Noise monitoring method and device, nonvolatile storage medium and computer equipment

Abstract

The invention discloses a noise monitoring method and device, a nonvolatile storage medium and computer equipment. The method comprises the steps that historical working data of target equipment are acquired, and the historical working data comprise vibration signals and sound signals; determining feature data and a data deviation value according to the historical working data; performing feature selection based on the feature data and the data deviation value to obtain a plurality of feature sets; training a preset original convolutional neural network model by using the plurality of feature sets to obtain a target monitoring model, the target monitoring model being used for monitoring a deviation value of the working data; acquiring to-be-monitored working data of target equipment and inputting the to-be-monitored working data into the target monitoring model to obtain a deviation value; and determining a noise monitoring result based on a magnitude relationship between the deviation value and a preset threshold value. According to the invention, the technical problems of low monitoring efficiency and high false alarm rate caused by the fact that current noise monitoring methods mostly depend on manual inspection and simple sensor data analysis are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of noise monitoring, and in particular to a noise monitoring method, device, non-volatile storage medium and computer equipment. Background Art

[0002] The operation of some equipment is often accompanied by vibration and noise. Long-term or abnormal vibration and noise not only affect the performance and life of the equipment, but may also have an adverse impact on the production environment and personnel health. Traditional vibration and noise monitoring methods mostly rely on manual inspections and simple sensor data analysis, which have problems such as low monitoring efficiency, slow response speed, and high false alarm rate.

[0003] To address the above-mentioned problems, no effective solution has been proposed yet. Summary of the invention

[0004] The embodiments of the present invention provide a noise monitoring method, apparatus, non-volatile storage medium and computer equipment to at least solve the technical problem that current noise monitoring methods mostly rely on manual inspections and simple sensor data analysis, resulting in low monitoring efficiency and high false alarm rate.

[0005] According to one aspect of an embodiment of the present invention, a noise monitoring method is provided, including: obtaining historical working data of a target device, wherein the historical working data includes a vibration signal and a sound signal; determining feature data and a data deviation value based on the historical working data; performing feature selection based on the feature data and the data deviation value to obtain a plurality of feature sets, wherein the plurality of feature sets include a vibration signal feature set and a sound signal feature set; using the plurality of feature sets to train a preset original convolutional neural network model to obtain a target monitoring model, wherein the target monitoring model is used to monitor the deviation value of the working data; collecting the working data to be monitored of the target device and inputting it into the target monitoring model to obtain a deviation value; and determining the noise monitoring result based on the relationship between the deviation value and a preset threshold value.

[0006] Optionally, determining characteristic data based on historical working data includes: decomposing the historical working data to obtain multiple components; performing time-frequency analysis on the multiple components to obtain characteristic data, wherein the characteristic data includes instantaneous frequency and instantaneous amplitude.

[0007] Optionally, determining a data deviation value based on historical working data includes: determining a working data sequence based on the working data and a preset time sequence; sampling the working data sequence to obtain the sampled data and determining future actual values ​​corresponding to the sampled data; making predictions based on the sampled data to obtain predicted values ​​corresponding to the sampled data; determining a deviation value corresponding to the sampled data based on future actual values ​​and predicted values ​​corresponding to the sampled data; and determining a data deviation value based on the deviation value corresponding to the sampled data.

[0008] Optionally, feature selection is performed based on feature data and data deviation values ​​to obtain multiple feature sets, including: combining feature data and data deviation values ​​into a data set according to a preset time sequence; converting the data set from a high-dimensional feature space to a low-dimensional feature space to obtain a reduced-dimensional data set; and classifying and combining the reduced-dimensional data set based on preset feature types to obtain multiple feature sets.

[0009] Optionally, the noise monitoring result is determined based on the size relationship between the deviation value and a preset threshold value. When the deviation value includes a vibration signal deviation value and a sound signal deviation value, the method includes: determining a first warning value based on the vibration signal deviation value and the preset threshold value; determining a second warning value based on the sound signal deviation value and the preset threshold value; determining a target warning value based on the first warning value and the second warning value; and determining that the noise monitoring result is abnormal when the target warning value exceeds the preset warning value.

[0010] Optionally, the original convolutional neural network model is a combination model of a convolutional neural network and a long short-term memory network.

[0011] According to another aspect of an embodiment of the present invention, a noise monitoring device is also provided, including: an acquisition module, used to acquire historical working data of a target device, wherein the historical working data includes vibration signals and sound signals; a first determination module, used to determine feature data and data deviation values ​​based on the historical working data; a selection module, used to perform feature selection based on the feature data and the data deviation values ​​to obtain multiple feature sets, wherein the multiple feature sets include vibration signal feature sets and sound signal feature sets; a training module, used to train a preset original convolutional neural network model using multiple feature sets to obtain a target monitoring model, wherein the target monitoring model is used to monitor the deviation value of the working data; an acquisition module, used to acquire the working data to be monitored of the target device and input it into the target monitoring model to obtain a deviation value; a monitoring module, used to determine the noise monitoring result based on the size relationship between the deviation value and a preset threshold value.

[0012] According to another aspect of an embodiment of the present invention, a non-volatile storage medium is provided. The non-volatile storage medium includes a stored program, wherein when the program is running, the device where the non-volatile storage medium is located is controlled to execute any one of the above-mentioned noise monitoring methods.

[0013] According to yet another aspect of an embodiment of the present invention, a computer device is provided. The computer device includes a processor, and the processor is used to run a program. When the program is run, any one of the above-mentioned noise monitoring methods is executed.

[0014] According to yet another aspect of the embodiments of the present invention, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, any one of the above-mentioned noise monitoring methods is implemented.

