Cage guide anomaly detection method based on audio signal analysis

Abstract

The invention relates to the technical field of intelligent detection, in particular to a cage guide anomaly detection method based on audio signal analysis, which comprises the following steps: S1, acquiring audio signals of equipment in different running states in real time; s2, the collected audio signals are preprocessed, and audio features related to the running state of the cage guide are extracted; s3, establishing a dynamic feature change model; s4, applying the dynamic feature change model established in the S3 to analysis of real-time audio signals, and judging whether abnormity occurs or not; and S5, if the audio signal is judged to be abnormal in the step S4, carrying out classification processing on the abnormal signal through a pre-trained classification model. According to the method, the dynamic feature change model and the incremental learning algorithm are combined, real-time and accurate analysis of the audio signals of the cage guide system is achieved, and various abnormal types can be effectively recognized and automatically classified.

Description

Technical Field

[0001] The present invention relates to the field of intelligent detection technology, and in particular to a tank channel abnormality detection method based on audio signal analysis. Background Art

[0002] With the continuous advancement of industrialization, tank equipment plays a vital role in many industrial productions, especially in the fields of chemical industry, energy, metallurgy, etc. During long-term operation, tank equipment may cause abnormal conditions due to equipment aging, environmental changes or improper operation. These abnormalities are often difficult to warn in advance through traditional monitoring methods. Traditional fault detection methods usually rely on the physical indicators of the equipment or offline diagnostic methods, which makes it difficult to conduct real-time and comprehensive monitoring of the equipment's operating status. In addition, many traditional detection methods have low accuracy and cannot effectively distinguish different types of abnormalities. They can often only identify fault signals and find it difficult to provide more detailed abnormality classification and timely feedback.

[0003] In the existing technology, audio signals, as an important reflection of the operating status of equipment, have potential value in anomaly detection. By analyzing the audio signals of the equipment, the operating characteristics of the equipment can be identified and whether the equipment is in normal working condition can be monitored. However, the existing audio signal analysis methods usually adopt fixed signal processing models, which cannot adapt to the complex changes under different operating conditions of the equipment. Moreover, traditional methods mostly rely on simple signal comparison and have difficulty in effectively handling noise interference and complex timing feature changes, resulting in low accuracy of anomaly detection. In addition, the existing technology lacks an automatic classification mechanism for abnormal signals, making it difficult to accurately identify the type and cause of the anomaly, resulting in subsequent processing measures not being targeted enough and delaying the timing of handling equipment failures. Summary of the Invention

[0004] Based on the above objectives, the present invention provides a tank channel abnormality detection method based on audio signal analysis.

[0005] A tank channel anomaly detection method based on audio signal analysis includes the following steps: S1: Multiple sensors are set up around the tank channel to collect audio signals of the equipment in different operating states in real time; S2: Preprocess the collected audio signal to remove noise and interference signals from the signal, and extract audio features related to the tank road operation status through a high-pass filtering signal processing method; S3: Based on the audio features extracted in S2, a dynamic feature change model is established. The dynamic feature change model uses a combination of incremental learning and long short-term memory network to analyze the frequency and amplitude characteristics of the audio signal over time and establish the temporal change pattern of the audio signal; S4: Apply the dynamic feature change model established in S3 to analyze the real-time audio signal. By comparing the features of the real-time audio signal with the features of historical normal signals, determine whether an abnormality occurs. S5: If S4 determines that the audio signal is abnormal, the abnormal signal is classified and processed using a pre-trained classification model.

[0006] Optionally, the S1 specifically includes: S11: installing audio collection sensors at both ends and the middle of the tank channel. The audio collection sensors are microphones that can capture sound signals generated during the operation of the equipment in real time and convert the collected analog audio signals into digital signals through an analog-to-digital converter; S12: Set the sampling frequency of the audio acquisition sensor to collect at least 1000 data points per second.

