Turbocharger surge identification method and system, electronic equipment and storage medium

Abstract

The invention provides a turbocharger surge identification method and system, electronic equipment and a storage medium. The turbocharger surge identification method comprises the steps that multi-channel sound signal data of a turbocharger are acquired; performing feature extraction on the preprocessed sound signal data to obtain multi-dimensional and multi-domain composite features, and performing dimension reduction processing on the multi-dimensional and multi-domain composite features to form a three-dimensional sensitive feature set; on the basis of the constructed supercharger surge identification model, surge results of different sound signal channels are predicted, so that surge prediction results of sound signal data of all the channels are obtained; and performing decision fusion on the surge prediction result of each channel through an improved DS evidence theory to obtain a fused surge identification result, and performing turbocharger surge identification according to the surge identification result. According to the method, high-precision recognition of the surge state of the turbocharger is achieved, the model adaptability and robustness in the multi-working-condition and multi-channel environment are remarkably improved, and reliable technical support is provided for industrial application.

Description

Technical Field

[0001] The present application belongs to the technical field of turbocharger surge fault identification, and in particular, relates to a turbocharger surge identification method, system, electronic equipment and storage medium. Background Art

[0002] Traditional turbocharger surge identification usually relies on subjective judgment by test personnel through hearing and vision. When a "buzzing" sound is heard, it is considered that the turbocharger has entered a surge state. However, this experience-based judgment method is not only low in accuracy and difficult to accurately determine the near-surge point range of the compressor, but may also lead to untimely and inaccurate fault diagnosis.

[0003] Existing surge identification methods mostly use the change of compressor outlet pressure signal as the basis for judgment. Although it can accurately identify the occurrence of surge to a certain extent, since the pressure sensor is an invasive sensor, its installation inside the outlet pipe will change the outlet path and may affect the performance of the compressor. This method is difficult to be widely used in actual engineering. Therefore, non-invasive acoustic signals are used as monitoring parameters for turbocharger surge identification. By analyzing the characteristics of acoustic signals, turbocharger surge identification based on single-channel acoustic signals is realized. However, single-channel acoustic signals have obvious limitations in information acquisition and are easily affected by external interference, resulting in inaccurate surge identification results, which affects the final identification and judgment of the turbocharger surge state. Summary of the invention

[0004] In view of this, the present application aims to propose a turbocharger surge identification method, system, electronic device and storage medium to solve the problem that a single channel has large identification error and cannot identify the turbocharger surge state with high accuracy.

[0005] To achieve the above purpose, the technical solution of this application is implemented as follows: In a first aspect, the present application provides a turbocharger surge identification method, comprising: Acquire multi-channel acoustic signal data of turbocharger; Extracting features from the preprocessed acoustic signal data to obtain multi-dimensional and multi-domain composite features, and performing dimensionality reduction processing on the multi-dimensional and multi-domain composite features to form a three-dimensional sensitive feature set; Based on the constructed supercharger surge recognition model, the surge results of different acoustic signal channels are predicted respectively to obtain the surge prediction results of the acoustic signal data of each channel, wherein the supercharger surge recognition model is a bidirectional gated recurrent unit classification model based on the whale optimization algorithm and is trained by the three-dimensional sensitive feature set; The surge prediction results of each channel are fused by decision making through the improved DS evidence theory to obtain a fused surge identification result, and turbocharger surge identification is performed according to the surge identification result.

[0006] In a second aspect, based on the same inventive concept, the present application also provides a turbocharger surge identification system, comprising: A signal acquisition module is configured to acquire multi-channel acoustic signal data of a turbocharger; A preprocessing module is configured to perform feature extraction on the preprocessed acoustic signal data to obtain multi-dimensional and multi-domain composite features, and perform dimensionality reduction processing on the multi-dimensional and multi-domain composite features to form a three-dimensional sensitive feature set; A model building module is configured to predict surge results of different acoustic signal channels respectively based on the constructed supercharger surge recognition model to obtain surge prediction results of acoustic signal data of each channel, wherein the supercharger surge recognition model is a bidirectional gated recurrent unit classification model based on a whale optimization algorithm and is trained by the three-dimensional sensitive feature set; The decision fusion module is configured to perform decision fusion on the surge prediction results of each channel through the improved DS evidence theory to obtain a fused surge identification result, and perform turbocharger surge identification according to the surge identification result.