[0015] In an embodiment of the present invention, a noise monitoring method is adopted, by obtaining historical working data of a target device, wherein the historical working data includes a vibration signal and a sound signal; determining feature data and a data deviation value based on the historical working data; performing feature selection based on the feature data and the data deviation value to obtain a plurality of feature sets, wherein the plurality of feature sets include a vibration signal feature set and a sound signal feature set; using the plurality of feature sets to train a preset original convolutional neural network model to obtain a target monitoring model, wherein the target monitoring model is used to monitor the deviation value of the working data; collecting the working data to be monitored of the target device and inputting it into the target monitoring model to obtain a deviation value; determining the noise monitoring result based on the relationship between the deviation value and the preset threshold value, thereby achieving the purpose of monitoring the working data based on the convolutional neural network model, thereby realizing the technical effect of improving the monitoring efficiency and accuracy, and further solving the technical problem that the current noise monitoring method mostly relies on manual inspections and simple sensor data analysis, resulting in low monitoring efficiency and high false alarm rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0017] Figure 1 A hardware structure block diagram of a computer terminal for implementing a noise monitoring method is shown;

[0018] Figure 2 is a flow chart of a noise monitoring method provided according to an embodiment of the present invention;

[0019] Figure 3 is a structural block diagram of a noise monitoring device provided according to an embodiment of the present invention. DETAILED DESCRIPTION

[0020] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0021] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0022] According to an embodiment of the present invention, a method embodiment of a noise monitoring method is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0023] The method embodiment provided in the first embodiment of the present application can be executed in a mobile terminal, a computer terminal or a similar computing device. Figure 1 FIG. 1 shows a hardware structure block diagram of a computer terminal for implementing a noise monitoring method. Figure 1 As shown, the computer terminal 10 may include one or more (102a, 102b, ..., 102n are used to illustrate) processors (the processor may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA), a memory 104 for storing data. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of the BUS bus), a network interface, a power supply and / or a camera. It can be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the above electronic device. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations are shown.

[0024] It should be noted that the one or more processors and / or other data processing circuits described above may generally be referred to herein as "data processing circuits". The data processing circuits may be embodied in whole or in part as software, hardware, firmware, or any other combination thereof. In addition, the data processing circuit may be a single independent processing module, or may be incorporated in whole or in part into any of the other components in the computer terminal 10. As described in the embodiments of the present application, the data processing circuit acts as a processor control (e.g., selection of a variable resistor terminal path connected to an interface).

[0025] The memory 104 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the noise monitoring method in the embodiment of the present invention. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory 104, that is, the noise monitoring method of the above-mentioned application is realized. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely arranged relative to the processor, and these remote memories may be connected to the computer terminal 10 via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0026] The display may be, for example, a touch screen liquid crystal display (LCD) that enables a user to interact with a user interface of the computer terminal 10 .

[0027] Figure 2 is a flow chart of a noise monitoring method according to an embodiment of the present invention. Figure 2 As shown, the method comprises the following steps:

[0028] Step S202, obtaining historical working data of the target device, wherein the historical working data includes vibration signals and sound signals.

[0029] In this step, the target device can be a device that emits vibration noise, such as motors, generators, conveyors and other equipment. Because the vibration noise emitted by such equipment can cause harm to the body of the staff, it is necessary to monitor the vibration noise and issue a timely warning when it exceeds a certain threshold. Collect the historical working data of the target device, the working data may include one type of data and two types of data, wherein the one type of data includes vibration signals and sound wave signals, and the second type of data includes temperature signals and location information. Specifically, data collection can be performed by setting up monitoring points on the target device. The historical working data obtained here can be used as a training set to train the model, and the trained model can be used for subsequent noise monitoring.

[0030] Step S204, determining characteristic data and data deviation value according to historical working data.

[0031] In this step, feature extraction can be performed on the historical working data to obtain feature data. And a type of data in the historical working data, namely vibration signals and sound signals, is subjected to deviation modal analysis to obtain data deviation values. Among them, the purpose of feature extraction on the working data is to extract the most meaningful features for subsequent analysis or model training from the original data. The method of feature extraction depends on the type of data and the target of analysis. In vibration and noise monitoring, feature extraction of working data mainly includes feature extraction of vibration signals, sound wave signals, and possible auxiliary signals (such as temperature and position information). For example, vibration signal feature extraction can include time domain features: including peak acceleration, root mean square acceleration, effective value acceleration, etc., which can reflect the intensity and fluctuation of vibration. Frequency domain features: Use Fourier transform (FFT) or wavelet transform to convert the signal to the frequency domain to extract features such as frequency, amplitude, and energy distribution. Time-frequency domain features: Complex features such as instantaneous frequency and instantaneous amplitude can be extracted. For the extraction of acoustic signal features, sound level features can be included, and the A-weighted sound level (dB(A)) can be calculated, which is a sound level measurement method that takes into account the sensitivity of the human ear to sounds of different frequencies. Spectral features: Through spectrum analysis, the main frequency, frequency distribution, bandwidth and other features of the acoustic signal can be extracted. Time-frequency features: The time-frequency features of the acoustic signal are extracted using methods such as short-time Fourier transform (STFT) or Mel-frequency cepstral coefficients (MFCC). These features are very useful for identifying specific noise patterns. Deviation analysis is an important means to detect changes in equipment performance and identify potential failure modes. By comparing with benchmark data, abnormal vibrations, temperature fluctuations, position offsets, etc. can be found, thereby warning of possible problems with the equipment. Specifically, modal analysis techniques (such as empirical mode decomposition EMD and Hibert transform) can be used to extract the inherent modes in the signal, that is, the intrinsic vibration modes of the signal. These modes are then quantified and characterized, such as calculating the instantaneous frequency and instantaneous amplitude. Compare the modal features of the current measurement with the modal features of the history or standard to find deviations or changes. Based on the deviation, the difference between the signal characteristics and the expected value is determined, that is, the data deviation value.

[0032] Step S206, performing feature selection based on the feature data and the data deviation value to obtain multiple feature sets, wherein the multiple feature sets include a vibration signal feature set and a sound signal feature set.

