arrangement is removed for the times of arrangement in the g formula,correspond toTime seriesThe permutation entropy of (a) can be defined in terms of Shannon entropy as:
when P is g =1/m! When Hp (m) reaches a maximum value ln (m!), the arrangement entropy Hp (m) is normalized by ln (m!), that is: hp = Hp (m)/ln (m!) isThe rank entropy value of (c).
The permutation entropy reference value is set to 0.6 and the resulting permutation upper value of the eigenfunctions is compared with the reference value. If the eigen function component is larger than the reference value, the eigen function component is a singular component, otherwise, the eigen function component is a non-singular component. After the first singular component is obtained, the permutation entropy calculation step is repeated on the intrinsic mode function component until the non-singular component is obtained.
X 'will be original signal' 1 (t) subtracting all singular components to obtain a new signal x "", with the singular components removed 1 (t) for x ″) 1 And (t) performing EMD decomposition to obtain the final required intrinsic mode function.
S132: optimizing the improved set empirical mode decomposition pair processed vibration signal x' (t) by a simulated annealing algorithm;
in order to solve the problem of poor multi-stability of improved ensemble empirical mode decomposition (CPEMD) parameters, optimized and improved ensemble empirical mode decomposition is carried out by using a simulated annealing algorithm (SA), an ensemble average parameter N is added with a white noise logarithm Ne, an entropy order m is replaced, an arrangement entropy threshold theta0 is an independent variable, a mean square error MSE and an orthogonality ort are taken as indexes, Y = MSE +8 × ort, and a Y minimum value is found by using the simulated annealing algorithm and an optimal parameter is solved.
Y is used as an objective function f () of the simulated annealing algorithm, then the set average parameter N is set as S1, the added white noise logarithm Ne is S2, the substitution entropy level m is S3, the permutation entropy threshold theta0 is S4, and S = [ S1S 2S 3S 4] comprises the following algorithm steps:
initializing: initial temperature T 0 (high enough), initial solution S (iteration start), iteration number L, termination temperature T 1 。
The sixth step is the sixth step.
And producing novel S'. S ' = [ S1' S2' S3' S4' ]
An increment Δ C = f (S') -f (S), where f () is an objective function.
If deltaC<0, then S' is received as the new current solution; otherwise, a value in [0,1 ] is generated]And comparing the random number epsilon in the interval with the probability P, if epsilon is more than P, receiving a new solution, and otherwise, keeping S. WhereinK is a constant.
Sixthly, judging whether the iteration times are met or not, and turning to the third step if the iteration times are not met;
see if the target temperature is met, if not, T is gradually reduced, and T>T 1 And then turning to the third step.
The best parameters and solutions are obtained.
S133: performing feature extraction on the processed vibration signal x' (t) by using a singular value decomposition method;
taking a vibration signal as an example, the initial vector matrix E is formed by the first 5 eigenmode function components after CPEMD decomposition by using the optimal parameters e×g The rank of the matrix is r', singular value decomposition is carried out, E = UDW T Where U is unitary matrix of order e × e, D is diagonal matrix of order e × g, and W is a semi-positive fixed diagonal matrix of order e × g T I.e., the conjugate transpose of W, is a unitary matrix of order g × g., Δ r′×r′ =diag(σ 1 ,σ 2 ,…,σ r′ ),Is a matrix E T E characteristic value. To obtain E e×g Singular value of σ i (i =1,2, \8230;, r'), the resulting characteristic component σ i As a set of features for a neural network. I of n signal 2 A characteristic component ofTake the first feature set as an example. N =1, i 2 =1,2, \8230h6, i.e. the first piece of data of the feature set is: sigma 12 、σ 12 、...、σ 16 The nth data is σ n2 、σ n2 、...、σ n6 。
S134: and storing the features obtained by decomposition into a feature set.
