1. An early failure diagnosis method, comprising:

preprocessing a sampling signal of a part to be detected, and inputting the preprocessed sampling signal into a stochastic resonance model to obtain a target signal;

segmenting the target signal to generate a training set and a test set, wherein the training set comprises a plurality of first sub-target signals carrying real labels, and the test set comprises a plurality of second sub-target signals;

establishing an initial feature extraction model, and training the initial feature extraction model through the training set to obtain a target feature extraction model;

performing feature extraction on a plurality of second sub-target signals in the test set through the target feature extraction model to obtain a plurality of second feature vectors corresponding to the plurality of second sub-target signals one to one;

and determining a target least square support vector machine fault diagnosis model, analyzing the plurality of second characteristic vectors through the target least square support vector machine fault diagnosis model, and determining whether the part to be detected has an early fault.

2. The early fault diagnosis method according to claim 1, wherein the preprocessing the sampled signal comprises:

envelope demodulation is carried out on the sampling signal to obtain multi-scale noise; and performing orthogonal discrete wavelet transform on the multi-scale noise based on the maximum output signal-to-noise ratio principle to generate target type noise with 1/f type noise distribution.

3. The early fault diagnosis method according to claim 2, characterized in that the orthogonal discrete wavelet transform:

in the formula, W_f(j, k) is the target type noise; f (t) is a continuous power value of the multi-scale noise; t is the sampling time;for passing the wavelet function psi (t) through 2^jInteger multiple compression and channel binary 2^-jk is translated into a function; j is a scaling factor; k is the translation factor.

4. The early fault diagnosis method according to claim 2, wherein the initial feature extraction model comprises a convolutional layer, a pooling layer, a fully-connected layer, and a softmax classifier; the training the initial feature extraction model through the training set to obtain a target feature extraction model comprises:

inputting the plurality of first sub-target signals in the training set into the initial feature extraction model, and extracting a plurality of first feature vectors corresponding to the plurality of first sub-target signals one by one through the convolutional layer, the pooling layer and the full-link layer;

classifying the plurality of first feature vectors through the softmax classifier, and generating a plurality of prediction labels in one-to-one correspondence with the plurality of first feature vectors;

determining a classification accuracy of the initial feature extraction model from the plurality of predictive labels and the plurality of real labels;

and determining whether the initial feature extraction model is trained or not according to the classification accuracy, and if the initial feature extraction model is trained, obtaining the target feature extraction model.

5. The early fault diagnosis method according to claim 4, wherein the initial feature extraction model includes model parameters; the step of determining whether the initial feature extraction model is trained according to the classification accuracy comprises the following steps:

judging whether the classification accuracy is smaller than a threshold classification accuracy;

and if the classification accuracy is smaller than the threshold classification accuracy, adjusting the model parameters, and training the initial feature extraction model after the model parameters are adjusted again until the training times are larger than the threshold training times or the classification accuracy of the initial feature extraction model after the training is larger than or equal to the threshold classification accuracy, and finishing the training of the initial feature extraction model.

6. The early fault diagnosis method of claim 1, wherein the determining a target least squares support vector machine fault diagnosis model comprises:

establishing an initial least square support vector machine fault diagnosis model, wherein the initial least square support vector machine fault diagnosis model comprises a kernel function, initial kernel function parameters and initial penalty factors;

and optimizing the initial kernel function parameters and the initial penalty factors through the particle swarm optimization to obtain target kernel function parameters and target penalty factors so as to determine the target least square support vector machine fault diagnosis model.

7. The early fault diagnosis method according to claim 6, wherein the kernel function is a gaussian-iran radial basis function, the gaussian-iran radial basis function being:

exp(-γ|u-v|²)

wherein γ is a free parameter; u and v are both the first feature vector; | u-v | is the distance between two first feature vectors.