[0007] Optionally, the S2 specifically includes: S21: performing noise filtering on the collected audio signal, specifically performing time-frequency analysis on the audio signal through short-time Fourier transform, decomposing the signal into several frequency bands, and identifying and removing low-frequency noise and environmental background noise; S22: Use a high-pass filter to attenuate the low-frequency portion in the frequency domain. Set the cutoff frequency of the high-pass filter to 300 Hz to remove low-frequency noise and irrelevant signal components, retaining only the signal in the frequency range related to the tankway operating status. S23: performing time domain smoothing on the signal processed by the high-pass filter, and smoothing the amplitude of the audio signal using a sliding average method to reduce fluctuations caused by instantaneous changes; S24: extracting audio features related to the tank channel operation status by calculating the frequency spectrum and time domain features of the preprocessed audio signal, including spectrum peaks, frequency response curves, and time domain standard deviations.

[0008] Optionally, the S24 specifically includes: S241: Calculate the frequency spectrum of the preprocessed audio signal by applying a fast Fourier transform to the audio signal to obtain its frequency spectrum; S242: Extracting the peak value in the spectrum. First, calculate the amplitude of the spectrum. Then, use the peak detection algorithm to extract the maximum amplitude value and the corresponding frequency in the spectrum. Then, record the frequency and amplitude of the spectrum peak. S243: Calculate a frequency response curve, where the frequency response curve represents the amplitude response of the audio signal at different frequencies. The frequency response curve is obtained by plotting a curve showing how the amplitude of the spectrum changes with frequency. S244: Calculate the standard deviation of the audio signal to evaluate the amplitude variation of the signal.

[0009] Optionally, the S3 specifically includes: S31: Based on the audio features extracted in S2, including spectrum peaks, frequency response curves, and time domain standard deviations, an audio feature dataset is constructed, where each data sample consists of a timestamp and a corresponding audio feature vector; S32: Use an incremental learning algorithm to train the audio feature dataset and gradually update the model parameters to adapt it to the new audio feature data; S33: Applying long short-term memory networks to audio feature data to capture temporal changes in audio signals and learn how audio signals change over time; S34: Combining incremental learning and long short-term memory networks, we first use incremental learning to train preliminary data, and then use long short-term memory networks to optimize the model to accurately capture the temporal characteristics of audio signals. S35: Continuously update the long short-term memory network model through incremental learning, enabling it to adapt to new audio feature data in real time and capture dynamic changes in audio signals; S36: Through continuous updating and optimization, the dynamic feature change model is established to ensure that the model can analyze the dynamic feature changes based on the real-time audio signal.

[0010] Optionally, the S32 specifically includes: S321: Input the audio feature dataset constructed in S31 into the incremental learning model and initialize the model parameters. The objective function used by the incremental learning model is: ,in, are model parameters, is the loss function, is the actual label, is the model output, is the input feature, is the sample size; S322: Select the corresponding learning rate according to the incremental learning strategy and the incremental update step size , and update the model parameters using the following formula: ,in, are the updated model parameters, are the current model parameters, is the learning rate, is the gradient of the current model parameters; S323: In each incremental training process, new audio feature data is used and its corresponding label , is updated incrementally using the following formula: ,in, Indicates the current parameters In the new input data The gradient on S324: Gradually adjust the model parameters by iterating the above incremental learning process , enabling the model to adapt to new audio feature data.

[0011] Optionally, the S33 specifically includes: S331: Input the audio feature data updated by incremental learning in S32 into the long short-term memory network to capture the temporal changes of the audio signal; S332: In the long short-term memory network, the hidden state and memory unit state of each time step are calculated step by step using the control mechanism of the input gate, forget gate, candidate memory unit and output gate; S333: Learning the temporal changes of audio signals through the hidden state sequence of long short-term memory networks; S334: After training is completed, the final hidden state of the long short-term memory network is used as the temporal feature of the audio signal; S335: Optimize the weight parameters in the long short-term memory network through the back-propagation algorithm, so that the network can better fit the temporal variation of the audio signal.