In the step S14, a long-short term memory artificial neural network LSTM is constructed and trained and tested:
as shown in FIG. 2, the long-short term memory artificial neural network graph is obtained by dividing the obtained feature set T (n) into a training set T 1 And test set T 2 Wherein the training set T 1 Is given by the label y 1 N test sets and T test sets 2 Is given by the label y 2 . The basic steps for constructing the long-short term memory artificial neural network LSTM are as follows: with training set T 1 And label y 1 For input, training is performed, and a test set T is input 2 Making network test and outputting predictive label y 2 ', the correct number is set as N 0 Let N stand for 0 =0, if y' 2 (a)-y 2 (a) =0 (a =1,2, \ 8230;, N) then N 0 =N 0 +1, cycle from 1 to N, from formulaWhere s is the accuracy of the classification prediction. The derivation of each parameter and weight of the long-term and short-term memory artificial neural network is as follows:
let x be the input of the neural network, o be the output, and y' be the labels of the training set.
Training a neural network: the signal is propagated forward, the error is propagated backward, and the steps are as follows:
1) Updating the forget gate output: f (t) = σ (w) xf x t +w hf h t-1 +b f ) Wherein σ is sigmoid function;
2) Update input gate two part output: i (t) = σ (w) xi x t +w hi h t-1 +b i ),g t =tanh(w xg x t +w hg h t-1 +b g );
3) Renewal of the cell State c t =c t-1 ⊙f t +g t ⊙i t ;
4) Updating output gate output: o (t) = σ (w) xo x t +w ho h t-1 +b o );h t =o t ⊙tanh(c t ) Keeping short-term memory;
forward propagation: the forward propagation formula at time t is:
c t =c t-1 ⊙f t +g t ⊙i t
m t =tanh(c t )
h t =o t ⊙m t
y t =w yh h t +by
the back propagation formula: j is an error.
updating the node gradient:
updating the parameter gradient:
as shown in fig. 3, the long-short term memory artificial neural network model is optimized using a simulated annealing algorithm:
using the classification prediction accuracy obtained by LSTM as simulated annealingAn objective function f () of the algorithm, and then the batch size is set as an argument x 1 Setting the number of hidden layer nodes as the independent variable x of the objective function 2 Optimizer type z, s = [ x1 x2 z ]]The method comprises the following steps:
initializing: initial temperature T 0 (sufficiently high), initial solution S (start of iteration), number of iterations L, termination temperature T 1 。
And the sixth step is carried out by L.
And producing novel S'. S '= [ x1' x2'z' ].
An increment Δ C = f (S') -f (S), where f () is an objective function.
If the delta C is larger than 0, receiving S' as a new current solution; otherwise, a value of [0,1 ] is generated]And comparing the random number epsilon in the interval with the probability P, if epsilon is more than P, receiving a new solution, and otherwise, keeping S. WhereinK is a constant.
Sixthly, turning to the third step if the iteration times are met or not; see if the target temperature is met, if not, T is gradually reduced, and T>T 1 And then turning to the third step.
Resulting in optimal classification prediction accuracy S.
In the embodiment, the optimization algorithm is carried out on the system with more parameters and poor stability to search the parameters and adjust the parameters, so that the optimal parameters are found, and the optimal system and the prediction accuracy are obtained.
Example 2
As shown in fig. 4, the method for predicting the wear degree of the crankshaft bearing of the present embodiment includes the following steps:
s21: placing an industrial acceleration sensor at a specified position of a diesel engine, and acquiring accelerated vibration data of a diesel engine vibration signal x (t); in the present embodiment, the 4 th crankshaft bearing of the diesel engine is considered as a subject. A PCB601A01 type ICP industrial acceleration sensor is selected and placed at the left and right positions of a 3-cylinder joint of an oil bottom and a cylinder body, and is attached to the surface of a test part through a powerful magnetic seat. The bearing clearance is defined as 0.05-0.11mm as the normal working condition, 0.11-0.15mm as the light abrasion, 0.15-0.20mm as the moderate abrasion, and more than 0.20mm as the heavy abrasion. The vibration signals of the crankshaft bearing under normal working conditions, slight abrasion, moderate abrasion and severe abrasion at the rotating speed of 1800r/min are collected, the sampling frequency of the sensor is 12.8kHz, and 16384 points are sampled every time as one piece of data after the sensor runs stably.