8. The early fault diagnosis method according to claim 7, wherein the optimizing the initial kernel function parameter and the initial penalty factor by the particle swarm optimization, and obtaining the target kernel function parameter and the target penalty factor comprises:

initializing initial positions, speeds and iteration times of each particle vector in the particle swarm;

designing a fitness function;

training the initial least square support vector machine fault diagnosis model by using the training set, obtaining a predictive diagnosis label corresponding to a first characteristic vector, and calculating the fitness value and the global optimal fitness value of each particle according to the fitness function;

updating the speed and position of each particle according to the fitness value of each particle, the global optimal fitness value and the historical optimal fitness value;

adjusting the initial kernel function parameters and the initial penalty factors according to the updated speed and position of each particle;

judging whether the times of adjusting the initial kernel function parameters and the initial penalty factors reach the maximum iteration times or not, if the iteration times do not reach the maximum iteration times, adjusting the initial kernel function parameters and the initial penalty factors again, and if the iteration times reach the maximum iteration times, the current kernel function parameters and the penalty factors are the target kernel function parameters and the target penalty factors respectively.

9. The early fault diagnosis method of claim 8, wherein the fitness function is a mean-square error function, the mean-square error function being:

wherein n is the number of the first feature vectors; y is_iIs the real label;is the predictive diagnostic label.

10. An early failure diagnosis apparatus characterized by comprising:

one or more processors;

a memory; and

one or more application programs, wherein the one or more application programs are stored in the memory and configured to be executed by the processor to implement the early failure diagnosis method of any of claims 1-9.

Background

With the development of the times, the automation degree in the production activities is continuously improved, and mechanical equipment is also developed towards the direction of integration. In engineering applications, once a component in a mechanical device fails in operation, it may have a great impact on the entire production system. Therefore, before the mechanical equipment is seriously worn, namely in the early stage of the occurrence of the fault, the fault of the equipment is accurately and timely identified, and the method has great significance for avoiding the occurrence of serious catastrophic accidents.

In practical application, the traditional methods such as spectrum analysis, envelope analysis, wavelet analysis, morphological filtering and the like cannot completely extract fault characteristics due to the interference of multiple fault sources and background noise, so that the conditions of misjudgment, missed judgment and the like are caused, and the operation evaluation of mechanical equipment is incomplete. Meanwhile, the early fault diagnosis algorithm mostly adopts a neural network for diagnosis, the neural network has the problems of low convergence speed, easy falling into local minimum and the like although having better nonlinear approximation capability, and the fault diagnosis of the neural network is not good in multi-classification effect, so that the diagnosis result is unreliable.

Disclosure of Invention

The invention provides an early fault diagnosis method and device, and aims to solve the technical problems that in the prior art, fault features are not extracted thoroughly, operation evaluation on mechanical equipment is incomplete, the multi-classification effect is poor, and the diagnosis result is unreliable.

In one aspect, the present invention provides an early fault diagnosis method, including:

preprocessing a sampling signal of a part to be detected, and inputting the preprocessed sampling signal into a stochastic resonance model to obtain a target signal;

segmenting the target signal to generate a training set and a test set, wherein the training set comprises a plurality of first sub-target signals carrying real labels, and the test set comprises a plurality of second sub-target signals;

establishing an initial feature extraction model, and training the initial feature extraction model through the training set to obtain a target feature extraction model;

performing feature extraction on a plurality of second sub-target signals in the test set through the target feature extraction model to obtain a plurality of second feature vectors corresponding to the plurality of second sub-target signals one to one;

and determining a target least square support vector machine fault diagnosis model, analyzing the plurality of second characteristic vectors through the target least square support vector machine fault diagnosis model, and determining whether the part to be detected has an early fault.

In a possible implementation manner of the present invention, the preprocessing the sampling signal includes:

envelope demodulation is carried out on the sampling signal to obtain multi-scale noise; and performing orthogonal discrete wavelet transform on the multi-scale noise based on the maximum output signal-to-noise ratio principle to generate target type noise with 1/f type noise distribution.