[0012] Optionally, the S36 specifically includes: S361: Continuously update various parameters in the dynamic feature change model based on the training results of incremental learning and long short-term memory networks; S362: Calculating the prediction error during each training process and adjusting the model weights using an optimization algorithm to minimize the model's prediction error. S363: Based on the incremental learning method and iterative updates of the long short-term memory network, the structure of the dynamic feature change model is gradually improved; S364: When the error of the model is lower than a preset threshold, it indicates that the dynamic feature change model has been optimized.

[0013] Optionally, the S4 specifically includes: The S4 specifically includes: S41: Inputting the audio features of the real-time collected audio signal processed in step S2 into the dynamic feature change model to obtain the dynamic features of the real-time audio signal; S42: Calculate the difference between the dynamic features of the real-time audio signal and the dynamic features of the historical normal signal, and measure the difference by calculating the Euclidean distance, which is: ,in, represents the difference measure, Represents the real-time audio signal eigenvalues, Indicates the historical normal audio signal eigenvalues, The dimension representing the audio features; S43: Based on the calculated difference metric Compare with the set threshold, if If it is greater than the threshold, it is judged as an abnormal signal.

[0014] Optionally, the S5 specifically includes: S51: If S4 determines that the audio signal is abnormal, the feature vector of the abnormal signal is input into the pre-trained classification model for classification; S52: The classification model receives the input abnormal signal features and outputs a category label based on the patterns learned during training. , this label indicates the type of abnormal signal; when Indicates a device failure. Indicates external interference, Indicates abnormal operation; S53: Different processing strategies are adopted according to the category labels obtained in S52. For abnormal signals of equipment failure type, a maintenance notice will be sent; for abnormal signals of external interference type, the signal will be enhanced or shielded; for abnormal signals of operation abnormality type, a real-time prompt or warning will be sent to the operator, and the operator will be informed of the existing operation errors.

[0015] Beneficial effects of the present invention: The present invention, by combining a dynamic feature change model with incremental learning technology, can capture the temporal changes of audio signals in real time, identify and classify various types of abnormal signals. Compared with traditional anomaly detection methods, this solution has higher accuracy and adaptability, can effectively eliminate the interference of environmental noise, and accurately reflect the actual operating status of the tank channel system.

[0016] The present invention uses an incremental learning algorithm to enable the model to be continuously updated and adapt to new operating conditions and audio features. This adaptive feature enables the system to maintain efficient anomaly detection capabilities in complex and changing environments, reduces the need for human intervention, and improves the stability and safety of the tank channel system. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0018] Figure 1Schematic diagram of a tank channel abnormality detection method according to an embodiment of the present invention; Figure 2 Schematic diagram of the process of establishing a dynamic feature change model according to an embodiment of the present invention. DETAILED DESCRIPTION

[0019] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments. It is also noted that, to provide a more detailed description, the following embodiments are best and preferred embodiments, and those skilled in the art may employ alternative methods for implementing certain known technologies. Furthermore, the accompanying drawings are intended only to provide a more detailed description of the embodiments and are not intended to limit the present invention.

[0020] It should be noted that references in the specification to "one embodiment," "an embodiment," "exemplary embodiments," "some embodiments," etc. indicate that the described embodiments may include specific features, structures, or characteristics, but not necessarily every embodiment will include such specific features, structures, or characteristics. Furthermore, when specific features, structures, or characteristics are described in conjunction with an embodiment, it is within the knowledge of persons skilled in the relevant art to implement such features, structures, or characteristics in conjunction with other embodiments (whether or not explicitly described).

[0021] In general, terms can be understood, at least in part, from their use in context. For example, depending at least in part on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in the singular sense, or can be used to describe a combination of features, structures, or characteristics in the plural sense. Additionally, the term "based on" can be understood as not necessarily intended to convey an exclusive set of factors, but can instead, depending at least in part on the context, allow for the presence of other factors that are not necessarily explicitly described.

