Pulsating pressure sensor measurement signal simulation device and method
By combining convolutional neural networks and Kalman filters, along with Fourier transform and random forest models, the accuracy and stability issues of pulsating pressure signal measurement under strong noise conditions were solved, achieving high-precision simulation of pulsating pressure signals.
Patent Information
- Application Number
- CN202511537868.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-10-27
- Publication Date
- 2026-01-27
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Existing technologies have low accuracy in measuring pulsating pressure signals in high-noise environments. Traditional filtering methods cannot effectively handle various types of noise, resulting in unstable and inaccurate measurement results.
A convolutional neural network is used to identify noise types and calculate intensity, Kalman filter parameters are adjusted in real time, and Fourier transform and random forest models are combined for feature fusion and prediction to generate pulsating pressure signals.
It significantly improves the simulation accuracy and reliability of pulsating pressure signals, making it suitable for fields such as industrial monitoring, medical and health care, and environmental monitoring.
Smart Images
Figure CN121412913A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of signal processing technology, and in particular to a device and method for simulating the measurement signal of a pulsating pressure sensor. Background Technology
[0002] Pulsating pressure sensor technology is widely used in fluid dynamics, automotive industry, aerospace, medical equipment, and other fields to monitor pressure changes in liquid or gas flow in real time. With the development of sensor technology, the measurement accuracy and real-time performance of pulsating pressure sensors have gradually improved. In recent years, with the continuous advancement of digital signal processing technology and artificial intelligence algorithms, the methods for acquiring and processing pulsating pressure signals have also undergone significant changes. Traditional pulsating pressure signal processing methods mainly rely on classical filtering techniques and mathematical models, such as Kalman filtering and Fourier transform, to optimize the measurement signal through noise reduction and feature extraction. However, these methods often have certain limitations in complex environments, such as poor adaptability to different noise types and limited ability to extract signal features.
[0003] A common problem in existing technologies is the low measurement accuracy of pulsating pressure signals in noisy environments. While traditional Kalman filters can effectively handle a certain range of noise, they cannot provide a dynamically adaptive solution for the diverse types of noise in complex environments. Furthermore, signal processing methods such as Fourier transforms offer relatively simplistic extraction of frequency domain features, failing to fully consider the deep integration of time and frequency domain features. Therefore, in practical applications, existing technologies often suffer from over-filtering or ineffective filtering of pulsating pressure signals, leading to instability and inaccuracy in the measurement results. These problems significantly limit the application efficiency of pulsating pressure sensors in high-noise, high-complexity environments. Summary of the Invention
[0004] In view of the aforementioned existing problems, the present invention is proposed.
[0005] Therefore, the present invention provides a method for simulating the measurement signal of a pulsating pressure sensor, which solves the problem of insufficient signal accuracy and stability under noise interference.
[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, the present invention provides a method for simulating pulsating pressure sensor measurement signals, comprising: acquiring multi-source pulsating pressure sensor data and preprocessing it; based on the preprocessed pulsating pressure data, using a convolutional neural network to identify and classify noise, obtaining the noise type of the pulsating pressure signal, and calculating the noise intensity; based on the noise type and intensity of the pulsating pressure signal, adjusting the parameters of a Kalman filter in real time to eliminate noise from the pulsating pressure signal; based on the denoised pulsating pressure signal, calculating time-domain features and using a Fourier transform method to calculate frequency-domain features to generate a pulsating pressure signal feature vector; fusing the pulsating pressure signal feature vector using a random forest model and optimizing it using a cross-validation method to generate a pulsating pressure signal prediction model; using the pulsating pressure signal prediction model to predict the pressure value of real-time data, and adding noise to the prediction result to simulate the pulsating pressure signal.
[0007] As a preferred embodiment of the method for simulating the measurement signal of the pulsating pressure sensor described in this invention, the specific steps for acquiring multi-source pulsating pressure sensor data and performing preprocessing are as follows. Multi-source pulsating pressure sensor data includes pulsating pressure data, temperature data, humidity data, vibration data, flow velocity data, timestamps, and location information; Check the data, correct and remove errors in the dataset, and complete the data cleaning process; Time synchronization is achieved by adjusting the timestamps of data from different sensors using interpolation methods. Low-pass filtering algorithms are used to remove noise components from signals, thus completing the filtering of data. Scaling the data to a specific range creates normalization.