[0007] In a third aspect, based on the same inventive concept, the present application also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method described in the first aspect is implemented.

[0008] In a fourth aspect, based on the same inventive concept, the present application also provides a non-transitory computer-readable storage medium, wherein the non-transitory computer-readable storage medium stores computer instructions, and the computer instructions are used to enable the computer to execute the method described in the first aspect.

[0009] Compared with the prior art, the turbocharger surge identification method, system, electronic device and storage medium described in this application have the following beneficial effects: The turbocharger surge identification method, system, electronic device and storage medium described in the present application achieve high-precision identification of the turbocharger surge state, significantly improve the model adaptability and robustness under multi-operating conditions and multi-channel environments, and provide reliable technical support for the realization of industrial applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] The drawings constituting a part of the present application are used to provide a further understanding of the present application. The illustrative embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings: Figure 1 A flow chart of a turbocharger surge identification method according to an embodiment of the present application; Figure 2 Schematic diagram of a turbocharger surge reproduction and multi-directional sound signal acquisition test system according to an embodiment of the present application; Figure 3 A schematic diagram of a signal processing process of a turbocharger surge identification method according to an embodiment of the present application; Figure 4 This is a schematic structural diagram of a turbocharger surge identification device according to an embodiment of the present application; Figure 5 This is a schematic diagram of the hardware structure of the electronic device described in the embodiment of the present application. DETAILED DESCRIPTION

[0011] In order to make the objectives, technical solutions and advantages of the present application more clearly understood, the present application is further described in detail below in combination with specific embodiments and with reference to the accompanying drawings.

[0012] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the embodiments of the present application should be the usual meanings understood by people with ordinary skills in the field to which the present application belongs. The "first", "second" and similar words used in the embodiments of the present application do not represent any order, quantity or importance, but are only used to distinguish different components. "Including" or "comprising" and similar words mean that the elements or objects appearing in front of the word cover the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0013] The embodiments of the present application are described in detail below with reference to the accompanying drawings.

[0014] See also Figure 1 and Figure 3 As shown, this embodiment provides a method for identifying turbocharger surge, which specifically includes the following steps: Step S1, obtaining multi-channel acoustic signal data of a turbocharger.

[0015] Specifically, in this embodiment, four noise sensors are used to collect acoustic signal data from four different positions of the turbocharger to obtain multi-channel acoustic signal raw data with labels under different working conditions. ,in, i For different number of channels,n is the number of acoustic signal data per channel. Through the data collected from multiple tests under multiple working conditions, a complete turbocharger surge acoustic signal database is established. The turbocharger surge reproduction and acoustic signal acquisition test system is shown in the attached figure. Figure 2 shown.

[0016] Step S2: extracting features from the pre-processed acoustic signal data to obtain multi-dimensional and multi-domain composite features, and performing dimensionality reduction processing on the multi-dimensional and multi-domain composite features to form a three-dimensional sensitive feature set.

[0017] Specifically, in this embodiment, this step specifically includes the following steps: Step S21, pre-processing the collected sound signal data, firstly performing noise reduction processing, using the wavelet packet threshold noise reduction method to reduce the noise of the sound signal and remove the interference of high-frequency noise. The wavelet packet noise reduction formula is:

[0018] Where: is the sound signal to be transformed, is the scale factor, is the translation factor, is the base wavelet.

[0019] Step S22: After noise reduction, the acoustic signal data is framed and windowed, and the framed data is standardized using a standardization method based on a sliding window to complete the preprocessing of the acoustic signal data.

[0020] Step S23: extract multi-dimensional and multi-domain composite features from the time domain, frequency domain, and cepstrum domain.

[0021] The extracted time domain features include dimensional and dimensionless indicators. There are 10 dimensional indicators extracted, including maximum value, minimum value, average value, root mean square value, variance, standard deviation, peak value, peak-to-peak value, short-time zero-crossing rate and short-time energy; there are 6 dimensionless indicators extracted, including skewness, kurtosis, impulse factor, peak factor, margin factor and waveform factor. The time domain features have 16 dimensions. The extracted frequency domain features include spectral kurtosis, centroid frequency, mean square frequency, root mean square frequency, frequency variance and frequency standard deviation, with a total of 6 dimensions. The extracted cepstral domain features include Mel-Frequency Cepstral Coefficients (MFCC) and Linear Prediction Cepstral Coefficients (LPCC). MFCC includes 13-dimensional features and LPCC includes 16-dimensional features. Based on the above features, a 51-dimensional multi-domain composite feature of turbocharger surge sound signal is constructed.