[0022] like Figure 1-Figure 2 As shown, a tank channel abnormality detection method based on audio signal analysis includes the following steps: S1: Multiple sensors are set up around the tank channel to collect audio signals of the equipment in different operating states in real time; S2: Preprocess the collected audio signal to remove noise and interference signals from the signal, and extract audio features related to the tank road operation status through a high-pass filtering signal processing method; S3: Based on the audio features extracted in S2, a dynamic feature change model is established. The dynamic feature change model uses a combination of incremental learning and long short-term memory (LSTM) networks to analyze the frequency and amplitude characteristics of the audio signal over time and establish the temporal change pattern of the audio signal. S4: Apply the dynamic feature change model established in S3 to analyze the real-time audio signal. By comparing the features of the real-time audio signal with the features of historical normal signals, determine whether an abnormality occurs. S5: If S4 determines that the audio signal is abnormal, the abnormal signal is classified and processed using a pre-trained classification model.

[0023] S1 specifically includes: S11: Audio collection sensors are installed at both ends and the middle of the tank channel. The audio collection sensors are microphones, which are used to capture the sound signals generated during the operation of the equipment in real time and convert the collected analog audio signals into digital signals through an analog-to-digital converter; S12: Set the sampling frequency of the audio acquisition sensor to collect at least 1000 data points per second to ensure sufficient resolution and accuracy for subsequent analysis and provide accurate data support for subsequent anomaly detection.

[0024] S2 specifically includes: S21: Filter the collected audio signal for noise. Specifically, perform time-frequency analysis on the audio signal through short-time Fourier transform (STFT), decompose the signal into several frequency bands, and identify and remove low-frequency noise and environmental background noise. The formula for short-time Fourier transform is: ,in, For time The audio signal at the moment, is the window function, is the frequency, is the time variable, For in time and frequency The complex spectrum at ; S22: Use a high-pass filter to attenuate the low-frequency portion in the frequency domain. Set the cutoff frequency of the high-pass filter to 300 Hz to remove low-frequency noise and irrelevant signal components, retaining only the signal in the frequency range related to the tankway operating status. S23: Performing time domain smoothing on the high-pass filtered signal, using a sliding average method to smooth the amplitude of the audio signal to reduce fluctuations caused by instantaneous changes and ensure signal stability and accuracy; S24: By calculating the frequency spectrum and time domain features of the preprocessed audio signal, audio features related to the tankway operating status are extracted, including spectrum peaks, frequency response curves, and time domain standard deviations, which serve as input features for subsequent modeling. By comprehensively applying time-frequency analysis, noise filtering, and high-pass filtering, low-frequency noise and environmental interference signals in the audio signal can be effectively removed, ensuring that the signal retains valid features closely related to the tankway operating status. This process provides clean and stable signal data for subsequent feature extraction and anomaly detection.

[0025] S24 specifically includes: S241: Calculate the frequency spectrum of the pre-processed audio signal by applying fast Fourier transform to the audio signal to obtain its frequency spectrum , the formula is: ,in, For time The audio signal at the moment, is the total number of sampling points of the signal, is the frequency, For the signal at frequency The amplitude at S242: Extract the peak value in the spectrum. First, calculate the amplitude of the spectrum. Then, use the peak detection algorithm to extract the maximum amplitude value and the corresponding frequency in the spectrum. Then, record the frequency and amplitude of the spectrum peak. The formula is: ; ,in, is the frequency of the spectrum peak, is the amplitude of the spectrum peak, is the amplitude of the spectrum; S243: Calculate a frequency response curve. The frequency response curve represents the amplitude response of the audio signal at different frequencies. The frequency response curve is obtained by plotting a curve showing how the amplitude of the spectrum changes with frequency. The formula is as follows: ,in, The frequency response curve is at frequency The value at the ; S244: Calculate the standard deviation of the audio signal to evaluate the amplitude change of the signal. The formula is: ,in, is the mean value of the audio signal, is the total number of sampling points of the signal, is the standard deviation of the audio signal, reflecting the amplitude change of the signal; by calculating the frequency spectrum of the audio signal through fast Fourier transform (FFT), the spectrum peak and frequency response curve can be effectively extracted, thereby reflecting the frequency characteristics of the audio signal.