S22: and inputting the diesel engine vibration signal x (t) into a crankshaft bearing wear degree prediction model, and outputting a prediction result of the crankshaft bearing wear degree.
The accuracy S of classification prediction of the abrasion degree of the crankshaft bearing of the diesel engine by adopting the abrasion degree prediction method of the crankshaft bearing reaches 96.875 percent, and is improved by more than 9 percent compared with the basic LSTM network prediction accuracy. The model can predict the wear degree of the crankshaft through the acquired vibration signals.
Example 3
As shown in fig. 5, the apparatus for establishing a prediction model of wear degree of a crankshaft bearing according to the present embodiment includes:
an acquiring unit 301, configured to acquire a diesel engine vibration signal x (t) acquired by an industrial acceleration sensor;
because there is noise interference outside, the processing unit 302 is configured to perform wavelet threshold denoising on the obtained vibration signal x (t), and select a soft threshold method with a threshold as an unbiased likelihood estimation threshold to perform processing, so as to obtain a processed vibration signal x' (t);
the extraction unit 303 is configured to perform decomposition by using a simulated annealing algorithm to optimize an improved ensemble empirical mode decomposition, and perform feature extraction on the processed vibration signal x' (t) by using a singular value decomposition method and store the extracted feature in a feature set;
and the generating unit 304 is used for taking the vibration signal characteristics with the characteristic concentration as the input of the long-short term memory artificial neural network model, optimizing the long-short term memory artificial neural network model by using a simulated annealing algorithm, and generating a crankshaft bearing wear degree judgment model.
The extraction unit 303 includes:
a first decomposition unit 3031 for improved collective empirical mode decomposition processing of the vibration signal x' (t);
a second decomposition unit 3032, configured to simulate an annealing algorithm to optimize the improved ensemble empirical mode decomposition on the processed vibration signal x' (t);
an extraction subunit 3033, configured to perform feature extraction on the processed vibration signal x' (t) by using a singular value decomposition method;
a storing unit 3034, configured to store the decomposed features into a feature set.
The device of the embodiment can solve the problems of more parameters and poor stability of improved ensemble empirical mode decomposition (CPEMD).
Example 4
As shown in fig. 6, the present embodiment provides a crankshaft bearing wear degree prediction device, including:
a vibration signal acquisition unit 401, configured to acquire a diesel engine vibration signal x (t) acquired by an industrial acceleration sensor;
and a prediction result output unit 402 for inputting the diesel engine vibration signal x (t) collected by the industrial acceleration sensor into the crankshaft bearing wear degree prediction model and outputting a crankshaft bearing wear degree prediction result.
The present embodiment prediction device suppresses the CEEMD pseudo component for CPEMD. The generation of the pseudo component is that positive white and negative white total noise are added in the signal, although decomposition averaging is carried out, the signal cannot be completely eliminated in calculation iteration, therefore, permutation entropy Hp is introduced to remove the residual white noise as an index so as to reduce the influence of the residual white noise, and SA-CPEMD improves the stability of CPEMD.
Embodiments 3 and 4 describe a device for establishing a crankshaft bearing wear degree prediction model and a crankshaft bearing wear degree prediction device provided in the embodiments of the present application from the perspective of functional modularization. Next, a terminal device for establishing a crankshaft bearing wear degree prediction model and a terminal device for predicting a crankshaft bearing wear degree provided in the embodiments of the present application will be described from the perspective of hardware processing.