[0033] In this step, the feature data and the data deviation value can be combined to obtain a data set. Then, principal component analysis and feature selection are performed on the data set to obtain multiple feature sets. Among them, principal component analysis and feature selection are two important techniques for data dimensionality reduction and model optimization in data analysis and machine learning. Principal component analysis is used to find linear combinations between variables in a data set to identify the main variation direction of the data. Principal component analysis can convert high-dimensional data into low-dimensional data while retaining as much data information and structure as possible. This process is achieved by constructing a series of orthogonal (unrelated) principal components, each of which is a linear combination of the original features and is sorted by the size of the data variability it explains. The first principal component explains the largest variation in the data, the second principal component explains the largest variation in the remaining data, and so on. Feature selection refers to selecting the most relevant and informative features from the original feature set for building a machine learning model. Unlike PCA, feature selection does not create new feature combinations, but directly selects the most effective subset from existing features. Principal component analysis and feature selection can be used in combination to more comprehensively optimize data and models. Principal component analysis can be used for preliminary dimensionality reduction, reducing the number of features and dealing with redundancy in high-dimensional data sets, while feature selection can further select the features most relevant to the target variable on this basis to improve the prediction accuracy and efficiency of the model.

[0034] For example, in vibration and noise monitoring, there may be a large number of sensor readings related to the equipment status, and there may be a high correlation between these readings. Using principal component analysis can reduce the dimensionality of the data and reduce the computational complexity. However, the principal components obtained by principal component analysis may contain features that do not directly reflect the health status of the equipment. Therefore, after principal component analysis, feature selection techniques such as model-based representation, regularization, or statistical tests can be used to select the principal components that are most relevant to the prediction of abnormal equipment signals, which can further improve the accuracy of the subsequently trained model.

[0035] Step S208, using multiple feature sets to train the preset original convolutional neural network model to obtain a target monitoring model, wherein the target monitoring model is used to monitor the deviation value of the working data.

[0036] In this step, the acquired feature set can be used to train the original convolutional neural network model (CNN model). For example, each feature set (such as a vibration comprehensive feature set, a sound wave comprehensive feature set, and a fully comprehensive feature set) can be converted into a format suitable for CNN input, which usually means that the data needs to be reorganized into a multidimensional array or tensor. Then the data set is divided into a training set and a validation set as needed to evaluate the performance of the model on unseen data. A CNN model architecture can be designed, which may include multiple convolutional layers, pooling layers, fully connected layers, etc. The preset original CNN model can be a basic architecture, which may need to be adjusted later according to the training results. Select a suitable loss function and optimizer, which will guide the learning process of the model to minimize the prediction error. Use each feature set to train the CNN model separately. The training process of each feature set may need to be carried out independently, or they may be combined together using some multi-task learning framework. For each feature set, the preprocessed data is fed into the model, and the model parameters are updated through forward propagation and backpropagation processes to minimize the loss function. Use the validation set to evaluate the performance of the model trained on each feature set, which may include indicators such as precision, recall, and F1 score. After the model training is completed, an independent test set is used to finally evaluate the performance of the model and finally obtain the target monitoring model.

[0037] The process of training a CNN model using multiple feature sets can be considered a feature enhancement method. By analyzing data from different angles or levels, the model can learn more comprehensive and richer signal features.

[0038] As an optional embodiment, the original convolutional neural network model is a combination model of a convolutional neural network and a long short-term memory network.

[0039] Specifically, the training process of the combined model of convolutional neural network and long short-term memory network, namely CNN-LSTM network model, is as follows:

[0040] a. Construct a convolutional neural network-long short-term memory network to generate a deviation value prediction model;

[0041] The CNN-LSTM network consists of three parallel CNN-LSTM branches, a fully connected layer, and an output layer;

[0042] The CNN-LSTM branch is composed of an input layer, a CNN layer, a flattening layer, and an LSTM layer in series; the full comprehensive feature set, the vibration comprehensive feature set, and the sound wave comprehensive feature set are input into the three CNN-LSTM branches respectively; the data passes through the input layer, the CNN layer, the flattening layer, and the LSTM layer in sequence, and is output by the LSTM layer;

[0043] The output data of the LSTM layer is used as the input of the fully connected layer, which integrates and maps the input data features; the fully connected layer is connected to the output layer; the output layer generates a deviation value prediction model;

[0044] b. Use the comprehensive feature set test data to optimize the deviation value prediction model, adjust the CNN-LSTM network weights and bias through the loss function and optimizer, and output the noise monitoring model through multiple training cycles.

[0045] Among them, the CNN-LSTM network is a deep learning architecture that is very effective in processing time series and spatial data. CNN is good at processing data with spatial structure, such as images and signals, while LSTM is good at processing time series data. Combining the two can take advantage of their advantages and process complex spatiotemporal series data to achieve better noise monitoring results.

[0046] Step S210, collecting the working data to be monitored of the target device and inputting it into the target monitoring model to obtain the deviation value.

[0047] In this step, the real-time working data of the target device can be collected and input into the target monitoring model as the data to be monitored to obtain the deviation value. Among them, the deviation value can be used to evaluate the difference between the current operating state of the equipment and the normal operating state. The target monitoring model can predict the deviation between the data to be monitored and the normal data during training, that is, the deviation value. This deviation can be reflected in abnormal changes in vibration signals, noise signals, or a combination of them. The size of the deviation value reflects the degree of deviation between the current state of the equipment and the normal state. If the deviation value exceeds a certain threshold, this may mean that the equipment is approaching or is already in a fault state. Therefore, the system can trigger different levels of warnings according to the size of the deviation prediction value, so that operators or maintenance teams can take timely measures to avoid or reduce the impact and losses caused by equipment failure.

[0048] Step S212: determining the noise monitoring result based on the magnitude relationship between the deviation value and the preset threshold.