[0008] As a preferred embodiment of the method for simulating the measurement signal of the pulsating pressure sensor described in this invention, the steps of using a convolutional neural network to identify and classify noise based on the preprocessed pulsating pressure data to obtain the noise type of the pulsating pressure signal and calculate the noise intensity are as follows: Build a CNN model, utilize a convolutional neural network architecture, and train the model using the Adam optimizer and cross-entropy loss function; The preprocessed pulsating pressure data is input into a CNN model for feature extraction. In CNN models, features output from convolutional layers are synthesized. Using a CNN model, noise is classified from the signal to determine the type of background noise. The intensity of background noise is calculated by using a weighted average method on the feature maps output by the CNN model.
[0009] In a preferred embodiment of the pulsating pressure sensor measurement signal simulation method of the present invention, the steps for real-time adjustment of the Kalman filter parameters based on the noise type and intensity of the pulsating pressure signal to eliminate noise are as follows: Based on the noise type and intensity of the pulsating pressure signal, the noise type influence factor is calculated through spectrum analysis, and the measured noise intensity is used as the influence factor. Based on the obtained influencing factors and noise intensity, a comprehensive analysis of the noise environment is conducted through statistical analysis, frequency domain analysis, time series analysis and machine learning methods. This analysis reveals the factors that have the greatest impact on the noise environment, the distribution of noise at different frequencies, and the variation of noise over time. Based on the results of noise analysis, an adjustment strategy is formulated to dynamically adjust the parameters of the Kalman filter; Based on the adjusted Kalman filter parameters, noise is eliminated from the pulsating pressure signal using a denoising algorithm.
[0010] As a preferred embodiment of the pulsating pressure sensor measurement signal simulation method of the present invention, the step of generating a pulsating pressure signal feature vector by calculating time-domain features based on the denoised pulsating pressure signal and calculating frequency-domain features using the Fourier transform method is as follows. The mean of the denoised pulsating pressure signal is calculated using the arithmetic mean. The standard deviation is calculated by squaring the population standard deviation. The maximum and minimum peak values of the pulsating pressure signal are calculated using the peak value calculation method. Kurtosis is obtained by calculating the sharpness of the distribution of the pulsating pressure signal using excess kurtosis. The mean, standard deviation, peak value, and kurtosis are integrated to form the time-domain characteristics of the pulsating pressure signal; Based on the denoised pulsating pressure signal, the time-frequency domain characteristics of the pulsating pressure signal are calculated using the Fourier transform method. By using a feature concatenation method, the time-domain features and frequency-domain features of the pulsating pressure signal are combined to generate a feature vector of the pulsating pressure signal.
[0011] As a preferred embodiment of the pulsating pressure sensor measurement signal simulation method of the present invention, the steps of fusing the pulsating pressure signal feature vector through a random forest model and optimizing it using cross-validation to generate a pulsating pressure signal prediction model are as follows. Initialize the parameters of the random forest model, obtain raw data from the pulsating pressure sensor, preprocess and divide the data into training sets, and train the random forest model using the training set data. Based on the trained random forest model, feature concatenation and weighted fusion are performed on the feature vector of pulsating pressure signal. The hyperparameter configuration is obtained by combining grid search with cross-validation, and the trained random forest model is then optimized. By combining the fused pulsating pressure signal feature vector with the optimized random forest model, a machine learning algorithm is used to obtain a pulsating pressure signal prediction model.
[0012] As a preferred embodiment of the method for simulating the measurement signal of a pulsating pressure sensor according to the present invention, the step of predicting the pressure value of real-time data using a pulsating pressure signal prediction model and adding noise to the prediction result to simulate the pulsating pressure signal is as follows: Real-time acquisition of multi-source pulsating pressure sensor data and generation of real-time pulsating pressure signal feature vectors; Based on the trained random forest model, the feature vectors of real-time pulsating pressure signals are fused. The fused real-time pulsating pressure signal feature vector is input into the pulsating pressure signal prediction model to predict pressure changes and generate the pressure value at the current moment. Noise is generated using a Gaussian noise model, and the noise intensity is calculated by analyzing the fluctuation of the pressure change prediction results to generate the final noise. By using an additive noise model, the final noise is added to the generated current pressure value to simulate a pulsating pressure signal.