In one possible implementation of the invention, the orthogonal discrete wavelet transform is:

in the formula, W_f(j, k) is the target type noise; f (t) is a continuous power value of the multi-scale noise; t is the sampling time;for passing the wavelet function psi (t) through 2^jInteger multiple compression and channel binary 2^-jk is translated into a function; j is a scaling factor; k is the translation factor.

In a possible implementation manner of the present invention, the initial feature extraction model includes a convolutional layer, a pooling layer, a full link layer, and a softmax classifier; the training the initial feature extraction model through the training set to obtain a target feature extraction model comprises:

inputting the plurality of first sub-target signals in the training set into the initial feature extraction model, and extracting a plurality of first feature vectors corresponding to the plurality of first sub-target signals one by one through the convolutional layer, the pooling layer and the full-link layer;

classifying the plurality of first feature vectors through the softmax classifier, and generating a plurality of prediction labels in one-to-one correspondence with the plurality of first feature vectors;

determining a classification accuracy of the initial feature extraction model from the plurality of predictive labels and the plurality of real labels;

and determining whether the initial feature extraction model is trained or not according to the classification accuracy, and if the initial feature extraction model is trained, obtaining the target feature extraction model.

In one possible implementation manner of the present invention, the initial feature extraction model includes model parameters; the step of determining whether the initial feature extraction model is trained according to the classification accuracy comprises the following steps:

judging whether the classification accuracy is smaller than a threshold classification accuracy;

and if the classification accuracy is smaller than the threshold classification accuracy, adjusting the model parameters, and training the initial feature extraction model after the model parameters are adjusted again until the training times are larger than the threshold training times or the classification accuracy of the initial feature extraction model after the training is larger than or equal to the threshold classification accuracy, and finishing the training of the initial feature extraction model.

In a possible implementation manner of the present invention, the determining a target least squares support vector machine fault diagnosis model includes:

establishing an initial least square support vector machine fault diagnosis model, wherein the initial least square support vector machine fault diagnosis model comprises a kernel function, initial kernel function parameters and initial penalty factors;

and optimizing the initial kernel function parameters and the initial penalty factors through the particle swarm optimization to obtain target kernel function parameters and target penalty factors so as to determine the target least square support vector machine fault diagnosis model.

In a possible implementation manner of the present invention, the kernel function is a gaussian-iran radial basis function, and the gaussian-iran radial basis function is:

exp(-γ|u-v|²)

wherein γ is a free parameter; u and v are both the first feature vector; | u-v | is the distance between two first feature vectors.

In a possible implementation manner of the present invention, the optimizing the initial kernel function parameter and the initial penalty factor by the particle swarm algorithm, and obtaining the target kernel function parameter and the target penalty factor includes:

initializing initial positions, speeds and iteration times of each particle vector in the particle swarm;

designing a fitness function;

training the initial least square support vector machine fault diagnosis model by using the training set, obtaining a predictive diagnosis label corresponding to a first characteristic vector, and calculating the fitness value and the global optimal fitness value of each particle according to the fitness function;

updating the speed and position of each particle according to the fitness value of each particle, the global optimal fitness value and the historical optimal fitness value;

adjusting the initial kernel function parameters and the initial penalty factors according to the updated speed and position of each particle;

judging whether the times of adjusting the initial kernel function parameters and the initial penalty factors reach the maximum iteration times or not, if the iteration times do not reach the maximum iteration times, adjusting the initial kernel function parameters and the initial penalty factors again, and if the iteration times reach the maximum iteration times, the current kernel function parameters and the penalty factors are the target kernel function parameters and the target penalty factors respectively.

In a possible implementation manner of the present invention, the fitness function is a mean square error function, and the mean square error function is:

wherein n is the number of the first feature vectors; y is_iIs the real label;is the predictive diagnostic label.

In another aspect, the present invention provides an early failure diagnosis apparatus including:

one or more processors;

a memory; and

one or more applications, wherein the one or more applications are stored in the memory and configured to be executed by the processor to implement any of the early failure diagnosis methods described above.