[0022] Step S24: To further reduce redundant information and enhance model calculation efficiency, the UMAP method is used to reduce the dimension of the extracted multidimensional features. X ={ x 1 , x 2 , ..., x n} is a multi-domain composite feature dataset of acoustic signals. For any x i ,definition and :

[0023] Where: for x i To its nearest neighbor distance; is a Riemann scale of the k-nearest-neighbor graph, under which k The data in the neighbor graph is approximately evenly distributed; for x i The distance to its nearest neighbor ensures that there is at least one point with an edge weight of 1. x i connect.

[0024] Defining a directed weighted graph ,use The symmetry definition of an undirected weighted graph Assuming Q is the weighted adjacency matrix of G, we can get a symmetric matrix:

[0025] Where: The undirected weighted neighborhood graph G can be defined by the adjacency matrix R; is the Hadamard product (dot product).

[0026] UMAP applies gravitational and repulsive forces along the boundaries and vertices respectively to evolve a point set { y i}, i =1, 2, ..., n Equivalent weighted graph H , two vertices i and j In coordinates y i and y j The gravitational force at is:

[0027] Where:a and b is a hyperparameter, the default value is a ≈1.929, b ≈0.7915. The repulsive force is:

[0028] Where: It is a minimum value, the default value is 0.001.

[0029] Through the above calculations, we get the fuzzy topological representation of high-dimensional data sets and low-dimensional data sets respectively. G and H , G and H The difference between them is calculated using cross entropy to match the topological structure of the determined original data and obtain the low-dimensional output of the data. Y ={ y 1 , y 2 , ..., y n}, thus realizing the dimension reduction and visualization of the multi-domain composite features of the turbocharger surge sound signal. According to the above method, the 3D sensitive feature set of turbocharger surge is obtained.

[0030] Step S3: predicting the surge results of different acoustic signal channels respectively based on the constructed supercharger surge recognition model to obtain the surge prediction results of the acoustic signal data of each channel, wherein the supercharger surge recognition model is a bidirectional gated recurrent unit classification model based on the whale optimization algorithm and is trained by the three-dimensional sensitive feature set.

[0031] Specifically, in this embodiment, this step specifically includes the following steps: Step S31: Construct a supercharger surge recognition model based on the obtained 3D sensitive feature set and select a bidirectional gated recurrent unit (BiGRU) model for supercharger surge recognition. BiGRU consists of two independent GRU units. GRU has two gating units, namely update gate and reset gate, to control the flow of information. The reset gate is used to update the hidden state of the previous moment. h t−1 With the current input y t Splicing, calculating the candidate state The update gate controls the hidden state of the previous moment h t−1 The candidate state at the current moment Integrate to get the current status information output value ht , and then get the prediction result of GRU. The calculation and update of each link of GRU are as follows: The update gate z t and reset gate r t The calculation process is:

[0032] Candidate status and current status h t The calculation process is:

[0033] Where: h t , h t−1 are the output values ​​of the hidden state at the current moment and the previous moment respectively; y t is the input at the current moment; is the sigmoid activation function; W z , W r and W h is the corresponding weight matrix.

[0034] The two GRUs process the input acoustic signal feature sample sequence in both the forward and reverse time order. Y ={ y 1 , y 2 , ..., y n} to process and merge their respective outputs to obtain the output sequence H ={ h 1 , h 2 , ..., h n}.

[0035] Step S32: The Whale Optimization Algorithm (WOA) is used to optimize the model parameters, thereby constructing a WOA-BiGRU-based supercharger surge recognition model for realizing surge recognition of single-channel acoustic signals. Specifically, the WOA algorithm is used to optimize the key parameters of the BiGRU model, including the learning rate, the loss rate, and the number of GRU layer units. The global search capability of the WOA algorithm is used to determine the optimal combination of BiGRU model parameters, thereby improving the performance of the model.

[0036] Step S33: using the constructed WOA-BiGRU classification model, respectively predict the surge results of four different acoustic signal channels to obtain the supercharger surge recognition results of each channel.