[0026] S3 specifically includes: S31: Based on the audio features extracted in S2, including spectrum peaks, frequency response curves, and time domain standard deviations, an audio feature dataset is constructed, where each data sample consists of a timestamp and a corresponding audio feature vector, which serves as input for subsequent modeling. S32: Use an incremental learning algorithm to train the audio feature dataset and gradually update the model parameters to adapt it to the new audio feature data; S33: Applying long short-term memory (LSTM) networks to audio feature data to capture temporal changes in audio signals and learn how audio signals change over time. S34: Combining incremental learning and long short-term memory networks, we first use incremental learning to train preliminary data, and then use long short-term memory networks to optimize the model to accurately capture the temporal characteristics of audio signals. S35: Continuously update the long short-term memory network model through incremental learning, enabling it to adapt to new audio feature data in real time and capture dynamic changes in audio signals; S36: Through continuous updating and optimization, the dynamic feature change model is established to ensure that the model can analyze the dynamic feature changes of the real-time audio signal, thereby determining whether there are abnormal changes in the audio signal.

[0027] S32 specifically includes: S321: Input the audio feature dataset constructed in S31 into the incremental learning model and initialize the model parameters. The objective function used by the incremental learning model is: ,in, are model parameters, is the loss function, is the actual label, is the model output, is the input feature, is the sample size; S322: Select the corresponding learning rate according to the incremental learning strategy and the incremental update step size , and update the model parameters using the following formula: ,in, are the updated model parameters, are the current model parameters, is the learning rate, is the gradient of the current model parameters; S323: In each incremental training process, new audio feature data is used and its corresponding label , is updated incrementally using the following formula: ,in, Indicates the current parameters In the new input data The gradient on S324: Gradually adjust the model parameters by iterating the above incremental learning process , so that the model can adapt to new audio feature data and continuously optimize the accuracy and generalization ability of the model; by gradually updating the model parameters so that it can adapt to new audio data in real time, the adaptability of the model in a dynamic environment is guaranteed. By optimizing the update mechanism of the model, the accuracy and generalization ability of the model can be effectively improved, and the adaptability of the system to audio signals under different operating conditions can be enhanced, thereby improving the accuracy and stability of tank channel anomaly detection.

[0028] S33 specifically includes: S331: Input the audio feature data updated by incremental learning in S32 into the long short-term memory network to capture the temporal changes of the audio signal; S332: In the long short-term memory network, the hidden state and memory unit state of each time step are calculated step by step using the control mechanism of the input gate, forget gate, candidate memory unit and output gate; S333: Learning the temporal changes of audio signals through the hidden state sequence of long short-term memory networks; S334: After training is completed, the final hidden state of the long short-term memory network is used as the temporal feature of the audio signal; S335: Optimize the weight parameters in the long short-term memory network through the back-propagation algorithm, so that the network can better fit the temporal variation of the audio signal.

[0029] The steps to use the long short-term memory network to learn the change pattern are as follows: First, the audio feature data updated by incremental learning in S32 is input into the long short-term memory network to capture the temporal changes of the audio signal. The input of the LSTM model is the time series of the audio feature dataset. ,in, Indicates the The audio feature vector of time steps, Indicates the length of the time series; Then, in the LSTM network, the hidden state at each time step is calculated based on the following formula and memory cell status : ; ; ; ; ; ;in, is the activation value of the input gate; is the activation value of the forget gate; is a candidate memory unit; is the memory unit state at the current moment; is the activation value of the output gate; is the hidden state at the current moment; is the LSTM weight matrix, is the bias term; represents the sigmoid function, anh represents the hyperbolic tangent activation function; Next, through the hidden state of the LSTM network Learning the temporal changes of audio signals over time, each time step in the network will update the long-term and short-term dependency information of the audio signal to capture the potential temporal patterns in the signal; and after the training is completed, the final hidden state of the LSTM network will be updated. As the time series feature vector of the audio signal, it serves as the input for subsequent model analysis and prediction; Finally, the back-propagation algorithm is used to optimize the weight parameters of the LSTM network. By minimizing the loss function, the model's ability to learn the temporal changes of the audio signal is gradually improved. The loss function usually uses the mean square error (MSE) or cross entropy loss function, and the formula is: ,in, is the actual label, is the predicted value of the model, is the sample size.