[0049] In this step, the noise monitoring result can be determined based on the magnitude relationship between the deviation value and the preset threshold. For example, a maximum threshold can be set, and when the deviation value exceeds the maximum threshold, the result is considered abnormal and needs to be checked by staff.

[0050] Through the above steps, the purpose of monitoring working data based on the convolutional neural network model is achieved, thereby realizing the technical effect of improving monitoring efficiency and accuracy, and further solving the technical problems that the current noise monitoring methods mostly rely on manual inspections and simple sensor data analysis, resulting in low monitoring efficiency and high false alarm rate.

[0051] As an optional embodiment, characteristic data is determined based on historical working data, including: decomposing the historical working data to obtain multiple components; performing time-frequency analysis on the multiple components to obtain characteristic data, wherein the characteristic data includes instantaneous frequency and instantaneous amplitude.

[0052] Optionally, the characteristic data can be determined by first decomposing the historical working data to obtain multiple vectors, and then performing time-frequency analysis on the vectors to obtain the characteristic data, wherein time-frequency analysis is a signal processing technique used to analyze the characteristics of the signal in both the time and frequency domains. Signal analysis is usually performed in a single domain (such as the time domain or the frequency domain), while time-frequency analysis can simultaneously consider the time and frequency characteristics of the signal. Through time-frequency analysis, it is possible to observe how the frequency composition of the signal changes over time, which is critical for identifying transient events, frequency modulation, or sudden noise sources in the signal.

[0053] For example, empirical mode decomposition (EMD) and Hibert transform are used to process the signal to be measured X(t), where the signal to be measured can be a vibration signal or a sound signal, and time-frequency analysis is performed to obtain a type of characteristic data, where the type of characteristic data includes instantaneous frequency ω(t) and instantaneous amplitude a(t);

[0054] The complex analytical signal Z(t) obtained by EMD decomposition and Hilbert transform of the measured signal X(t) is expressed as:

[0055]

[0056] where c i (t) is the component vector corresponding to the i-th IMF, r(t) is the residual value of the decomposition of the measured signal, j is the imaginary unit, and the Hilbert transform P is the Cauchy principal value at the transformation, τ is the integral variable, and t is the time parameter;

[0057] Calculate the instantaneous amplitude a(t) and instantaneous phase θ(t):

[0058]

[0059] The instantaneous frequency can be calculated by taking the derivative of the instantaneous phase θ(t):

[0060]

[0061] Specifically in actual situations, a type of data is subjected to EMD decomposition and Hibert transform, and the characteristic data (instantaneous amplitude of vibration, instantaneous frequency of vibration, instantaneous amplitude of sound wave, instantaneous frequency of sound wave) can be calculated, which can be: (0.75G, 10Hz, 97dB(A), 1200Hz); (1.15G, 15Hz, 102dB(A), 1100Hz); (0.45G, 8Hz, 88dB(A), 1000Hz); (1.45G, 20Hz, 107dB(A), 1050Hz); (0.65G, 12Hz, 87dB(A), 950Hz).

[0062] The acquired working data usually includes two types of data, where the second type of data usually includes temperature signals and position information. The second type of data is processed to obtain the second type of feature data, which includes the temperature change rate R T (t) and position deviation B P (t). The specific feature extraction method is as follows:

[0063] Calculate the rate of temperature change:

[0064]

[0065] Where T(t) and T(t-1) are the temperatures at the current time and the previous time respectively, and Δt is the time interval;

[0066] Calculate the position deviation:

[0067] B P (t)=|P(t)-M(P(t))|

[0068] Where P(t) is the current position and M(P(t)) is the average value of the position movement.

[0069] In the actual evaluation, the two types of data are processed to obtain (temperature change rate, position deviation):

[0071] As an optional embodiment, determining a data deviation value based on historical working data includes: determining a working data sequence based on the working data and a preset time sequence; sampling the working data sequence to obtain the sampled data and determining future actual values ​​corresponding to the sampled data; making predictions based on the sampled data to obtain predicted values ​​corresponding to the sampled data; determining a deviation value corresponding to the sampled data based on future actual values ​​and predicted values ​​corresponding to the sampled data; and determining a data deviation value based on the deviation value corresponding to the sampled data.

[0072] Optionally, to determine the data deviation value, the working data sequence can be obtained based on the working data and time sequence, and then sampling and feature extraction can be performed, that is, the working data is sampled and key features such as vibration signals, sound wave signals, temperature signals and position information are extracted from it. Then, a deviation model can be constructed using methods such as Halton sequence sampling and radial basis proxy model, which can predict what the sensor data of the device should be under a given state. Calculate the deviation value, input the working data into the deviation model, and the model will predict the expected value of these data under normal conditions. Then, by comparing the difference between the model prediction value and the actual collected value, the deviation value is calculated. The deviation value reflects the distance between the monitoring data and the normal working state.

[0073] Specifically, we can first perform de-randomized Halton sequence sampling on a type of data in the working data, namely, vibration signals and sound signals, to obtain j The VC sequence with {j=0, 1, 2, ..., N} as the basis represents the N-dimensional Halton sequence U N :

[0074] U N =[H b1 (n), H b2 (n),…,H bN (n)]

[0075] where integer b is a base number, b≥2, a i (n)∈{0, 1, 2, ..., b-1}, the VC sequence with basis b is expressed as

[0076] Perform digital permutation and scrambling operations on N-dimensional Halton sequences;

[0077] The radial basis proxy model is used to perform deviation modal analysis on a type of sampled data, and the Gaussian radial basis function is combined with the quadratic polynomial to obtain an improved Gaussian radial basis function:

[0078]

[0079] where ω i is the Gaussian radial basis function coefficient, ω r =(ω r1 ,ω r2 ,…,ω rN ) T , Gaussian function x i is the sample point, is the i-th sample mean, is the width parameter of the radial basis function, cmax is the maximum distance between the selected center points, g 2i is the coefficient of the quadratic term, g 1i is the coefficient of the first-order term, g 0 is a constant term;

[0080] ω r Satisfy the interpolation conditions:

[0081] f i =y i , i∈{1, 2, …, N}

[0082] Calculate the deviation value P i :

[0083] P i =f i -y i

[0084] where y i is the exact value, f i is the predicted value.