[0037] Step S4: performing decision fusion on the surge prediction results of each channel by using the improved DS evidence theory to obtain a fused surge identification result, and performing turbocharger surge identification according to the surge identification result.

[0038] Specifically, in this embodiment, this step specifically includes the following steps: Step S41, by performing joint weight calculation on the predicted state probabilities of different channels, a revised basic probability distribution is obtained. is a mutually exclusive finite non-empty set, called the recognition frame, which contains all possible results θ j , the recognition framework in this paper It is composed of the state information of the supercharger, including normal state and surge state. Basic Probability Assignment (BPA), also known as mass function (mass, m), is m from the power set →[0, 1], while satisfying the following constraints:

[0039] Where: It reflects that the sum of the basic probability assignments of all propositions is equal to 1, which ensures that the sum of the probabilities in the entire proposition space is 1; A It is a power set A proposition in m ( A ) indicates that the evidence is A The support of is the basic probability distribution of the proposition. If m ( A )>0, then A It is called the focal element.

[0040] The supercharger state is predicted by the BiGRU classification model using the data from each channel, and the probability of the predicted result is used as the original BPA distribution of the DS evidence theory. For evidence from different sources, Dempster proposed to use orthogonal sum to combine the quality functions of different evidences to achieve exchangeability and associativity. m 1 and m 2 is the quality function of the two pieces of evidence on the discriminating framework, then the combination m 1 and m 2 The combination rule is expressed as:

[0041] Where: K The values ​​represent the normalization coefficients, defined as:

[0042] coefficient K Indicates the degree of conflict between evidences. K The larger the value, the more severe the conflict. K =0 means that the BPA of different evidence sources are completely consistent. K =1 means that the BPA distributions of different evidence sources are completely conflicting.

[0043] Step S42: In order to eliminate conflicts and paradoxes, the DS evidence theory is improved by establishing new synthesis rules and modifying the initial basic probability distribution. The Josselme distance matrix and Deng entropy fuzzy preference matrix are introduced, and the joint weight is calculated to modify the original BPA.

[0044] Step S421: The Jousellme distance is a widely used distance. The Jousellme distance is used to calculate the distance between two evidence sources. d J ( m 1 , m 2 ), and its calculation formula is:

[0045] Where: m 1 , m 2 is a sequence vector in the power set space; D for A positive definite matrix of order , where:

[0046] According to the obtained Jousselme distance matrix, calculate the support degree of the evidence Ds i , thereby determining the objective weight. Ds i Indicates the degree of support of the evidence. Ds i The larger it is, the more supported the evidence is.

[0047]

[0048] The objective weight of each sensor can be determined after the degree of support from the evidence D i for:

[0049] Jousselme distance measures the distance between evidences by considering the focal element of the evidence, but ignores the information carried by the evidence itself.

[0050] Step S422, the application of information entropy can effectively consider the information contained in the evidence, providing an opportunity to solve this problem. Fuzzy preference relationship is a framework for dealing with decision-making problems. Using fuzzy preference relationships based on various information entropies to correct basic probability distribution is an effective method. This paper uses a fuzzy preference matrix based on Deng entropy to calculate the weight of each channel. The calculation formula of Deng entropy is:

[0051] Where: |A| is the size of the focal element A, E d Is the entropy value.

[0052] According to the calculated Deng entropy value, calculate the variance of the Deng entropy value V i . The off-diagonal elements in the fuzzy preference matrix p ij ∈[0,1] is calculated as follows:

[0053] The obtained fuzzy preference matrix P for:

[0054] Calculate the consistency matrix based on the fuzzy preference matrix:

[0055] According to the consistency relationship matrix, we can get information about each candidate target. m i The sort value of RV i , which is defined as follows:

[0056] Where: RV i is the ranking value obtained based on the Deng entropy fuzzy preference matrix; is the consistency matrix.

[0057] The ranking value calculated by the above formula can be used as the credibility of evidence or as the weight of evidence to indicate the importance of evidence.

[0058] Step S43: The objective weight obtained by the Jousselme distance matrix Jw i And the ranking value obtained based on the Deng entropy fuzzy preference matrix RV i Perform joint weight calculation to obtain joint weight JD i .

[0059]

[0060] Using joint weights JD i The original BPA is modified to obtain a modified BPA, and the modified BPA is used to perform decision fusion on multiple evidence bodies through the Dempster synthesis rule to obtain the final prediction result.

