Signal processing method and system based on deep neural networks

Abstract

This application discloses a signal processing method and system based on a deep neural network, relating to the technical field of signal processing. The signal processing method includes collecting raw signal data, performing initial processing on the signal data including filtering and denoising, extracting key features of the processed signal, and converting the signal into a format suitable for deep learning processing. This application integrates signal processing technology and deep learning methods to handle complex nonlinear relationships, is applicable to various signal processing tasks, improves model performance and accuracy with large amounts of data and computational resources, reduces the burden of manual feature design, and better captures important information in signals. The deep neural network has powerful nonlinear modeling capabilities, can handle complex nonlinear relationships, is applicable to various signal processing tasks, can model and process complex signals more accurately than conventional methods, has broad applicability, and can play a role in various fields and applications.

Description

Technical Field

[0001] This application claims the priority of a Chinese patent application filed with the China National Intellectual Property Administration on April 16, 2024, with the application number 202410454369.3 and the title of the invention being "Signal Processing Method Based on Deep Neural Network", and the entire content thereof is incorporated herein by reference.

[0002] This application relates to the technical field of signal processing, and specifically, to a signal processing method and system based on deep neural networks.

Background Art

[0003] A deep neural network is generally an artificial neural network composed of multiple layers including an input layer, a hidden layer, and an output layer. Each layer contains a plurality of neurons, there are connections between the neurons of adjacent layers, each connection has a weight, and through training data, the weights in the network are adjusted so that the network can learn the feature representation of the input data and perform corresponding classification, identification, or prediction tasks.

[0004] Signal processing refers to processes such as acquiring, converting, transmitting, storing, and interpreting signals, and these signals may be data from various sources such as sensors, communication systems, and within the body. The purpose of signal processing is to extract useful information from these data and analyze and interpret it.

[0005]

[0006] Deep neural networks can be used for tasks such as feature extraction, classification, and regression in signal processing. Deep learning models can be used for tasks such as speech recognition of speech signals, target detection of image signals, and disease diagnosis of biomedical signals. Deep learning has demonstrated excellent performance in the field of signal processing due to its powerful feature extraction ability and non-linear modeling ability.Generally, related signal processing methods typically require the manual design of feature extractors and are highly dependent on domain knowledge and experience, making them unsuitable for complex signal scenarios. They may perform poorly when processing nonlinear or high-dimensional data, failing to adequately mine the latent information in the data. For some complex signal processing tasks, related signal processing methods may require significant manual and time investment for parameter tuning and algorithm optimization, while simultaneously having poor generalization capabilities, potentially resulting in poor generalization performance for new signal types or datasets with large variations.

[0007] As described above, in order to solve the above problems, it is necessary to provide a signal processing method based on deep neural networks. [Overview of the project] [Problems that the invention aims to solve]

[0008] The purpose of this application is to provide a signal processing method and system based on a deep neural network in order to solve the problems raised in the background art described above. [Means for solving the problem]

[0009] To achieve the above objective, this application provides the following technical solutions.

[0010] A signal processing method based on a deep neural network, wherein the signal processing method is implemented based on a signal processing system. Step S1 involves collecting raw signal data, Step S2 involves performing initial processing on the signal data, including filtering and noise reduction. Step S3 involves extracting key features from the processed signal, Step S4 involves converting the signal into a format suitable for deep learning processing, Step S5 involves classifying, identifying, or predicting a signal using a deep neural network, Step S6 involves training a deep learning model and optimizing its performance, Step S7 involves analyzing the DSP code and ensuring its accuracy and efficiency. Step S8 involves analyzing power supply noise, designing filters, and optimizing power supply quality. Step S9 involves monitoring the power supply and system temperature to ensure a stable power supply and avoid overheating. Step S10 converts the processing result into an appropriate output format, Step S11 controls the output of the result, including display, storage, or transmission. Step S12 provides a user interface that allows parameter setting and viewing of results, The system includes step S13, which monitors the overall system status and ensures stable and highly efficient operation.