[0030] S36 specifically includes: S361: Based on the training results of incremental learning and long short-term memory networks, the parameters of the dynamic feature change model are continuously updated. Specifically, each time new audio signal feature data is input, the model performs comprehensive learning based on the current audio feature vector and historical audio features, updating the weights and biases in the model to optimize the model's adaptability to changes in temporal features. S362: By calculating the prediction error during each training process, the model weights are adjusted using an optimization algorithm (such as gradient descent) to minimize the model's prediction error. Specifically, through the backpropagation algorithm, the model adjusts all network parameters based on the error between the actual observed value and the predicted value, allowing the model to more accurately fit the new audio feature data. S363: Using incremental learning methods and iterative updates of the long-short-term memory network, the structure of the dynamic feature change model is gradually improved, enabling it to more stably capture long-term and short-term changes in audio signals. After each update, the dynamic feature change model optimizes the model based on the new audio signal, thereby enhancing its dynamic response to the tank channel operating status. S364: When the model error is lower than the preset threshold, it indicates that the dynamic feature change model has been optimized and can be further applied. The final expression of the dynamic feature change model is: ,in, Is the input audio feature vector, indicating the time Audio signal characteristics; For the moment The updated model parameters include all parameters of the LSTM network (such as weight matrix, bias, etc.); For the moment The model output is the predicted value. The above steps continuously update and optimize the parameters of the dynamic feature change model to make the model more adaptable and accurate, and able to cope with changes in audio signals under different operating states. The model can gradually adapt to new audio feature data through incremental learning and use the powerful timing learning ability of the LSTM network to accurately capture the timing characteristics of audio signals. Ultimately, the optimized dynamic feature change model can provide more accurate anomaly detection in practical applications and improve the robustness and responsiveness of the system.

[0031] S4 specifically includes: S41: Inputting the audio features of the real-time collected audio signal processed in step S2 into the dynamic feature change model to obtain the dynamic features of the real-time audio signal; S42: Calculate the difference between the dynamic features of the real-time audio signal and the dynamic features of the historical normal signal, and measure the difference by calculating the Euclidean distance, which is: ,in, represents the difference measure, Represents the real-time audio signal eigenvalues, Indicates the historical normal audio signal eigenvalues, The dimension representing the audio features; S43: Based on the calculated difference metric Compare with the set threshold, if If the value is greater than the threshold, it is determined to be an abnormal signal. By applying the dynamic feature change model to real-time audio signal analysis, it is possible to accurately determine whether an abnormality occurs in real time, and use the Euclidean distance metric to accurately measure the difference between the real-time signal and the historical normal signal, thereby effectively improving the accuracy and response speed of anomaly detection.

[0032] S5 specifically includes: S51: If S4 determines that the audio signal is abnormal, the feature vector of the abnormal signal is input into a pre-trained classification model for classification. The classification model is trained based on a large amount of labeled abnormal signal data and can automatically determine the type of abnormal signal based on the audio characteristics (including spectral characteristics, time domain characteristics, etc.) of the abnormal signal. S52: The classification model receives the input abnormal signal features and outputs a category label based on the patterns learned during training. , this label indicates the type of abnormal signal; when Indicates a device failure. Indicates external interference, Indicates abnormal operation; S53: Different processing strategies are adopted according to the category labels obtained in S52. For abnormal signals of equipment failure type, a maintenance notification will be sent; for abnormal signals of external interference type, the signal will be enhanced or shielded; for abnormal signals of operation abnormality type, a real-time prompt or warning will be sent to the operator, and the operator will be informed of the existing operation errors; the above steps can quickly and accurately classify and process abnormal signals by using pre-trained classification models, effectively distinguish different types of abnormal signals, and thus take targeted countermeasures, thereby improving the detection accuracy and response speed of the system and reducing the need for manual intervention.