[0061] Experiments show that this method can achieve high-precision identification of turbocharger surge state, significantly improve the adaptability and robustness of the model under multi-operating conditions and multi-channel environments, and provide reliable technical support for the realization of industrial applications.

[0062] It should be noted that the above describes some embodiments of the present application. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the above embodiments and still achieve the desired results. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0063] Based on the same inventive concept, corresponding to any of the above-mentioned embodiments and methods, an embodiment of the present application further provides a turbocharger surge identification system.

[0064] like Figure 4 As shown, the turbocharger surge identification system comprises: The signal acquisition module 11 is configured to acquire multi-channel acoustic signal data of the turbocharger.

[0065] Specifically, in this embodiment, the module designs and constructs a high-precision turbocharger surge reproduction and multi-directional sound signal acquisition test system, which can accurately reproduce the turbocharger surge phenomenon under various speed conditions. The system is equipped with a multi-channel sound signal acquisition device (i.e., noise sensor) to fully capture the complete sound signal data of the turbocharger during the surge reproduction process from four different directions. At the same time, based on the collected sound signal data, a complete surge sound signal database covering different working conditions is established, providing high-quality data support for the subsequent feature extraction and recognition model construction.

[0066] The preprocessing module 12 is configured to perform feature extraction on the preprocessed acoustic signal data to obtain multi-dimensional and multi-domain composite features, and perform dimensionality reduction processing on the multi-dimensional and multi-domain composite features to form a three-dimensional sensitive feature set.

[0067] Specifically, in this embodiment, the module extracts multiple composite features from the time domain, frequency domain and cepstrum domain of the acoustic signal to establish a multi-domain feature set covering 51-dimensional features. Among them, the wavelet packet threshold denoising technology is used to perform deep denoising on the signal, which significantly improves the purity and credibility of the feature signal. In order to further reduce redundant information and enhance the calculation efficiency of the model, the uniform manifold approximation and projection method (UMAP) is used to reduce the dimension of the preferred features, and finally a low-dimensional set with 3 sensitive features is formed, which can accurately characterize the surge state of the turbocharger and provide a solid foundation for the construction of the recognition model.

[0068] The model building module 13 is configured to predict the surge results of different acoustic signal channels respectively based on the constructed supercharger surge recognition model to obtain the surge prediction results of the acoustic signal data of each channel, wherein the supercharger surge recognition model is a bidirectional gated recurrent unit classification model based on the whale optimization algorithm and is trained by the three-dimensional sensitive feature set.

[0069] Specifically, in this embodiment, the module is based on a set of sensitive features and uses a combination of machine learning and deep learning to establish a turbocharger surge recognition model suitable for single-channel acoustic signals. By repeatedly training and verifying the model, the feature weight distribution is optimized to improve the recognition accuracy and stability of the model. Specifically, the model uses a bidirectional gated recurrent unit (BiGRU) optimized by the whale optimization algorithm (WOA) to fully exploit the time series characteristics of the acoustic signal and its intrinsic correlation. By calculating the surge recognition rate and misjudgment rate, the robustness and accuracy of the model are verified, which can accurately evaluate the operating status of the turbocharger and output diagnostic results, providing a reliable basis for subsequent research.

[0070] The decision fusion module 14 is configured to perform decision fusion on the surge prediction results of each channel through the improved DS evidence theory to obtain a fused surge identification result, and perform turbocharger surge identification according to the surge identification result.

[0071] Specifically, in this embodiment, based on the single-channel recognition model, this module further proposes a decision fusion method based on multi-channel acoustic signals to improve the global accuracy and robustness of recognition. For the acoustic signal data collected from different channels, independent surge recognition models are trained respectively, and the improved Dempster-Shafer (DS) evidence theory is used to make weighted fusion decisions on the recognition results of each channel, effectively reducing the impact of single channel recognition errors.

[0072] For the convenience of description, the above system is described by dividing the functions into various modules. Of course, when implementing the embodiments of the present application, the functions of each module can be implemented in the same or multiple software and / or hardware.

[0073] The system of the above embodiment is used to implement the corresponding method in any of the above embodiments, and has the beneficial effects of the corresponding method embodiment, which will not be repeated here.