[0011] A signal processing system based on a deep neural network, the signal processing system including a data acquisition module, a signal processing module, a machine learning module, a DSP code audit module, a power filtering optimization module, a power supply and thermal problem detection module, an output processing module, and a system management module. The data acquisition module is positioned to acquire and preprocess signals. The signal processing module is positioned to perform feature extraction and signal transformation. The machine learning module is configured to perform deep learning and model training. The DSP code audit module is configured to perform code checks and performance evaluations. The power filtering optimization module is positioned to perform power noise analysis and filter design. The power supply and thermal problem detection module is positioned to monitor the power supply and temperature. The output processing module is positioned to perform data format conversion and output control. The system management module is positioned to provide the user interface and monitor the system.

[0012] In an exemplary embodiment, the data acquisition module further includes a signal acquisition unit and a preprocessing unit. The signal acquisition unit is positioned to acquire the original signal from the sensor or other data source. The pre-processing unit is positioned to perform initial processing on the original signal, including filtering and denoising, in order to improve data quality.

[0013] In an exemplary embodiment, the signal processing module further includes a feature extraction unit and a signal conversion unit. The feature extraction unit is positioned to extract key features, including frequency and amplitude, from the preprocessed signal. The signal conversion unit is positioned to convert signals into a format suitable for deep learning model processing (including time-frequency domain conversion).

[0014] In an exemplary embodiment, the machine learning module further includes a deep learning unit and a model training unit. The deep learning unit is configured to perform signal classification, identification, or prediction tasks using deep neural networks. Model training units are configured to train and optimize deep learning models using training datasets.

[0015] In an exemplary embodiment, the DSP code audit module further includes a code check unit and a performance evaluation unit. The code check unit is configured to perform static analysis on DSP (Digital Signal Processing) codes and check for potential errors and non-standard codes. The performance evaluation unit is positioned to assess the execution efficiency and resource consumption of the DSP code and ensure that performance requirements are met.

[0016] In an exemplary embodiment, the power filter optimization module further includes a power noise analysis unit and a filter design unit, The power noise analysis unit is arranged to analyze the noise in the power line and determine the filtering needs, The filter design unit is arranged to design an appropriate filter to reduce the impact of power noise on signal processing.

[0017] In an exemplary embodiment, the power supply and heat problem detection module further includes a power supply monitoring unit and a temperature monitoring unit, The power supply monitoring unit is arranged to monitor the voltage and current of the power supply in real time and detect abnormalities in the power supply, The temperature monitoring unit is arranged to monitor the temperature of the system and prevent performance degradation and damage due to overheating.

[0018] In an exemplary embodiment, the output processing module further includes a data format conversion unit and an output control unit, The data format conversion unit is arranged to convert the processing result into a format suitable for output or display, The output control unit is arranged to control the output method of the result including display, storage or transmission of the display.

[0019] In an exemplary embodiment, the system management module further includes a user interface unit and a system monitoring unit, The user interface unit is arranged to provide a user interaction interface that allows the user to set parameters and view the results, The system monitoring unit is arranged to monitor the state of the system including the state of the hardware and the operating state of the software, and ensure the stable operation of the system.

Advantages of the Invention

[0020] Compared to conventional technologies, the beneficial effects of this application are as follows: This application integrates signal processing technology and deep learning methods, fully utilizing the advantages of both. It can automatically learn high-level feature representations of signals, possesses strong generalization capabilities, can handle complex nonlinear relationships, and is applicable to various signal processing tasks. With large amounts of data and computational resources, it improves the performance and accuracy of the model, while simultaneously reducing the burden of manual feature design and better capturing important information in signals. The deep neural network has strong nonlinear modeling capabilities, can handle complex nonlinear relationships, and is applicable to various signal processing tasks. Compared to conventional methods, it can model and process complex signals more accurately, has broad applicability, and can play a role in various fields and applications. [Brief explanation of the drawing]

[0021] [Figure 1] The topology diagram of the signal processing system of this application is shown. [Figure 2] This is a flowchart of the signal processing method based on a deep neural network as described in this application. [Modes for carrying out the invention]

[0022] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the drawings of the embodiments of this application, and it is clear that the embodiments described are only a selection of embodiments of this application, not all embodiments. All other embodiments obtained by a person skilled in the art without creative work based on the embodiments of this application are within the technical scope of this application.