[0013] Secondly, the present invention provides a device for simulating pulsating pressure sensor measurement signals, comprising a data acquisition module, a noise identification and classification module, a noise cancellation module, a feature vector generation module, a prediction model generation module, and a simulated pulsating pressure signal module; the data acquisition module is used to acquire multi-source pulsating pressure sensor data and perform preprocessing; the noise identification and classification module is used to identify and classify noise based on the preprocessed pulsating pressure data using a convolutional neural network, obtain the noise type of the pulsating pressure signal, and calculate the noise intensity; the noise cancellation module is used to adjust the Kal in real time based on the noise type and intensity of the pulsating pressure signal. The parameters of the man filter are used to eliminate noise from the pulsating pressure signal; the feature vector generation module is used to calculate the time-domain features based on the denoised pulsating pressure signal, and then use the Fourier transform method to calculate the frequency-domain features to generate the pulsating pressure signal feature vector; the prediction model generation module is used to fuse the pulsating pressure signal feature vector through a random forest model and optimize it using cross-validation to generate a pulsating pressure signal prediction model; the simulated pulsating pressure signal module is used to predict the pressure value of real-time data through the pulsating pressure signal prediction model, and add noise to the prediction result to simulate the pulsating pressure signal.
[0014] Thirdly, the present invention provides a computer device including a memory and a processor, wherein the memory stores a computer program, wherein when the computer program is executed by the processor, it implements any step of the pulsating pressure sensor measurement signal simulation method as described in the first aspect of the present invention.
[0015] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the pulsating pressure sensor measurement signal simulation method as described in the first aspect of the present invention.
[0016] The beneficial effects of this invention are as follows: through innovative noise identification and classification methods and noise cancellation strategies that adjust Kalman filter parameters in real time, the simulation accuracy and reliability of pulsating pressure signals are significantly improved, making it applicable to multiple fields such as industrial monitoring, medical and health care, and environmental monitoring. These innovations not only address the shortcomings of existing technologies but also provide new ideas and technical means for research and applications in related fields. Attached Figure Description
[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0018] Figure 1 This is a flowchart of the method for simulating the measurement signal of the pulsating pressure sensor in Example 1.
[0019] Figure 2 This is a schematic diagram of the pulsating pressure sensor measurement signal simulation device in Example 1. Detailed Implementation
[0020] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.
[0021] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.
[0022] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.
[0023] Example 1, referring to Figure 1 and Figure 2 This is the first embodiment of the present invention, which provides a method for simulating the measurement signal of a pulsating pressure sensor, including the following steps: S1. Collect data from multi-source pulsating pressure sensors and perform preprocessing; Multi-source pulsating pressure sensor data includes pulsating pressure data, temperature data, humidity data, vibration data, flow velocity data, timestamps, and location information; It should be noted that pulsating pressure data is acquired in real-time by a pulsating pressure sensor, recording pressure fluctuations over time. Temperature data is acquired using a temperature sensor to collect ambient temperature data; temperature and pressure signals typically exhibit a certain correlation, affecting sensor accuracy and output. Humidity data is acquired using a humidity sensor; changes in humidity can affect the sensor's response characteristics, especially in high-humidity environments, where sensitivity may vary. Vibration data is acquired using a vibration sensor to obtain environmental vibration information; vibration has a certain impact on the pressure signal and must therefore be considered in conjunction with the pressure signal. Flow velocity data is recorded by a flow velocity sensor to measure the flow rate of gas or liquid; changes in flow velocity directly affect pressure changes and must be synchronized with other data. Timestamps are required for each data point to ensure time synchronization and alignment of data from different sensors. Location information is data collected from the sensor's location; especially in dynamic environments, changes in location can affect pressure data.
[0024] Check the data, correct and remove errors in the dataset, and complete the data cleaning process; It should be noted that the completeness of the data output from each sensor should be checked. If there are missing data points, missing values need to be filled or incomplete data discarded. Statistical methods should be used to detect and remove obvious outliers. For small-scale erroneous data, interpolation or fitting methods can be used to correct them, ensuring data consistency and accuracy.