In another aspect, the present invention further provides a computer-readable storage medium, on which a computer program is stored, the computer program being loaded by a processor to execute the steps in the early failure diagnosis method described in any one of the above.

According to the method, firstly, a sampling signal of a part to be detected is preprocessed, then the preprocessed sampling signal is input to a stochastic resonance model, a signal frequency spectrum is obtained, and the energy of noise in the sampling signal is transferred to a fault characteristic frequency component through a stochastic resonance effect, so that the weak signal is strengthened while the noise is weakened, the early weak fault characteristic enhancement is realized, the incomplete signal extraction is avoided, and the reliability of fault diagnosis is improved; furthermore, the target least square support vector machine fault diagnosis model is determined, the second feature vectors are analyzed through the target least square support vector machine fault diagnosis model, whether the part to be detected has an early fault or not is determined, and compared with the method for diagnosing the fault through a neural network, the method and the device can identify various complex fault types, further improve the reliability of early fault diagnosis, and reduce the risk and economic loss of mechanical equipment operation.

Drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed to be used in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.

FIG. 1 is a flow chart illustrating an early fault diagnosis method according to an embodiment of the present invention;

FIG. 2 is a flow diagram illustrating one embodiment of generating a target type of noise provided by embodiments of the present invention;

FIG. 3 is a schematic structural diagram of an initial feature extraction model according to an embodiment of the present invention;

fig. 4 is a schematic flowchart of an embodiment of S103 according to the present invention;

fig. 5 is a flowchart illustrating an embodiment of S404 according to the present invention;

fig. 6 is a schematic flowchart of an embodiment of S105 according to the present invention;

fig. 7 is a flowchart illustrating an embodiment of S602 according to an embodiment of the present invention;

fig. 8 is a schematic structural diagram of an early failure diagnosis apparatus according to an embodiment of the present invention.

Detailed Description

The technical solution in 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. It is to be understood that the described embodiments are merely exemplary of the invention, and not restrictive of the full scope of the invention. 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.

In the description of the present invention, "a plurality" means two or more unless specifically defined otherwise.

Reference herein to "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment can be included in at least one embodiment of the invention. The appearances of the phrase in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments. It is explicitly and implicitly understood by one skilled in the art that the embodiments described herein can be combined with other embodiments.

The present invention provides a method and an apparatus for early fault diagnosis, which will be described in detail below.

As shown in fig. 1, a schematic flow chart of an embodiment of an early fault diagnosis method provided by an embodiment of the present invention includes:

s101, preprocessing a sampling signal of a part to be detected, and inputting the preprocessed sampling signal into a stochastic resonance model to obtain a target signal;

the sampling signal may be an acceleration signal or a vibration signal.

S102, segmenting the target signal to generate a training set and a test set, wherein the training set comprises a plurality of first sub-target signals carrying real labels, and the test set comprises a plurality of second sub-target signals;

s103, establishing an initial feature extraction model, and training the initial feature extraction model through a training set to obtain a target feature extraction model;

s104, performing feature extraction on a plurality of second sub-target signals in the test set through a target feature extraction model to obtain a plurality of second feature vectors corresponding to the second sub-target signals one by one;

and S105, determining a target least square support vector machine fault diagnosis model, analyzing the plurality of second characteristic vectors through the target least square support vector machine fault diagnosis model, and determining whether the part to be detected has an early fault.

According to the early fault diagnosis method provided by the embodiment of the invention, firstly, a sampling signal of a part to be detected is preprocessed, then, the preprocessed sampling signal is input to a stochastic resonance model to obtain a signal frequency spectrum, and the energy of noise of the sampling signal is transferred to a fault characteristic frequency component through a stochastic resonance effect, so that the weak signal is strengthened while the noise is weakened, the early weak fault characteristic enhancement is realized, the incomplete signal extraction is avoided, and the reliability of fault diagnosis is improved; furthermore, the target least square support vector machine fault diagnosis model is determined, the second feature vectors are analyzed through the target least square support vector machine fault diagnosis model, whether the part to be detected has an early fault or not is determined, and compared with the method for diagnosing the fault through a neural network, the method and the device can identify various complex fault types, further improve the reliability of early fault diagnosis, and reduce the risk and economic loss of mechanical equipment operation.