[0023] Example 1 As shown in Figure 1, a signal processing system is established that includes a data acquisition module, a signal processing module, a machine learning module, a DSP code audit module, a power filtering optimization module, a power supply and thermal problem detection module, an output processing module, and a system management module. Here, the data acquisition module collects and preprocesses signals, the signal processing module performs feature extraction and signal transformation, the machine learning module performs deep learning and model training, the DSP code audit module performs code checking and performance evaluation, the power supply filtering optimization module performs power supply noise analysis and filter design, the power supply and thermal problem detection module monitors the power supply and temperature, the output processing module performs data format conversion and output control, and the system management module provides a user interface and monitors the system.

[0024] The signal acquisition unit collects the original signal from the sensor or other data source (e.g., a heart sound signal sensor or some one-dimensional signal data source), and the preprocessing unit performs initial processing on the original signal, including filtering and noise reduction, to improve data quality.

[0025] Furthermore, the feature extraction unit extracts key features, including frequency and amplitude, from the pre-processed signal, and the signal conversion unit converts the signal into a format suitable for deep learning model processing (including time-frequency domain conversion).

[0026] The deep learning unit uses deep neural networks to perform signal classification, identification, or prediction tasks, while the model training unit trains and optimizes deep learning models using training datasets.

[0027] Furthermore, a code check unit performs static analysis on DSP (Digital Signal Processing) code to check for potential errors and non-standard code, and a performance evaluation unit evaluates the execution efficiency and resource consumption of the DSP code to ensure that performance requirements are met.

[0028] Furthermore, the power supply noise analysis unit analyzes noise in the power lines and determines the filtering needs, and the filter design unit designs appropriate filters to reduce the impact of power supply noise on signal processing.

[0029] Furthermore, a power supply monitoring unit monitors the power supply voltage and current in real time to detect power supply abnormalities, and a temperature monitoring unit monitors the system temperature to prevent performance degradation or damage due to overheating.

[0030] Furthermore, a data format conversion unit converts the processing results into a format suitable for output or display, and an output control unit controls the output method of the results, including display on a display, storage, or transmission.

[0031] Furthermore, the user interface unit provides a user interaction interface that allows the user to set parameters and view results, while the system monitoring unit monitors the system status, including the hardware status and software operating status, to ensure stable system operation. For example, if the original signal acquired is from a heart sound signal sensor, the system monitoring unit monitors the status of the heart sound signal analysis system where the heart sound signal sensor is located, ensuring stable operation of the heart sound signal analysis system and improving the accuracy of subsequent heart sound signal analysis results.

[0032] Example 2 As shown in Figure 2, in actual applications, the signal processing method based on the above information processing system specifically includes the following steps.