[0025] Time synchronization is achieved by adjusting the timestamps of data from different sensors using interpolation methods. It should be noted that since different sensors may sample at different frequencies, timestamps need to be interpolated at this stage to ensure that the data from all sensors are aligned within the same time window. Linear interpolation or spline interpolation methods are used to fill the time difference between different sensors, thus synchronizing the data in time.
[0026] Low-pass filtering algorithms are used to remove noise components from signals, thus completing the filtering of data. It should be noted that a low-pass filter is used to remove high-frequency noise components from the sensor data. For low-frequency drift that may exist in the pressure data, a high-pass filter can be used to remove this low-frequency noise, enhancing the signal's validity.
[0027] Scaling the data to a specific range creates normalization; It should be noted that the data from each sensor are scaled to the same dimensional range to prevent certain data from dominating subsequent calculations due to excessively large dimensions, thus ensuring a balance of influence from various data sources. For example, data such as temperature, humidity, flow rate, and vibration are normalized to the [0, 1] interval.
[0028] S2. Based on the preprocessed pulsating pressure data, a convolutional neural network is used to identify and classify noise, obtain the noise type of the pulsating pressure signal, and calculate the noise intensity. Build a CNN model, utilize a convolutional neural network architecture, and train the model using the Adam optimizer and cross-entropy loss function; It should be noted that the input data is fed into the CNN network layer by layer, passing through convolutional layers, pooling layers, fully connected layers, and other computational steps, finally outputting the predicted value. The gradient of the loss function is calculated, and the network weights are adjusted using the Adam optimizer to minimize the loss function. The gradient is calculated layer by layer using the backpropagation algorithm, and the weights are updated accordingly. Iterative training is performed using batch training data. A learning rate scheduler is used to gradually decrease the learning rate to improve convergence.
[0029] The preprocessed pulsating pressure data is input into a CNN model for feature extraction. It should be noted that through multiple convolutional and pooling operations, CNNs can automatically learn and extract temporal features, periodic variation features, and local structural features from pulsating pressure signals. Convolutional layers extract local spatially relevant features from the pulsating pressure signal, such as signal abrupt changes, noise patterns, or short-duration pulses. Each convolutional operation focuses on different levels of detail in the signal.
[0030] In CNN models, features output from convolutional layers are synthesized. It should be noted that by concatenating features, CNN models can retain more feature information, which helps capture the complex patterns of pulsating pressure signals and effectively improves the diversity and accuracy of feature representation. Weighted averaging allows for focusing on more valuable features based on the importance of each feature map, thereby improving the CNN model's ability to recognize pulsating pressure signals, especially in noisy environments. Max pooling reduces the size of feature maps while retaining the most discriminative feature information, which helps improve the CNN model's noise resistance and generalization ability.
[0031] Using a CNN model, noise is classified from the signal to determine the type of background noise. It should be noted that the CNN model outputs a probability distribution representing the probability value of each noise type. Based on the maximum probability value, the CNN model can determine which noise type the pulsating pressure signal belongs to. The output category can be mapped to the corresponding noise type based on the labels in the training dataset.
[0032] The intensity of background noise is calculated by using a weighted average method on the feature maps output by the CNN model. It should be noted that the intensity value of the noise is calculated based on the probability distribution of the noise type and the specific characteristics of the signal. This intensity value can be combined with factors such as signal amplitude variation and spectral characteristics to obtain the relative intensity of each noise type.
[0033] S3. Based on the noise type and intensity of the pulsating pressure signal, adjust the parameters of the Kalman filter in real time to eliminate noise in the pulsating pressure signal; Based on the noise type and intensity of the pulsating pressure signal, the noise type influence factor is calculated through spectrum analysis, and the measured noise intensity is used as the influence factor. It should be noted that the noise type influence factor reflects the degree of influence of noise type on the pulsating pressure signal. Based on the spectrum and noise intensity distribution, the influence factor of each noise type in the entire signal is calculated. The noise intensity value obtained from actual measurements is used as the adjustment parameter for the influence factor. The noise intensity value directly affects the weight of the noise type, and thus affects the noise cancellation effect on the signal.