Example 5
This embodiment provides a terminal device, terminal device includes: a processor and a memory; a memory for storing program code and transmitting the program code to the processor; and the processor is used for calling the instructions in the memory to execute the method for establishing the crankshaft bearing wear degree prediction model provided by the embodiment of the application.
Example 6
The present embodiment provides a terminal device, including: a processor and a memory; a memory for storing program code and transmitting the program code to the processor; and the processor is used for calling the instructions in the memory to execute the crankshaft bearing wear degree prediction method provided by the embodiment of the application.
It should be noted that, in the present specification, the embodiments are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments may be referred to each other. For the system or the device disclosed by the embodiment, the description is simple because the system or the device corresponds to the method disclosed by the embodiment, and the relevant points can be referred to the method part for description.
It is further noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrases "comprising one of 8230; \8230;" 8230; "does not exclude the presence of additional like elements in a process, method, article, or apparatus that comprises the element.
The steps of a method or algorithm described in connection with the embodiments disclosed herein may be embodied directly in hardware, in a software module executed by a processor, or in a combination of the two. A software module may reside in Random Access Memory (RAM), memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.
The above are merely examples of the present application and are not intended to limit the present application. Various modifications and changes may occur to those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of claims of the present application.
Claims (8)
1. A method of establishing a model for predicting the wear level of a crankshaft bearing, comprising the steps of:
s11: acquiring a diesel engine vibration signal x (t) acquired by an industrial acceleration sensor;
s12: carrying out wavelet threshold denoising on the obtained vibration signal x (t), and selecting a soft threshold method with a threshold as an unbiased likelihood estimation threshold to process to obtain a processed vibration signal x' (t);
s13: decomposing by using a set empirical mode decomposition optimized and improved by a simulated annealing algorithm, and performing feature extraction on the processed vibration signal x' (t) by using a singular value decomposition method and storing the feature extraction into a feature set;
s14: and (3) taking the feature set as the input of the long-short term memory artificial neural network model, and optimizing the long-short term memory artificial neural network model by using a simulated annealing algorithm to generate a crankshaft bearing wear degree prediction model.
2. The method for establishing a predictive model of the wear degree of a crankshaft bearing according to claim 1, wherein said step S13 comprises the steps of:
s131: processing the vibration signal x' (t) by improved set empirical mode decomposition;
s132: optimizing the improved set empirical mode decomposition pair processed vibration signal x' (t) by a simulated annealing algorithm;
s133: performing feature extraction on the processed vibration signal x' (t) by using a singular value decomposition method;
s134: and storing the features obtained by decomposition into a feature set.
3. A method for predicting the wear degree of a crankshaft bearing is characterized by comprising the following steps:
s21: acquiring a diesel engine vibration signal x (t) acquired by an industrial acceleration sensor;
s22: inputting the diesel engine vibration signal x (t) into a crankshaft bearing wear degree prediction model, and outputting a prediction result of the crankshaft bearing wear degree, wherein the crankshaft bearing wear degree prediction model is established according to the method for establishing the crankshaft bearing wear degree prediction model in any one of claims 1-2.
4. An apparatus for modeling wear of a crankshaft bearing, the apparatus comprising:
the acquisition unit (301) is used for acquiring a diesel engine vibration signal x (t) acquired by the industrial acceleration sensor;
a processing unit (302) for performing wavelet threshold denoising on the obtained vibration signal x (t), and selecting a soft threshold method with a threshold as an unbiased likelihood estimation threshold for processing to obtain a processed vibration signal x' (t);
the extraction unit (303) is used for decomposing the set empirical mode decomposition optimized and improved by using a simulated annealing algorithm, and performing characteristic extraction on the processed vibration signal x' (t) by using a singular value decomposition method and storing the characteristic extraction into a characteristic set;
and the generating unit (304) is used for taking the vibration signal characteristics in the characteristic set as the input of the long-short term memory artificial neural network model, optimizing the long-short term memory artificial neural network model by using a simulated annealing algorithm and generating a crankshaft bearing wear degree prediction model.