[0085] In the actual evaluation, a type of data is subjected to deviation modal analysis and then calculated to obtain (vibration signal deviation value, sound wave signal deviation value): (0.1G, 1dB(A)); (0.05G, 2dB(A)); (0.03G, 3dB(A)); (0.08G, 1dB(A)); (0.02G, 0dB(A)).

[0086] As an optional embodiment, feature selection is performed based on feature data and data deviation values ​​to obtain multiple feature sets, including: combining feature data and data deviation values ​​into a data set according to a preset time sequence; converting the data set from a high-dimensional feature space to a low-dimensional feature space to obtain a reduced-dimensional data set; and classifying and combining the reduced-dimensional data set based on preset feature types to obtain multiple feature sets.

[0087] Optionally, principal component analysis (PCA) can be used to convert the composed data set from a high-dimensional feature space to a low-dimensional feature space, which can remove noise and unnecessary complexity. Feature selection is then performed, that is, the most relevant features are selected from the principal components, or directly from the original features. For example, an evaluation criterion can be defined: the goal of feature selection is to identify the features that have the strongest explanatory power for the target variable (such as the normal or abnormal state of the equipment). Common evaluation criteria include correlation, information gain, coefficients of LASSO regression, etc. Feature screening is then performed, using the selected evaluation criteria to screen the most relevant features from the principal components or original features. This may involve multiple iterations and model training to find the best feature combination. A comprehensive feature set is constructed, and the screened features are combined together to form a comprehensive feature set. This feature set should contain features that can best reflect the changes in the equipment state, while reducing the dimensionality of the data and improving the training efficiency and prediction performance of the model. Among them, the comprehensive feature set can include multiple feature sets, such as a full comprehensive feature set, a vibration comprehensive feature set, and an acoustic comprehensive feature set.

[0088] Specifically, the timestamp information of the data deviation value may be extracted, and the first-class feature data and the second-class feature data obtained by the advanced signal processing may be matched according to the timestamp information, and the first-class feature data, the second-class feature data and the data deviation value may be aligned according to the time series;

[0089] The first-class feature data, second-class data and data deviation values ​​are combined into a data set, and principal component analysis and popular learning fusion algorithm PCA-LPP are used for feature selection;

[0090] Define the data set X = [x 1 , x 2 , …, x i , …, x M ]∈R, where x i ∈R N , M represents the number of samples, N represents the sample dimension, and the data is standardized.

[0091] Define the low-dimensional sample Y in the projection space = [y 1 ,y 2 , …, y i , …, y M ]∈R,y i ∈R L , L<<N, L represents the feature dimension in the low-dimensional space.

[0092] The overall objective function J(w) is the objective function J mapped by the PCA algorithm PCA (w) and the projection vector of LPP supervise the local objective function J LPP (w) Difference:

[0093]

[0094] Among them, β represents a unit vector, represents the mean vector, the covariance matrix K = diag[k ii ] is a diagonal matrix with diagonal elements k ii =[m ij ].

[0095] pass Laplace matrix L a = KD, simplifying J(w) to The objective function is connected to the constraints via the Lagrange multiplier method:

[0096] F(w,λ)=w T (G-XL a X T )w+λ(1-w T XD T w)

[0097] Derivative F(w, λ) with respect to w:

[0098] F ‘ (w) = 2(G-XL a X T )w+2λ(-XDX T w)

[0099] By F ‘ (w)=0 constructs generalized eigenvalue:

[0100] (G-XL a X T )w=λXDX T w

[0101] The eigenvectors λ are sorted in descending order according to the eigenvalues. Assuming that the dimension is reduced to d, the eigenvectors corresponding to the d largest eigenvalues ​​λ are obtained. Then according to W = [w 0 , w 1 , …, y d-1 ] form to construct the projection matrix, then the low-dimensional sample after projection is expressed as:

[0102] y i =W T x i

[0103] According to the contribution ranking, d principal components are selected to replace all the principal components. The screened principal components, data deviation values ​​and second-category data are combined into a full comprehensive feature set. The screened principal components, data deviation values ​​and second-category data are classified into a vibration comprehensive feature set and an acoustic comprehensive feature set according to vibration signals and acoustic wave signals. The three types of comprehensive feature sets are divided into training data and test data.

[0104] In practical applications, the original six features (instantaneous frequency of vibration signal, instantaneous amplitude of vibration signal, instantaneous frequency of acoustic wave signal, instantaneous amplitude of acoustic wave signal, temperature change rate, and position deviation) are selected according to PCA-LPP and reduced to four features (instantaneous frequency of vibration signal, instantaneous frequency of acoustic wave signal, temperature change rate, and position deviation). The comprehensive feature sets {vibration signal deviation value, instantaneous frequency of vibration signal, temperature change rate, position deviation temperature signal, and position information}, {sound wave signal deviation value, instantaneous frequency of acoustic wave signal, temperature change rate, position deviation temperature signal, and position information}, and {vibration signal deviation value, instantaneous frequency of vibration signal, sound wave signal deviation value, instantaneous frequency of acoustic wave signal, temperature change rate, position deviation temperature signal, and position information} are obtained through classification and combination.

[0105] As an optional embodiment, the noise monitoring result is determined based on the size relationship between the deviation value and the preset threshold value. When the deviation value includes the vibration signal deviation value and the sound signal deviation value, the method includes: determining a first warning value according to the vibration signal deviation value and the preset threshold value; determining a second warning value according to the sound signal deviation value and the preset threshold value; determining a target warning value according to the first warning value and the second warning value; and determining that the noise monitoring result is abnormal when the target warning value exceeds the preset warning value.