[0033] The present invention encompasses any alternatives, modifications, equivalents, and solutions that fall within the spirit and scope of the present invention. To provide a thorough understanding of the present invention, specific details are described in detail below in connection with the preferred embodiments of the present invention, but those skilled in the art will be able to fully understand the present invention without these detailed descriptions. Furthermore, to avoid unnecessary confusion regarding the essence of the present invention, well-known methods, processes, procedures, components, and circuits have not been described in detail.

[0034] 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 principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. A tank channel anomaly detection method based on audio signal analysis, characterized in that: The following steps are involved: S1: Multiple sensors are set up around the tank channel to collect audio signals of the equipment in different operating states in real time; S2: Preprocess the collected audio signal to remove noise and interference signals from the signal, and extract audio features related to the tank road operation status through a high-pass filtering signal processing method; S3: Based on the audio features extracted in S2, a dynamic feature change model is established. The dynamic feature change model uses a combination of incremental learning and long short-term memory network to analyze the frequency and amplitude characteristics of the audio signal over time and establish the temporal change pattern of the audio signal; S4: Apply the dynamic feature change model established in S3 to analyze the real-time audio signal. By comparing the features of the real-time audio signal with the features of historical normal signals, determine whether an abnormality occurs. S5: If S4 determines that the audio signal is abnormal, the abnormal signal is classified and processed using a pre-trained classification model.

2. The tank channel abnormality detection method based on audio signal analysis according to claim 1 is characterized in that: Said S1 specifically includes: S11: installing audio collection sensors at both ends and the middle of the tank channel. The audio collection sensors are microphones that can capture sound signals generated during the operation of the equipment in real time and convert the collected analog audio signals into digital signals through an analog-to-digital converter; S12: Set the sampling frequency of the audio acquisition sensor to collect at least 1000 data points per second.

3. The tank channel abnormality detection method based on audio signal analysis according to claim 2 is characterized in that: The S2 specifically includes: S21: performing noise filtering on the collected audio signal, specifically performing time-frequency analysis on the audio signal through short-time Fourier transform, decomposing the signal into several frequency bands, and identifying and removing low-frequency noise and environmental background noise; S22: Use a high-pass filter to attenuate the low-frequency portion in the frequency domain. Set the cutoff frequency of the high-pass filter to 300 Hz to remove low-frequency noise and irrelevant signal components, retaining only the signal in the frequency range related to the tankway operating status. S23: performing time domain smoothing on the signal processed by the high-pass filter, and smoothing the amplitude of the audio signal using a sliding average method to reduce fluctuations caused by instantaneous changes; S24: extracting audio features related to the tank channel operation status by calculating the frequency spectrum and time domain features of the preprocessed audio signal, including spectrum peaks, frequency response curves, and time domain standard deviations.

4. The method for detecting tank channel anomalies based on audio signal analysis according to claim 3 is characterized in that: The S24 specifically includes: S241: Calculate the frequency spectrum of the preprocessed audio signal by applying a fast Fourier transform to the audio signal to obtain its frequency spectrum; S242: Extracting the peak value in the spectrum. First, calculate the amplitude of the spectrum. Then, use the peak detection algorithm to extract the maximum amplitude value and the corresponding frequency in the spectrum. Then, record the frequency and amplitude of the spectrum peak. S243: Calculate a frequency response curve, where the frequency response curve represents the amplitude response of the audio signal at different frequencies. The frequency response curve is obtained by plotting a curve showing how the amplitude of the spectrum changes with frequency. S244: Calculate the standard deviation of the audio signal to evaluate the amplitude variation of the signal.

5. The method for detecting tank channel anomalies based on audio signal analysis according to claim 1 is characterized in that: The S3 specifically includes: S31: Based on the audio features extracted in S2, including spectrum peaks, frequency response curves, and time domain standard deviations, an audio feature dataset is constructed, where each data sample consists of a timestamp and a corresponding audio feature vector; S32: Use an incremental learning algorithm to train the audio feature dataset and gradually update the model parameters to adapt it to the new audio feature data; S33: Applying long short-term memory networks to audio feature data to capture temporal changes in audio signals and learn how audio signals change over time; S34: Combining incremental learning and long short-term memory networks, we first use incremental learning to train preliminary data, and then use long short-term memory networks to optimize the model to accurately capture the temporal characteristics of audio signals. S35: Continuously update the long short-term memory network model through incremental learning, enabling it to adapt to new audio feature data in real time and capture dynamic changes in audio signals; S36: Through continuous updating and optimization, the dynamic feature change model is established to ensure that the model can analyze the dynamic feature changes based on the real-time audio signal.