[0074] Based on the same inventive concept, corresponding to any of the above-mentioned embodiments and methods, an embodiment of the present application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method described in any of the above embodiments is implemented.

[0075] Figure 5 A more specific schematic diagram of the hardware structure of an electronic device provided in this embodiment is shown, and the device may include: a processor 1010, a memory 1020, an input / output interface 1030, a communication interface 1040, and a bus 1050. The processor 1010, the memory 1020, the input / output interface 1030, and the communication interface 1040 are connected to each other through the bus 1050 in the device.

[0076] The processor 1010 can be implemented by a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this specification.

[0077] The memory 1020 may be implemented in the form of ROM (Read Only Memory), RAM (Random Access Memory), static storage device, dynamic storage device, etc. The memory 1020 may store an operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented by software or firmware, the relevant program codes are stored in the memory 1020 and are called and executed by the processor 1010.

[0078] The input / output interface 1030 is used to connect the input / output module to realize information input and output. The input / output module can be configured in the device as a component (not shown in the figure), or it can be externally connected to the device to provide corresponding functions. The input device may include a keyboard, a mouse, a touch screen, a microphone, various sensors, etc., and the output device may include a display, a speaker, a vibrator, an indicator light, etc.

[0079] The communication interface 1040 is used to connect a communication module (not shown) to realize communication interaction between the device and other devices. The communication module can realize communication through wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.).

[0080] The bus 1050 includes a path that transmits information between the various components of the device (eg, the processor 1010 , the memory 1020 , the input / output interface 1030 , and the communication interface 1040 ).

[0081] It should be noted that, although the above device only shows the processor 1010, the memory 1020, the input / output interface 1030, the communication interface 1040 and the bus 1050, in the specific implementation process, the device may also include other components necessary for normal operation. In addition, it can be understood by those skilled in the art that the above device may also only include the components necessary for implementing the embodiments of the present specification, and does not necessarily include all the components shown in the figure.

[0082] The electronic device of the above embodiment is used to implement the corresponding method in any of the above embodiments, and has the beneficial effects of the corresponding method embodiment, which will not be described in detail here.

[0083] Based on the same inventive concept, corresponding to any of the above-mentioned embodiments, the present application also provides a non-transitory computer-readable storage medium, wherein the non-transitory computer-readable storage medium stores computer instructions, and the computer instructions are used to enable the computer to execute the method described in any of the above embodiments.

[0084] The computer-readable medium of this embodiment includes permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, read-only compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device.

[0085] The computer instructions stored in the storage medium of the above embodiments are used to enable the computer to execute the method described in any of the above embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0086] A person skilled in the art should understand that the discussion of any of the above embodiments is merely illustrative and is not intended to imply that the scope of the present application (including the claims) is limited to these examples. In line with the concept of the present application, the technical features in the above embodiments or different embodiments may be combined, the steps may be implemented in any order, and there are many other variations of the different aspects of the embodiments of the present application as described above, which are not provided in detail for the sake of simplicity.

[0087] In addition, to simplify the description and discussion, and in order not to make the embodiments of the present application difficult to understand, the known power / ground connections to the integrated circuit (IC) chip and other components may or may not be shown in the provided drawings. In addition, the device may be shown in the form of a block diagram to avoid making the embodiments of the present application difficult to understand, and this also takes into account the fact that the details of the implementation of these block diagram devices are highly dependent on the platform on which the embodiments of the present application are to be implemented (that is, these details should be fully within the scope of understanding of those skilled in the art). Where specific details (e.g., circuits) are set forth to describe exemplary embodiments of the present application, it is obvious to those skilled in the art that the embodiments of the present application can be implemented without these specific details or with changes in these specific details. Therefore, these descriptions should be considered illustrative rather than restrictive.

[0088] Although the present application has been described in conjunction with specific embodiments of the present application, many alternatives, modifications and variations of these embodiments will be apparent to those skilled in the art from the foregoing description. For example, other memory architectures (e.g., dynamic RAM (DRAM)) may use the discussed embodiments.

[0089] The embodiments of the present application are intended to cover all such substitutions, modifications and variations that fall within the broad scope of the appended claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the embodiments of the present application should be included in the scope of protection of the present application.