[0106] Optionally, when the deviation value includes the vibration signal deviation value and the sound signal deviation value, they can be judged separately first, and then the noise monitoring result can be determined. For example, the relationship between the vibration signal deviation value and the preset threshold value can be judged first to obtain the first warning value, and then the relationship between the sound signal deviation value and the preset threshold value can be judged to obtain the second warning value. Then, the noise monitoring result is judged according to the sum of the first warning value and the second warning value.

[0107] For example, the data to be monitored is input into the target monitoring model for deviation prediction, the warning value is determined based on the comparison result between the predicted deviation value and the threshold, and different levels of warnings are triggered based on the comprehensive warning value;

[0108] Calculate the warning value:

[0109]

[0110] Where a is the warning value, f(x) is the model prediction value, y is the actual value, θ min ,θ low and θhigh They are the minimum threshold, the low threshold, and the high threshold;

[0111] The warning trigger conditions are:

[0112]

[0113] The warning value includes vibration signal and sound signal, a z is the vibration signal warning value, a v is the sound signal warning value. The comprehensive warning value is the sum of the vibration signal and sound signal warning values. Three levels of warnings are triggered according to the comprehensive warning value.

[0114] Specifically, in actual situations, two groups of working data to be monitored are given: one type of data is (vibration signal, sound wave signal, timestamp): 1.1G (root mean square acceleration), 98dB (A), 2024-08-06T13:00:00Z; 0.6G (root mean square acceleration), 87dB (A), 2024-08-06T14:00:00Z;

[0115] The second type of data is (temperature signal, position signal, timestamp): 47°C, conveyor control system (3, -2, 5), 2024-08-06T13:00:00Z; 41°C, conveyor control system (-5, 8, -3), 2024-08-06T14:00:00Z;

[0116] The two sets of data were input into the deviation prediction model trained by CNL-STM to obtain the deviation prediction values ​​(vibration signal deviation value, acoustic signal deviation value): 0.1G, 2dB(A); 0.1G, 3dB(A);

[0117] Through the vibration signal and the sound wave signal corresponding to θ low ,θ min and θ high They are set to 5%, 10%, 20% and 2%, 5%, 8% respectively. The vibration signal warning value and sound signal warning value are 1, 0 and 2, 1 respectively. The corresponding comprehensive warning values ​​are 1 and 3. The triggering conditions are "no warning" and "level one warning" respectively.

[0118] It should be noted that, for the above-mentioned method embodiments, for the sake of simplicity, they are all described as a series of action combinations, but those skilled in the art should know that the present invention is not limited by the described action sequence, because according to the present invention, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present invention.

[0119] Through the description of the above implementation methods, those skilled in the art can clearly understand that the noise monitoring method according to the above embodiment can be implemented by means of software plus a necessary general hardware platform, and of course can also be implemented by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, a magnetic disk, or an optical disk), and includes a number of instructions for enabling a terminal device (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present invention.

[0120] According to an optional embodiment of the present invention, a noise monitoring system is further provided, which includes a sensor collection module, a cloud platform, a cloud data processing module, a cloud model module and a cloud supervision module, which are specifically as follows:

[0121] Sensing acquisition module: used to collect the working data of the monitoring target, pre-process the working data, and transmit the working data to the cloud platform;

[0122] Cloud platform: used to receive and process the work data; the cloud platform includes a cloud data processing module, a cloud model module and a cloud supervision module, and the data between the modules are interconnected;

[0123] Cloud data processing module: used for processing the working data to obtain a comprehensive feature set; the data processing includes advanced signal processing and deviation modal analysis;

[0124] Cloud model module: used for inputting the characteristic data and the data deviation value into the CNN-LSTM network to obtain a vibration noise monitoring model, and inputting the data to be monitored into the vibration noise monitoring model to obtain a deviation prediction value;

[0125] Cloud monitoring module: used to store, view and manage the working data, the data to be monitored and the deviation prediction value; used to provide graded warnings to users according to the deviation prediction value.

[0126] According to an embodiment of the present invention, a noise monitoring device for implementing the above noise monitoring method is also provided. Figure 3 is a structural block diagram of a noise monitoring device provided according to an embodiment of the present invention. Figure 3 As shown, the noise monitoring device includes: an acquisition module 302, a first determination module 304, a selection module 306, a training module 308, a collection module 310 and a monitoring module 312. The noise monitoring device is described below.

[0127] The acquisition module 302 is used to acquire historical working data of the target device, wherein the historical working data includes vibration signals and sound signals.

[0128] The first determination module 304 is connected to the acquisition module 302 and is used to determine the characteristic data and the data deviation value according to the historical working data.

[0129] The selection module 306 is connected to the first determination module 304 and is used to perform feature selection based on the feature data and the data deviation value to obtain multiple feature sets, wherein the multiple feature sets include a vibration signal feature set and a sound signal feature set.

[0130] The training module 308 is connected to the selection module 306 and is used to train the preset original convolutional neural network model using multiple feature sets to obtain a target monitoring model, wherein the target monitoring model is used to monitor the deviation value of the working data.

[0131] The acquisition module 310 is connected to the training module 308 and is used to acquire the working data to be monitored of the target device and input it into the target monitoring model to obtain the deviation value.

[0132] The monitoring module 312 is connected to the acquisition module 310 and is used to determine the noise monitoring result based on the magnitude relationship between the deviation value and a preset threshold.

[0133] It should be noted that the acquisition module 302, the first determination module 304, the selection module 306, the training module 308, the acquisition module 310 and the monitoring module 312 correspond to steps S202 to S212 in the embodiment, and the examples and application scenarios implemented by the multiple modules and the corresponding steps are the same, but are not limited to the contents disclosed in the above embodiment. It should be noted that the above modules as part of the device can be run in the computer terminal 10 provided in the embodiment.

[0134] An embodiment of the present invention may provide a computer device. Optionally, in this embodiment, the computer device may be located in at least one network device among multiple network devices of a computer network. The computer device includes a memory and a processor.