[0034] Based on the obtained influencing factors and noise intensity, a comprehensive analysis of the noise environment is conducted through statistical analysis, frequency domain analysis, time series analysis and machine learning methods. This analysis reveals the factors that have the greatest impact on the noise environment, the distribution of noise at different frequencies, and the variation of noise over time. It should be noted that, based on the denoised signal data, time series analysis methods are used to analyze the variation pattern of noise over time. If the noise exhibits periodic variations over time, a seasonal decomposition method is used to extract the seasonal components of the noise and analyze its periodic fluctuation characteristics. Clustering algorithms are used to classify noise types into different groups based on their frequency domain characteristics and time series characteristics. By analyzing the differences between different groups, the variation pattern of noise under different environments can be identified.
[0035] Based on the results of noise analysis, an adjustment strategy is formulated to dynamically adjust the parameters of the Kalman filter; It should be noted that the noise intensity and noise type influence factors obtained through the previous noise analysis are dynamically adjusted at each time step according to the changes in noise, using the process noise covariance matrix and the observation noise covariance matrix.
[0036] Based on the adjusted Kalman filter parameters, noise is eliminated from the pulsating pressure signal using a denoising algorithm; It should be noted that with each time step, the Kalman filter dynamically updates the state estimate based on the adjusted noise model, effectively suppressing noise and preserving the main trends of the real signal. In each iteration, the Kalman filter gradually improves the noise cancellation effect by updating the estimate, ultimately obtaining an output that is as close as possible to the real pulsating pressure signal.
[0037] S4. Based on the denoised pulsating pressure signal, the time domain features are calculated, and the frequency domain features are calculated using the Fourier transform method to generate the pulsating pressure signal feature vector. The mean of the denoised pulsating pressure signal is calculated using the arithmetic mean. It should be noted that the arithmetic mean provides the center position of the signal, i.e., the average level of the signal data. After denoising, the mean helps to remove or ignore the influence of random fluctuations on the signal, thus better capturing the true characteristics of signal variation, especially important when processing periodic or steady-state signals. The mean provides a quantitative description of the stability of pulsating pressure signals. The mean is often used as a base reference value for subsequent analyses of standard deviation and kurtosis.
[0038] The standard deviation is calculated by squaring the population standard deviation. It should be noted that the mean of the denoised pulsating pressure signal is obtained by calculating the arithmetic mean in the previous step, which forms the basis for calculating the standard deviation. The difference between each signal value and the mean is calculated, and each deviation is squared to eliminate the canceling effect of negative differences. All squared differences are summed and divided by the total number of samples to obtain the variance. The standard deviation is obtained by taking the square root of the variance.
[0039] The maximum and minimum peak values of the pulsating pressure signal are calculated using the peak value calculation method. It should be noted that a time window for the pulsating pressure signal needs to be determined. This window can be the entire time range of the signal or a specific portion of the signal. The entire pulsating pressure signal data is scanned, and fluctuations in the signal value are identified at each time point, recording the maximum and minimum values. Within the time window, the maximum value in the signal is selected, representing the highest pressure value. The minimum value within the time window is then selected, representing the lowest pressure value.
[0040] Kurtosis is obtained by calculating the sharpness of the distribution of the pulsating pressure signal using excess kurtosis. It should be noted that the mean of the pulsating pressure signal is the center of the pulsating pressure signal dataset. The standard deviation measures the average distance of the signal data from the mean, representing the volatility of the pulsating pressure signal data. The fourth central moment measures the degree of deviation of the signal data from the fourth power of the mean, and is used to describe the tail distribution of the data. The excess kurtosis value is calculated using the fourth central moment and the standard deviation.
[0041] The mean, standard deviation, peak value, and kurtosis are integrated to form the time-domain characteristics of the pulsating pressure signal; It should be noted that the pulsating pressure signal is first denoised and standardized to ensure signal quality and data consistency. Based on the preprocessed signal data, the mean, standard deviation, peak value, and kurtosis are calculated sequentially. The calculated mean, standard deviation, peak value, and kurtosis of each time-domain feature are then combined to form the time-domain feature.