5. An arrangement for establishing a predictive model of the wear extent of a crankshaft bearing according to claim 4, characterized in that the extraction unit (303) comprises:
a first decomposition unit (3031) for improved collective empirical mode decomposition processing of the vibration signal x' (t);
a second decomposition unit (3032) for optimizing the improved ensemble empirical mode decomposition pair processed vibration signal x' (t) by the simulated annealing algorithm;
an extraction subunit (3033) for performing feature extraction on the processed vibration signal x' (t) by using a singular value decomposition method;
and a storage unit (3034) for storing the characteristics obtained by the decomposition into a characteristic set.
6. A crankshaft bearing wear degree prediction apparatus, characterized by comprising:
a vibration signal acquisition unit (401) for acquiring a diesel engine vibration signal x (t) acquired by an industrial acceleration sensor;
a prediction result output unit (402) for inputting the diesel engine vibration signal x (t) collected by the industrial acceleration sensor into a crankshaft bearing wear degree prediction model, which is established according to the method for establishing a crankshaft bearing wear degree prediction model of any one of claims 1 to 2, and outputting a crankshaft bearing wear degree prediction result.
7. A terminal device, the terminal device comprising:
a processor and a memory;
the memory is used for storing program codes and transmitting the program codes to the processor;
the processor is used for calling the instructions in the memory to execute the method for establishing the crankshaft bearing wear degree prediction model according to any one of claims 1-2.
8. A terminal device, the terminal device comprising:
a processor and a memory;
the memory is used for storing program codes and transmitting the program codes to the processor;
the processor is used for calling instructions in the memory to execute the crankshaft bearing wear degree prediction method in claim 3.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210903822.5A CN115326397B (en) | 2022-07-28 | 2022-07-28 | Method and related device for establishing crankshaft bearing wear degree prediction model and prediction method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210903822.5A CN115326397B (en) | 2022-07-28 | 2022-07-28 | Method and related device for establishing crankshaft bearing wear degree prediction model and prediction method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN115326397A true CN115326397A (en) | 2022-11-11 |
CN115326397B CN115326397B (en) | 2023-10-27 |
Family
ID=83920545
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210903822.5A Active CN115326397B (en) | 2022-07-28 | 2022-07-28 | Method and related device for establishing crankshaft bearing wear degree prediction model and prediction method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115326397B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115577643A (en) * | 2022-11-23 | 2023-01-06 | 广东电网有限责任公司中山供电局 | Temperature prediction method and device for cable terminal |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP3465127A1 (en) * | 2016-05-24 | 2019-04-10 | Nuovo Pignone Technologie S.r.l. | Method and system to monitor the health status of a rolling bearing of a machinery, and machinery equipped with said system |
CN110610035A (en) * | 2019-08-28 | 2019-12-24 | 武汉科技大学 | Rolling bearing residual life prediction method based on GRU neural network |
CN113435322A (en) * | 2021-06-25 | 2021-09-24 | 西安交通大学 | Main shaft bearing fault detection method, system, equipment and readable storage medium |
CN113465923A (en) * | 2021-06-08 | 2021-10-01 | 华能国际电力股份有限公司上海石洞口第一电厂 | Residual life prediction method for grinding roller bearing of coal mill |
US20210350050A1 (en) * | 2020-05-06 | 2021-11-11 | Wuhan University | Method and system for predicting gas content in transformer oil based on joint model |
CN114264478A (en) * | 2021-12-21 | 2022-04-01 | 北京石油化工学院 | Diesel engine crankshaft bearing wear degree prediction method and system |
US20220114421A1 (en) * | 2020-01-08 | 2022-04-14 | Chongqing Unversity Of Posts And Telecommunications | Method for estimating the capacity of lithium battery based on convolution long-short-term memory neural network |