6. The method for detecting tank channel anomalies based on audio signal analysis according to claim 5, characterized in that: The S32 specifically includes: S321: Input the audio feature dataset constructed in S31 into the incremental learning model and initialize the model parameters. The objective function used by the incremental learning model is: ,in, are model parameters, is the loss function, is the actual label, is the model output, is the input feature, is the sample size; S322: Select the corresponding learning rate according to the incremental learning strategy and the incremental update step size , and update the model parameters using the following formula: ,in, are the updated model parameters, are the current model parameters, is the learning rate, is the gradient of the current model parameters; S323: In each incremental training process, new audio feature data is used and its corresponding label , is updated incrementally using the following formula: ,in, Indicates the current parameters In the new input data The gradient on S324: Gradually adjust the model parameters by iterating the above incremental learning process , enabling the model to adapt to new audio feature data.

7. The method for detecting tank channel anomalies based on audio signal analysis according to claim 6, characterized in that: The S33 specifically includes: S331: Input the audio feature data updated by incremental learning in S32 into the long short-term memory network to capture the temporal changes of the audio signal; S332: In the long short-term memory network, the hidden state and memory unit state of each time step are calculated step by step using the control mechanism of the input gate, forget gate, candidate memory unit and output gate; S333: Learning the temporal changes of audio signals through the hidden state sequence of long short-term memory networks; S334: After training is completed, the final hidden state of the long short-term memory network is used as the temporal feature of the audio signal; S335: Optimize the weight parameters in the long short-term memory network through the back-propagation algorithm, so that the network can better fit the temporal variation of the audio signal.

8. The method for detecting tank channel anomalies based on audio signal analysis according to claim 7, characterized in that: The S36 specifically includes: S361: Continuously update various parameters in the dynamic feature change model based on the training results of incremental learning and long short-term memory networks; S362: Calculating the prediction error during each training process and adjusting the model weights using an optimization algorithm to minimize the model's prediction error. S363: Based on the incremental learning method and iterative updates of the long short-term memory network, the structure of the dynamic feature change model is gradually improved; S364: When the error of the model is lower than a preset threshold, it indicates that the dynamic feature change model has been optimized.

9. The method for detecting tank channel anomalies based on audio signal analysis according to claim 1, characterized in that: The S4 specifically includes: S41: Inputting the audio features of the real-time collected audio signal processed in step S2 into the dynamic feature change model to obtain the dynamic features of the real-time audio signal; S42: Calculate the difference between the dynamic features of the real-time audio signal and the dynamic features of the historical normal signal, and measure the difference by calculating the Euclidean distance, which is: ,in, represents the difference measure, Represents the real-time audio signal eigenvalues, Indicates the historical normal audio signal eigenvalues, The dimension representing the audio features; S43: Based on the calculated difference metric Compare with the set threshold, if If it is greater than the threshold, it is judged as an abnormal signal.

10. The method for detecting tank channel anomalies based on audio signal analysis according to claim 1, characterized in that: The S5 specifically includes: S51: If S4 determines that the audio signal is abnormal, the feature vector of the abnormal signal is input into the pre-trained classification model for classification; S52: The classification model receives the input abnormal signal features and outputs a category label based on the patterns learned during training. , this label indicates the type of abnormal signal; when Indicates a device failure. Indicates external interference, Indicates abnormal operation; S53: Different processing strategies are adopted according to the category labels obtained in S52. For abnormal signals of equipment failure type, a maintenance notice will be sent; for abnormal signals of external interference type, the signal will be enhanced or shielded; for abnormal signals of operation abnormality type, a real-time prompt or warning will be sent to the operator, and the operator will be informed of the existing operation errors.

Images