[0033] S1. Collect raw signal data. S1.1. Configure the sensor or data source and start the data acquisition task. S1.2. Establish a data collection channel using data collection software. S1.3. Set data acquisition parameters, including sampling rate and sampling accuracy. S1.4. Start the data acquisition task and begin recording the original signal data. S2. Initialize the signal data. S2.1. Import the raw data into MATLAB. S2.2. Low-pass filtering is performed to remove high-frequency noise components. S2.3. Denoise the filtered data using techniques such as wavelet denoising or moving average. S2.4. Ensure that the processed data retains valid signal information while filtering out noise interference. S3. Extract key features of the signal. S3.1. Use the Fast Fourier Transform to extract key frequency and amplitude features from the preprocessed signal. S3.2. Determine that the feature extraction parameters include window length and overlap rate. S3.3. Normalize the extracted features to ensure they fall within a certain range. S3.4. Verify whether the extracted features are representative, and perform feature selection or dimensionality reduction processing as necessary. S4. Convert the signal to a format suitable for deep learning processing. S4.1. Based on the input requirements of the deep learning model, organize the signal data into an appropriate data structure, including tensors. S4.2. Considering the time-series nature of the signal data, organize it into time-series or sequence data. S4.3. Standardize or normalize the data to accelerate the convergence speed of the model and improve the training effect. S4.4. The processed dataset is split into a training set, a validation set, and a test set, and used for training and evaluating the model. S5. Perform signal processing using a deep neural network. S5.1. Select a recurrent neural network deep learning model architecture. S5.2. Set the network structure of the model, including parameters such as the number of layers and the number of neurons. S5.3. Train the model using the training set and monitor the change in the loss function during the training process. S5.4. Evaluate the model's performance using the validation set, and perform parameter tuning and optimization to improve the model's accuracy and generalization ability. S6. Train the deep learning model to optimize its performance. S6.1. Split the dataset and set the proportions for training, validation, and testing. S6.2. Adjust the model parameters using the Adam optimizer. S6.3. Monitor training loss and validation loss during the training process to ensure that the model can achieve good performance on both the training and validation sets. S6.4. Select the optimal combination of model parameters using cross-validation or hyperparameter search techniques. Analyze the S7.DSP code. Use S7.1.Cppcheck to check the DSP code and identify potential code defects. Refer to the documentation and specifications for the S7.2 DSP processor to ensure that the code meets hardware requirements and best practices. S7.3. Conduct a code audit, reviewing the code with team members to identify potential errors and propose improvements. S7.4. Conduct a code quality assessment, including evaluating code complexity and maintainability metrics. S8. Analyze power supply noise and design a filter. S8.1. Measure the noise level in the power line using an oscilloscope or spectrum analyzer. S8.2. Analyze the noise spectrum to determine the frequency range and filter type that requires filtering. S8.3. Design the filter and select appropriate filter parameters to meet the design requirements. S8.4. Use MATLAB to perform performance verification and optimization on the designed filter. S9. Monitor the power supply and system temperature. S9.1. Install the voltmeter and ammeter and connect them to the system's power supply circuit. S9.2. Set monitoring parameters, including alarm thresholds for voltage, current, and temperature. S9.3. Monitor the power supply voltage, current, and system temperature in real time and record historical data. S9.4. If an abnormal situation is detected, an alarm will be triggered and appropriate emergency measures will be taken. S10. Convert the processing results to an appropriate output format. S10.1. Use MATLAB to organize the processed results into an acceptable output format. S10.2. If necessary, convert the results into text, graph, image, or video format to facilitate subsequent analysis or display. S10.3. Ensure the visualization effect and readability of the output data, and perform necessary typesetting and formatting. S10.4. Verify the completeness and accuracy of the output results and ensure that they match the processing results of the raw data. S11. Control the output of the results. S11.1. The output control module sets the output mode, which includes display, storage, or transmission to a remote server. S11.2. Based on the system needs and user preferences, set the output parameters, including the display resolution and storage format. S11.3. Output control is performed to ensure that the results are output in the desired format. S11.4. Monitor the output process, and timely address any errors or abnormal situations that may occur to ensure the stability and reliability of the output. S12. Provide a user interface. S12.1. Develop user interface software to enable users to interact with the system. S12.2. Design a user interface, including a graphical user interface (GUI) or a command-line interface (CLI), to meet the needs of different users. S12.3. Implement a parameter setting function, allowing users to adjust system parameters and algorithm settings. S12.4. Provide a results viewing function so that users can easily view and analyze the processing results. S13. Monitor the overall system status. S13.1. System monitoring software shall be deployed to monitor the status of the system's hardware and software operation in real time. S13.2. Set monitoring indicators and thresholds to promptly detect system abnormalities or failures. S13.3. Monitor the system status and periodically check the system operation logs and alarm information. S13.4. If an abnormal situation is detected, necessary measures will be taken, including restarting the system and repairing software errors, to ensure the stable and efficient operation of the system.

[0034] Although embodiments of this application have been illustrated and described, those skilled in the art can make various changes, modifications, substitutions and variations to these embodiments without departing from the principles and spirit of this application, and the scope of this application is limited by the appended claims and equivalents.

Claims

1. A signal processing method based on a deep neural network, wherein the signal processing method is implemented based on a signal processing system, Step S1 involves collecting raw signal data, Step S2 involves performing initial processing on the signal data, including filtering and noise reduction. Step S3 involves extracting key features of the processed signal, Step S4 involves converting the signal into a format suitable for deep learning processing, Step S5 involves classifying, identifying, or predicting a signal using a deep neural network, Step S6 involves training a deep learning model and optimizing its performance, Step S7 involves analyzing the DSP code to ensure its accuracy and efficiency, Step S8 involves analyzing power supply noise, designing filters, and optimizing power supply quality. Step S9 involves monitoring the power supply and system temperature to ensure a stable power supply and avoid overheating. Step S10 converts the processing result into an appropriate output format, Step S11 controls the output of the result, including display, storage, or transmission. Step S12 provides a user interface that allows parameter setting and viewing of results, Step S13 involves monitoring the overall system status and ensuring stable and highly efficient operation. A method for processing signals based on a deep neural network, characterized by including the following.