-
2022
- 2022-07-28 CN CN202210903822.5A patent/CN115326397B/en active Active
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP3465127A1 (en) * | 2016-05-24 | 2019-04-10 | Nuovo Pignone Technologie S.r.l. | Method and system to monitor the health status of a rolling bearing of a machinery, and machinery equipped with said system |
CN110610035A (en) * | 2019-08-28 | 2019-12-24 | 武汉科技大学 | Rolling bearing residual life prediction method based on GRU neural network |
US20220114421A1 (en) * | 2020-01-08 | 2022-04-14 | Chongqing Unversity Of Posts And Telecommunications | Method for estimating the capacity of lithium battery based on convolution long-short-term memory neural network |
US20210350050A1 (en) * | 2020-05-06 | 2021-11-11 | Wuhan University | Method and system for predicting gas content in transformer oil based on joint model |
CN113465923A (en) * | 2021-06-08 | 2021-10-01 | 华能国际电力股份有限公司上海石洞口第一电厂 | Residual life prediction method for grinding roller bearing of coal mill |
CN113435322A (en) * | 2021-06-25 | 2021-09-24 | 西安交通大学 | Main shaft bearing fault detection method, system, equipment and readable storage medium |
CN114264478A (en) * | 2021-12-21 | 2022-04-01 | 北京石油化工学院 | Diesel engine crankshaft bearing wear degree prediction method and system |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115577643A (en) * | 2022-11-23 | 2023-01-06 | 广东电网有限责任公司中山供电局 | Temperature prediction method and device for cable terminal |
Also Published As
Publication number | Publication date |
---|---|
CN115326397B (en) | 2023-10-27 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN103559888B (en) | Based on non-negative low-rank and the sound enhancement method of sparse matrix decomposition principle | |
CN110309603B (en) | Short-term wind speed prediction method and system based on wind speed characteristics | |
CN110212528B (en) | Power distribution network measurement data missing reconstruction method | |
US20210043215A1 (en) | System and method for efficient processing of universal background models for speaker recognition | |
CN112364291B (en) | Empirical mode decomposition method and device for pre-filtering extreme point optimization set | |
CN112434891A (en) | Method for predicting solar irradiance time sequence based on WCNN-ALSTM | |
CN115326397A (en) | Method for establishing crankshaft bearing wear degree prediction model and prediction method and related device | |
CN114169110A (en) | Motor bearing fault diagnosis method based on feature optimization and GWAA-XGboost | |
CN115587666A (en) | Load prediction method and system based on seasonal trend decomposition and hybrid neural network | |
CN114298134A (en) | Wind power prediction method and device and electronic equipment | |
CN113935489A (en) | Variational quantum model TFQ-VQA based on quantum neural network and two-stage optimization method thereof | |
CN114264478A (en) | Diesel engine crankshaft bearing wear degree prediction method and system | |
Regazzoni et al. | A physics-informed multi-fidelity approach for the estimation of differential equations parameters in low-data or large-noise regimes | |
CN114166509A (en) | Motor bearing fault prediction method | |
CN111881620B (en) | User software behavior simulation system based on reinforcement learning algorithm and GAN model and working method thereof | |
CN117056669A (en) | Denoising method and system for vibration signal | |
CN115829157A (en) | Chemical water quality index prediction method based on variational modal decomposition and auto former model | |
CN114720129A (en) | Rolling bearing residual life prediction method and system based on bidirectional GRU | |
Abbas | System identification using optimally designed functional link networks via a fast orthogonal search technique. | |
CN115859048A (en) | Noise processing method and device for partial discharge signal | |
Abbas et al. | Volterra system identification using adaptive genetic algorithms | |
KR20220111897A (en) | Power prediction device with mode decomposition and neural network | |
Lim et al. | Analyzing deep neural networks with noisy labels | |
CN111126659A (en) | Power load prediction method and system | |
CN117200208B (en) | User-level short-term load prediction method and system based on multi-scale component feature learning |
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 |