The preprocessing of the sampling signal in S101 specifically includes: envelope demodulation is carried out on the sampling signal to obtain multi-scale noise; and performing orthogonal discrete wavelet transform on the multi-scale noise based on the maximum output signal-to-noise ratio principle to generate target type noise with 1/f type noise distribution.

The 1/f type noise is more likely to induce stochastic resonance effects than other noises, and thus the fault characteristic frequency can be further highlighted.

As shown in fig. 2, it can be seen that: the sampling signal is subjected to envelope demodulation to obtain multi-scale noise, then the preprocessed generated target type noise is input to the stochastic resonance model to obtain a signal frequency spectrum, and compared with the sampling signal, the signal frequency spectrum has obvious characteristics, so that the stochastic noise and pulse interference are effectively filtered, the early weak fault characteristics under strong background noise are highlighted, and the purpose of improving the accuracy of early fault diagnosis is achieved.

Specifically, the orthogonal discrete wavelet transform:

in the formula, W_f(j, k) is a target type noise; f (t) is the continuous power value of the multi-scale noise; t is the sampling time;for passing the wavelet function psi (t) through 2^jIntegral multiple voltageContraction of meridian binary number 2^-jk is translated into a function; j is a scaling factor; k is the translation factor.

Further, the output signal-to-noise ratio SNR is:

in the formula, P_signalIs the effective power of the sampled signal; p_noiseIs the effective power of the noise signal; n is the number of discrete samples of the multi-scale noise; x (n) is the discrete power value of the multi-scale noise.

Further, in some embodiments of the present invention, as shown in FIG. 3, the initial feature extraction model includes two convolutional layers, one pooling layer, one fully-connected layer, and a softmax classifier; as shown in fig. 4, S103 includes:

s401, inputting a plurality of first sub-target signals in a training set into an initial feature extraction model, and extracting a plurality of first feature vectors corresponding to the first sub-target signals one by one through a convolution layer, a pooling layer and a full-connection layer;

s402, classifying the plurality of first feature vectors through a softmax classifier, and generating a plurality of prediction labels in one-to-one correspondence with the plurality of first feature vectors;

s403, determining the classification accuracy of the initial feature extraction model according to the plurality of predicted labels and the plurality of real labels;

s404, determining whether the initial feature extraction model is trained or not according to the classification accuracy, and if the initial feature extraction model is trained, obtaining a target feature extraction model.

Further, the initial feature extraction model comprises model parameters; as shown in fig. 5, in some embodiments of the invention, S404 comprises:

s501, judging whether the classification accuracy is smaller than a threshold classification accuracy;

s502, if the classification accuracy is smaller than the threshold classification accuracy, adjusting model parameters, and training the initial feature extraction model after the model parameters are adjusted again until the training times are larger than the threshold training times or the classification accuracy of the initial feature extraction model after the training is performed again is larger than or equal to the threshold classification accuracy, and finishing the training of the initial feature extraction model.

Specifically, the model parameters may be adjusted using a small batch gradient descent method when adjusting the model parameters.

It should be noted that: the Loss function Loss of the initial feature extraction model is:

wherein M is the number of classes; p is a radical of_cTo predict the probability of correctness; y is_c0 or 1, 1 when the prediction is correct, and 0 when the prediction is wrong.

Specifically, in some embodiments of the present invention, M is 2 and the categories are "fault" or "healthy", respectively. Of course, in some other embodiments of the present invention, M may be adjusted according to actual conditions.

Further, in some embodiments of the present invention, as shown in fig. 6, S105 includes:

s601, establishing an initial least square support vector machine (LS-SVM) fault diagnosis model, wherein the initial least square support vector machine fault diagnosis model comprises a kernel function, initial kernel function parameters and initial penalty factors;

s602, optimizing the initial kernel function parameters and the initial penalty factors through a Particle Swarm Optimization (PSO), and obtaining target kernel function parameters and target penalty factors to determine a target least square support vector machine fault diagnosis model.