2. The signal processing system includes a data acquisition module, a signal processing module, a machine learning module, a DSP code audit module, a power filtering optimization module, a power supply and thermal problem detection module, an output processing module, and a system management module. The data acquisition module is arranged to acquire and preprocess signals, The aforementioned signal processing module is arranged to perform feature extraction and signal conversion. The aforementioned machine learning module is configured to perform deep learning and model training. The aforementioned DSP code audit module is configured to perform code checks and performance evaluations. The aforementioned power supply filtering optimization module is arranged to perform power supply noise analysis and filter design. The aforementioned power supply and thermal problem detection module is arranged to monitor the power supply and temperature. The output processing module is arranged to perform data format conversion and output control. The signal processing method based on a deep neural network according to claim 1, characterized in that the system management module is arranged to provide a user interface and monitor the system.

3. The data acquisition module further includes a signal acquisition unit and a preprocessing unit, The signal acquisition unit is arranged to acquire the original signal from the sensor. The signal processing method based on a deep neural network according to claim 2, characterized in that the preprocessing unit is arranged to perform initial processing on the original signal.

4. The signal processing module further includes a feature extraction unit and a signal conversion unit, The feature extraction unit is arranged to extract key features from the pre-processed signal. The signal processing method based on a deep neural network according to claim 3, characterized in that the signal conversion unit is arranged to convert the signal into a format suitable for deep learning model processing.

5. The aforementioned machine learning module further includes a deep learning unit and a model training unit. The deep learning unit is configured to perform signal classification, identification, or prediction tasks using a deep neural network. The signal processing method based on a deep neural network according to claim 4, characterized in that the model training unit is arranged to train and optimize a deep learning model using a training dataset.

6. The DSP code audit module further includes a code check unit and a performance evaluation unit, The code check unit is configured to perform static analysis on the DSP code. The deep neural network-based signal processing method according to claim 5, characterized in that the performance evaluation unit is arranged to evaluate the execution efficiency and resource consumption of the DSP code.

7. The power supply filtering optimization module further includes a power supply noise analysis unit and a filter design unit, The power supply noise analysis unit is arranged to analyze noise in the power lines, The signal processing method based on a deep neural network according to claim 6, characterized in that the filter design unit is arranged to design an appropriate filter.

8. The power supply and thermal problem detection module further includes a power supply monitoring unit and a temperature monitoring unit. The aforementioned power supply monitoring unit is positioned to monitor the power supply voltage and current in real time. The signal processing method based on a deep neural network according to claim 7, characterized in that the temperature monitoring unit is arranged to monitor the temperature of the system.

9. The output processing module further includes a data format conversion unit and an output control unit. The data format conversion unit is arranged to convert the processing result into a format suitable for output or display. The signal processing method based on a deep neural network according to claim 8, characterized in that the output control unit is arranged to control the output method of the result.

10. The system management module further includes a user interface unit and a system monitoring unit. The user interface unit is arranged to provide a user interaction interface. The signal processing method based on a deep neural network according to claim 9, characterized in that the system monitoring unit is arranged to monitor the status of the system.

11. A signal processing system based on a deep neural network, the signal processing system includes a data acquisition module, a signal processing module, a machine learning module, a DSP code audit module, a power filtering optimization module, a power supply and thermal problem detection module, an output processing module, and a system management module. The data acquisition module is arranged to acquire and preprocess signals, The aforementioned signal processing module is arranged to perform feature extraction and signal conversion. The aforementioned machine learning module is configured to perform deep learning and model training. The aforementioned DSP code audit module is configured to perform code checks and performance evaluations. The aforementioned power supply filtering optimization module is arranged to perform power supply noise analysis and filter design. The aforementioned power supply and thermal problem detection module is arranged to monitor the power supply and temperature. The output processing module is arranged to perform data format conversion and output control. The signal processing system based on a deep neural network is characterized in that the system management module is arranged to provide a user interface and monitor the system.

Images