[0135] Among them, the memory can be used to store software programs and modules, such as the program instructions / modules corresponding to the noise monitoring method and device in the embodiment of the present invention. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory, that is, the above-mentioned noise monitoring method is realized. The memory may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory may further include a memory remotely arranged relative to the processor, and these remote memories can be connected to the computer terminal via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0136] The processor can call the information and application programs stored in the memory through the transmission device to perform the following steps: obtain the historical working data of the target device, wherein the historical working data includes vibration signals and sound signals; determine the feature data and data deviation values ​​according to the historical working data; perform feature selection based on the feature data and the data deviation values ​​to obtain multiple feature sets, wherein the multiple feature sets include vibration signal feature sets and sound signal feature sets; use the multiple feature sets to train the preset original convolutional neural network model to obtain a target monitoring model, wherein the target monitoring model is used to monitor the deviation value of the working data; collect the working data to be monitored of the target device and input it into the target monitoring model to obtain the deviation value; determine the noise monitoring result based on the relationship between the deviation value and the preset threshold value.

[0137] Optionally, the processor may also execute program code of the following steps: determining characteristic data based on historical working data, including: decomposing the historical working data to obtain multiple components; performing time-frequency analysis on the multiple components to obtain characteristic data, wherein the characteristic data includes instantaneous frequency and instantaneous amplitude.

[0138] Optionally, the processor may also execute program code for the following steps: determining a data deviation value based on historical working data, including: determining a working data sequence based on the working data and a preset time sequence; sampling the working data sequence to obtain sampled data and determining future actual values ​​corresponding to the sampled data; making predictions based on the sampled data to obtain predicted values ​​corresponding to the sampled data; determining a deviation value corresponding to the sampled data based on future actual values ​​and predicted values ​​corresponding to the sampled data; and determining a data deviation value based on the deviation value corresponding to the sampled data.

[0139] Optionally, the processor may also execute the program code of the following steps: performing feature selection based on feature data and data deviation values ​​to obtain multiple feature sets, including: combining feature data and data deviation values ​​into a data set according to a preset time sequence; converting the data set from a high-dimensional feature space to a low-dimensional feature space to obtain a reduced-dimensional data set; and classifying and combining the reduced-dimensional data set based on preset feature types to obtain multiple feature sets.

[0140] Optionally, the processor may also execute program code for the following steps: determining the noise monitoring result based on the size relationship between the deviation value and a preset threshold value, and when the deviation value includes a vibration signal deviation value and a sound signal deviation value, including: determining a first warning value based on the vibration signal deviation value and the preset threshold value; determining a second warning value based on the sound signal deviation value and the preset threshold value; determining a target warning value based on the first warning value and the second warning value; and determining that the noise monitoring result is abnormal when the target warning value exceeds the preset warning value.

[0141] Optionally, the processor may also execute the program code of the following steps: the original convolutional neural network model is a combination model of a convolutional neural network and a long short-term memory network.

[0142] An embodiment of the present invention provides a noise monitoring method, which obtains historical working data of a target device, wherein the historical working data includes a vibration signal and a sound signal; determines feature data and a data deviation value based on the historical working data; performs feature selection based on the feature data and the data deviation value to obtain multiple feature sets, wherein the multiple feature sets include a vibration signal feature set and a sound signal feature set; uses the multiple feature sets to train a preset original convolutional neural network model to obtain a target monitoring model, wherein the target monitoring model is used to monitor the deviation value of the working data; collects the working data to be monitored of the target device and inputs it into the target monitoring model to obtain a deviation value; determines the noise monitoring result based on the relationship between the deviation value and a preset threshold value, thereby achieving the purpose of monitoring the working data based on the convolutional neural network model, thereby realizing the technical effect of improving the monitoring efficiency and accuracy, and further solving the technical problem that the current noise monitoring method mostly relies on manual inspections and simple sensor data analysis, resulting in low monitoring efficiency and high false alarm rate.

[0143] A person of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing the hardware related to the terminal device through a program, and the program can be stored in a non-volatile storage medium, and the storage medium may include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0144] The embodiment of the present invention further provides a non-volatile storage medium. Optionally, in this embodiment, the non-volatile storage medium can be used to store the program code executed by the noise monitoring method provided in the above embodiment.

[0145] Optionally, in this embodiment, the non-volatile storage medium may be located in any computer terminal in a computer terminal group in a computer network, or in any mobile terminal in a mobile terminal group.

[0146] Optionally, in this embodiment, the non-volatile storage medium is configured to store program codes for executing the following steps: obtaining historical working data of the target device, wherein the historical working data includes vibration signals and sound signals; determining feature data and data deviation values ​​based on the historical working data; performing feature selection based on the feature data and the data deviation values ​​to obtain multiple feature sets, wherein the multiple feature sets include vibration signal feature sets and sound signal feature sets; using multiple feature sets to train a preset original convolutional neural network model to obtain a target monitoring model, wherein the target monitoring model is used to monitor the deviation value of the working data; collecting the working data to be monitored of the target device and inputting it into the target monitoring model to obtain the deviation value; and determining the noise monitoring result based on the relationship between the deviation value and the preset threshold.

[0147] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for executing the following steps: determining characteristic data based on historical working data, including: decomposing the historical working data to obtain multiple components; performing time-frequency analysis on the multiple components to obtain characteristic data, wherein the characteristic data includes instantaneous frequency and instantaneous amplitude.

[0148] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for executing the following steps: determining a data deviation value based on historical working data, including: determining a working data sequence based on the working data and a preset time sequence; sampling the working data sequence to obtain the sampled data and determining future actual values ​​corresponding to the sampled data; making predictions based on the sampled data to obtain predicted values ​​corresponding to the sampled data; determining a deviation value corresponding to the sampled data based on future actual values ​​and predicted values ​​corresponding to the sampled data; and determining a data deviation value based on the deviation value corresponding to the sampled data.