Claims

1. A method for identifying turbocharger surge, characterized in that: include: Acquire multi-channel acoustic signal data of turbocharger; Extracting features from the preprocessed acoustic signal data to obtain multi-dimensional and multi-domain composite features, and performing dimensionality reduction processing on the multi-dimensional and multi-domain composite features to form a three-dimensional sensitive feature set; Based on the constructed supercharger surge recognition model, the surge results of different acoustic signal channels are predicted respectively to obtain the surge prediction results of the acoustic signal data of each channel, wherein the supercharger surge recognition model is a bidirectional gated recurrent unit classification model based on the whale optimization algorithm and is trained by the three-dimensional sensitive feature set; The surge prediction results of each channel are fused by decision making through the improved DS evidence theory to obtain a fused surge identification result, and turbocharger surge identification is performed according to the surge identification result.

2. The method according to claim 1, characterized in that The acquired acoustic signal data is preprocessed, including: The acoustic signal data is subjected to deep denoising by using a wavelet packet threshold denoising method, and the denoised acoustic signal data is subjected to frame division and windowing processing to obtain standardized acoustic signal data.

3. The method according to claim 1, characterized in that: The preprocessed acoustic signal data is subjected to feature extraction to obtain multi-dimensional and multi-domain composite features, and the multi-dimensional and multi-domain composite features are subjected to dimensionality reduction processing to form a three-dimensional sensitive feature set, including: By extracting multiple composite features from the time domain, frequency domain, and cepstrum domain of the acoustic signal data, a multi-domain composite feature set covering 51 dimensional features is constructed; The dimension of the selected features is reduced through uniform manifold approximation and projection method to form a three-dimensional sensitive feature set.

4. The method according to claim 1, characterized in that: The bidirectional gated recurrent unit classification model is composed of two independent GRU units, each of which has two gating units, including an update gate and a reset gate, wherein the reset gate is used to concatenate the hidden state of the previous moment with the input of the current moment to calculate the candidate state, and the update gate integrates the hidden state of the previous moment with the candidate state of the current moment to obtain the current state information output value, thereby obtaining the prediction result of the GRU unit; The two GRU units process the input acoustic signal feature sample sequence from the time forward order and time reverse order respectively, and merge their respective outputs to obtain the output sequence.

5. The method according to claim 1, characterized in that The model parameters are optimized by the whale optimization algorithm to build a supercharger surge identification model based on WOA-BiGRU; The surge results of different sound signal channels are predicted respectively according to the constructed supercharger surge identification model to obtain the surge prediction results of the sound signal data of each channel.

6. The method according to claim 1, characterized in that The step of performing decision fusion on the surge prediction results of each channel by using the improved DS evidence theory to obtain a fused surge identification result, and performing turbocharger surge identification according to the surge identification result includes: The probability of surge prediction results of acoustic signal data of each channel is used as the original basic probability distribution distribution of DS evidence theory; By introducing the Jousselme distance matrix and the Deng entropy fuzzy preference matrix, the predicted state probabilities of different channels are jointly weighted. The original basic probability distribution is modified according to the calculated joint weight to obtain the revised basic probability distribution. The revised basic probability distribution is used to perform decision fusion on multiple evidence bodies through the Dempster synthesis rule to obtain the fused surge recognition result.

7. A turbocharger surge identification system, characterized in that: include: A signal acquisition module is configured to acquire multi-channel acoustic signal data of a turbocharger; A preprocessing module is configured to perform feature extraction on the preprocessed acoustic signal data to obtain multi-dimensional and multi-domain composite features, and perform dimensionality reduction processing on the multi-dimensional and multi-domain composite features to form a three-dimensional sensitive feature set; A model building module is configured to predict surge results of different acoustic signal channels respectively based on the constructed supercharger surge recognition model to obtain surge prediction results of acoustic signal data of each channel, wherein the supercharger surge recognition model is a bidirectional gated recurrent unit classification model based on a whale optimization algorithm and is trained by the three-dimensional sensitive feature set; The decision fusion module is configured to perform decision fusion on the surge prediction results of each channel through the improved DS evidence theory to obtain a fused surge identification result, and perform turbocharger surge identification according to the surge identification result.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the method according to any one of claims 1 to 6 is implemented.

9. A non-transitory computer-readable storage medium, characterized in that: in, The non-transitory computer-readable storage medium stores computer instructions, and the computer instructions are used to cause a computer to execute the method according to any one of claims 1 to 6.

Images