The initial kernel function parameters and the initial penalty factors are optimized by using the PSO, the complex parameter adjustment links can be reduced by obtaining the target kernel function parameters and the target penalty factors, the optimization speed is increased, and the fault diagnosis precision of the target least square support vector machine fault diagnosis model is improved.

Specifically, in some embodiments of the present invention, the kernel function is a gaussian-iran Radial Basis Function (RBF) that is:

exp(-γ|u-v|²)

wherein γ is a free parameter; u and v are both first feature vectors; | u-v | is the distance between two first feature vectors.

Further, in some embodiments of the present invention, as shown in fig. 7, S602 includes:

s701, initializing the initial position, the speed and the iteration times of each particle vector in the particle swarm;

s702, designing a fitness function;

s703, training an initial least square support vector machine fault diagnosis model by using a training set, obtaining a predictive diagnosis label corresponding to the first characteristic vector, and calculating a fitness value and a global optimal fitness value of each particle according to a fitness function;

s704, updating the speed and the position of each particle according to the fitness value of each particle, the global optimal fitness value and the historical optimal fitness value;

s705, adjusting initial kernel function parameters and initial penalty factors according to the updated speed and position of each particle;

s706, judging whether the times of adjusting the initial kernel function parameters and the initial penalty factors reach the maximum iteration times, if the iteration times do not reach the maximum iteration times, adjusting the initial kernel function parameters and the initial penalty factors again, and if the iteration times reach the maximum iteration times, the current kernel function parameters and the penalty factors are respectively target kernel function parameters and target penalty factors.

Specifically, the fitness function is a mean square error function, which is:

wherein n is the number of the first eigenvector；y_iIs a real label;to predict diagnostic markers.

The smaller the mean square error is, the better the fitness is, and the better the prediction performance is.

An embodiment of the present invention further provides an early fault diagnosis apparatus, where the early fault diagnosis apparatus includes:

one or more processors;

a memory; and

one or more application programs, wherein the one or more application programs are stored in the memory and configured to be executed by the processor to perform the steps of the early fault diagnosis method described in any of the early fault diagnosis method embodiments above.

As shown in fig. 8, a schematic structural diagram of an early failure diagnosis device according to an embodiment of the present invention is shown, specifically:

the early failure diagnosis apparatus may include components such as a processor 801 of one or more processing cores, a memory 802 of one or more computer-readable storage media, a power supply 803, and an input unit 804. Those skilled in the art will appreciate that the early failure diagnostic device configuration shown in fig. 8 does not constitute a limitation of the early failure diagnostic device and may include more or fewer components than shown, or some components in combination, or a different arrangement of components. Wherein:

the processor 801 is a control center of the early failure diagnosis apparatus, connects respective parts of the entire early failure diagnosis apparatus using various interfaces and lines, and performs various functions of the early failure diagnosis apparatus and processes data by running or executing software programs and/or modules stored in the memory 802 and calling data stored in the memory 802, thereby performing overall monitoring of the early failure diagnosis apparatus. Alternatively, processor 801 may include one or more processing cores; preferably, the processor 801 may integrate an application processor, which mainly handles operating systems, operating user interfaces, application programs, etc., and a modem processor, which mainly handles wireless communications. It will be appreciated that the modem processor described above may not be integrated into the processor 801.

The memory 802 may be used to store software programs and modules, and the processor 801 executes various functional applications and data processing by operating the software programs and modules stored in the memory 802. The memory 802 may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application program required by at least one function (such as a sound playing function, an image playing function, etc.), and the like; the storage data area may store data created from use of the early failure diagnosis apparatus, and the like. Further, the memory 802 may include high speed random access memory and may also include non-volatile memory, such as at least one magnetic disk storage device, flash memory device, or other volatile solid state storage device. Accordingly, the memory 802 may also include a memory controller to provide the processor 801 access to the memory 802.