[0149] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for executing the following steps: performing feature selection based on feature data and data deviation values ​​to obtain multiple feature sets, including: combining the feature data and data deviation values ​​into a data set according to a preset time sequence; converting the data set from a high-dimensional feature space to a low-dimensional feature space to obtain a reduced-dimensional data set; and classifying and combining the reduced-dimensional data set based on preset feature types to obtain multiple feature sets.

[0150] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for executing the following steps: determining the noise monitoring result based on the size relationship between the deviation value and the preset threshold, and when the deviation value includes a vibration signal deviation value and a sound signal deviation value, including: determining a first warning value based on the vibration signal deviation value and the preset threshold; determining a second warning value based on the sound signal deviation value and the preset threshold; determining a target warning value based on the first warning value and the second warning value; and determining that the noise monitoring result is abnormal when the target warning value exceeds the preset warning value.

[0151] Optionally, in this embodiment, the non-volatile storage medium is configured to store program codes for executing the following steps: the original convolutional neural network model is a combination model of a convolutional neural network and a long short-term memory network.

[0152] An embodiment of the present invention also provides a computer program product, including a computer program. Optionally, in this embodiment, when the computer program is executed by a processor, it can achieve: obtaining historical working data of a target device, wherein the historical working data includes vibration signals and sound signals; determining feature data and data deviation values ​​based on the historical working data; performing feature selection based on the feature data and the data deviation values ​​to obtain multiple feature sets, wherein the multiple feature sets include vibration signal feature sets and sound signal feature sets; using multiple feature sets to train a preset original convolutional neural network model to obtain a target monitoring model, wherein the target monitoring model is used to monitor the deviation value of the working data; collecting the working data to be monitored of the target device and inputting it into the target monitoring model to obtain the deviation value; and determining the noise monitoring result based on the relationship between the deviation value and the preset threshold.

[0153] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0154] In the above embodiments of the present invention, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0155] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units can be a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0156] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0157] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0158] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a non-volatile storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk and other media that can store program codes.

[0159] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A noise monitoring method, characterized in that: include: Acquiring historical working data of the target device, wherein the historical working data includes a vibration signal and a sound signal; Determining characteristic data and data deviation values ​​according to the historical working data; Based on the feature data and the data deviation value, feature selection is performed to obtain a plurality of feature sets, wherein the plurality of feature sets include a vibration signal feature set and a sound signal feature set; The preset original convolutional neural network model is trained using the multiple feature sets to obtain a target monitoring model, wherein the target monitoring model is used to monitor the deviation value of the working data; Collecting the working data to be monitored of the target device and inputting it into the target monitoring model to obtain a deviation value; Based on the magnitude relationship between the deviation value and a preset threshold, a noise monitoring result is determined.

2. The method according to claim 1, characterized in that The determining of characteristic data according to the historical working data includes: Decomposing the historical work data to obtain multiple components; Performing time-frequency analysis on the multiple components to obtain the characteristic data, wherein the characteristic data includes instantaneous frequency and instantaneous amplitude.

3. The method according to claim 1, characterized in that The step of determining the data deviation value according to the historical working data includes: Determine a working data sequence according to the working data and a preset time sequence; Sampling the working data sequence to obtain sampled data and determining future actual values ​​corresponding to the sampled data; Perform prediction based on the sampled data to obtain a predicted value corresponding to the sampled data; Determine the deviation value corresponding to the sampled data according to the future actual value and the predicted value corresponding to the sampled data; The data deviation value is determined according to the deviation value corresponding to the sampled data.

4. The method according to claim 1, characterized in that: The feature selection is performed based on the feature data and the data deviation value to obtain multiple feature sets, including: The characteristic data and the data deviation value are combined into a data set according to a preset time sequence; Converting the data set from a high-dimensional feature space to a low-dimensional feature space to obtain a reduced-dimensional data set; Based on preset feature types, the reduced-dimensional data sets are classified and combined to obtain the multiple feature sets.

5. The method according to claim 1, characterized in that The determining of the noise monitoring result based on the magnitude relationship between the deviation value and the preset threshold, when the deviation value includes a vibration signal deviation value and a sound signal deviation value, comprises: Determining a first warning value according to the vibration signal deviation value and the preset threshold; Determining a second warning value according to the sound signal deviation value and the preset threshold; Determining a target warning value according to the first warning value and the second warning value; When the target warning value exceeds the preset warning value, the noise monitoring result is determined to be abnormal.

6. The method according to any one of claims 1 to 5, characterized in that: The original convolutional neural network model is a combination model of a convolutional neural network and a long short-term memory network.

7. A noise monitoring device, characterized in that: include: An acquisition module, used to acquire historical working data of a target device, wherein the historical working data includes a vibration signal and a sound signal; A first determination module, used to determine characteristic data and data deviation values ​​according to the historical working data; A selection module, configured to perform feature selection based on the feature data and the data deviation value to obtain a plurality of feature sets, wherein the plurality of feature sets include a vibration signal feature set and a sound signal feature set; A training module, used to train a preset original convolutional neural network model using the multiple feature sets to obtain a target monitoring model, wherein the target monitoring model is used to monitor the deviation value of the working data; A collection module, used for collecting the working data to be monitored of the target device and inputting it into the target monitoring model to obtain a deviation value; The monitoring module is used to determine the noise monitoring result based on the magnitude relationship between the deviation value and a preset threshold.

8. A non-volatile storage medium, characterized in that: The non-volatile storage medium includes a stored program, wherein when the program is executed, the device where the non-volatile storage medium is located is controlled to execute the noise monitoring method according to any one of claims 1 to 6.

9. A computer device, characterized in that: include: Memory and processor, The memory stores a computer program; The processor is used to execute the computer program stored in the memory, and when the computer program is run, the processor executes the noise monitoring method according to any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the noise monitoring method according to any one of claims 1 to 6 is implemented.

Images