The early failure diagnosis apparatus further includes a power supply 803 for supplying power to each component, and preferably, the power supply 803 may be logically connected to the processor 801 through a power management system, so as to implement functions of managing charging, discharging, and power consumption through the power management system. The power supply 803 may also include one or more dc or ac power sources, recharging systems, power failure detection circuitry, power converters or inverters, power status indicators, and any like components.

The early failure diagnosis apparatus may further include an input unit 804, and the input unit 804 may be used to receive input numeric or character information and generate a keyboard, mouse, joystick, optical or trackball signal input in connection with operating user settings and function control.

Although not shown, the early failure diagnosis apparatus may further include a display unit and the like, which will not be described in detail herein. Specifically, in this embodiment, the processor 801 in the early failure diagnosis apparatus loads an executable file corresponding to one or more processes of an application program into the memory 802 according to the following instructions, and the processor 801 runs the application program stored in the memory 802, thereby implementing various functions as follows:

acquiring a sampling signal of a part to be detected, and carrying out envelope demodulation on the sampling signal to obtain multi-scale noise;

preprocessing the multi-scale noise to generate target type noise;

inputting the target type noise into a stochastic resonance model to obtain a target signal;

segmenting the target signal to generate a training set and a test set, wherein the training set comprises a plurality of first sub-target signals carrying real labels, and the test set comprises a plurality of second sub-target signals;

establishing an initial feature extraction model, and training the initial feature extraction model through the training set to obtain a target feature extraction model;

performing feature extraction on a plurality of second sub-target signals in the test set through the target feature extraction model to obtain a plurality of second feature vectors corresponding to the plurality of second sub-target signals one to one;

and determining a target least square support vector machine fault diagnosis model, analyzing the plurality of second characteristic vectors through the target least square support vector machine fault diagnosis model, and determining whether the part to be detected has an early fault.

It will be understood by those skilled in the art that all or part of the steps of the methods of the above embodiments may be performed by instructions or by associated hardware controlled by the instructions, which may be stored in a computer readable storage medium and loaded and executed by a processor.

To this end, an embodiment of the present invention provides a computer-readable storage medium, which may include: read Only Memory (ROM), Random Access Memory (RAM), magnetic or optical disks, and the like. The pseudo random M array construction method comprises a step of constructing a pseudo random M array, and a step of constructing the pseudo random M array. For example, the computer program may be loaded by a processor to perform the steps of:

acquiring a sampling signal of a part to be detected, and carrying out envelope demodulation on the sampling signal to obtain multi-scale noise;

preprocessing the multi-scale noise to generate target type noise;

inputting the target type noise into a stochastic resonance model to obtain a target signal;

segmenting the target signal to generate a training set and a test set, wherein the training set comprises a plurality of first sub-target signals carrying real labels, and the test set comprises a plurality of second sub-target signals;

establishing an initial feature extraction model, and training the initial feature extraction model through the training set to obtain a target feature extraction model;

performing feature extraction on a plurality of second sub-target signals in the test set through the target feature extraction model to obtain a plurality of second feature vectors corresponding to the plurality of second sub-target signals one to one;

and determining a target least square support vector machine fault diagnosis model, analyzing the plurality of second characteristic vectors through the target least square support vector machine fault diagnosis model, and determining whether the part to be detected has an early fault.

In a specific implementation, each unit or structure may be implemented as an independent entity, or may be combined arbitrarily to be implemented as one or several entities, and the specific implementation of each unit or structure may refer to the foregoing method embodiment, which is not described herein again.

The method and the device for early fault diagnosis provided by the invention are described in detail above, a specific example is applied in the text to explain the principle and the implementation mode of the invention, and the description of the above embodiment is only used to help understand the method and the core idea of the invention; meanwhile, for those skilled in the art, according to the idea of the present invention, there may be variations in the specific embodiments and the application scope, and in summary, the content of the present specification should not be construed as a limitation to the present invention.