[0042] Based on the denoised pulsating pressure signal, the time-frequency domain characteristics of the pulsating pressure signal are calculated using the Fourier transform method. It should be noted that a Fourier transform is applied to the denoised pulsating pressure signal to convert the signal from the time domain to the frequency domain. The Fourier transform result will provide amplitude and phase information for each frequency component. By analyzing the frequency domain characteristics, key frequency components in the pulsating pressure signal can be identified, such as the system's operating frequency, noise frequency, and other frequency components.
[0043] By using the feature concatenation method, the time-domain features and frequency-domain features of the pulsating pressure signal are combined to generate a feature vector of the pulsating pressure signal; It should be noted that time-domain and frequency-domain features reflect different aspects of a signal. By splicing them together, the information from both domains is fused, enabling subsequent processing methods to make judgments based on a more comprehensive signal representation. Pulsating pressure signals can be very complex, containing various components such as periodic changes, sudden events, and noise. Combining time-domain and frequency-domain features allows for a more comprehensive capture of the patterns in these complex signals, aiding in the processing of signals with multiple variation modes.
[0044] S5. The feature vector of the pulsating pressure signal is fused using a random forest model and optimized using cross-validation to generate a pulsating pressure signal prediction model. Initialize the parameters of the random forest model, obtain raw data from the pulsating pressure sensor, preprocess and divide the data into training sets, and train the random forest model using the training set data. It should be noted that the training set contains a portion of the data, typically 70% to 80% of the original data, on which the model will learn and fine-tune its parameters. The test set usually contains the remaining 20% to 30% of the data, used to validate the model's performance on unseen data and avoid overfitting. Using the partitioned training set data, multiple decision trees are constructed using the random forest algorithm. Each tree randomly draws a subset of samples and features from the training data, ensuring that each tree is independent during training and thus preventing overfitting.
[0045] Based on the trained random forest model, feature concatenation and weighted fusion are performed on the feature vector of pulsating pressure signal. It should be noted that during model training, the model has already learned the importance of each feature. When a new pulsating pressure signal enters the model, features from different sources are first concatenated into a unified feature vector. Each feature carries latent information learned by the model during training. For the concatenated feature vector, the random forest uses its built-in ensemble learning mechanism to weight the different features. Each decision tree adaptively weights the features based on the feature importance learned during training.
[0046] The hyperparameter configuration is obtained by combining grid search with cross-validation, and the trained random forest model is then optimized. It should be noted that grid search automatically evaluates the cross-validation results of all hyperparameter combinations and ultimately selects the best-performing hyperparameter configuration, typically the one that optimizes the model's average evaluation metric on the validation set. Cross-validation, by dividing the data into different training and validation sets, helps avoid overfitting the model to a single dataset, thereby improving the model's generalization ability and making its performance more stable on unknown data.
[0047] By combining the fused pulsating pressure signal feature vector with the optimized random forest model, a machine learning algorithm is used to obtain a pulsating pressure signal prediction model. It should be noted that historical data is used to train the optimized random forest model. During this stage, the random forest learns the patterns and trends in the signal based on the input feature vectors and generates a set of decision trees. When a new pulsating pressure signal feature vector is input into the trained random forest model, the model uses a voting mechanism among the decision trees to synthesize the prediction results of multiple trees and ultimately arrives at an accurate prediction value. This prediction value represents the changing trend of the pulsating pressure signal at the current moment.
[0048] S6. Using a pulsating pressure signal prediction model, pressure values are predicted from real-time data, and noise is added to the prediction results to simulate pulsating pressure signals. Real-time acquisition of multi-source pulsating pressure sensor data and generation of real-time pulsating pressure signal feature vectors; It should be noted that the time-domain characteristics of the pulsating pressure signal are extracted using statistical methods such as arithmetic mean, standard deviation, peak value, and kurtosis. These characteristics reflect the basic trend and fluctuation amplitude of the signal. Fourier transform and other methods are then used to transform the signal from the time domain to the frequency domain, thereby analyzing the frequency components of the signal.
[0049] Based on the trained random forest model, the feature vectors of real-time pulsating pressure signals are fused. It should be noted that the features extracted from the time domain and frequency domain are concatenated to form a complete feature vector. In the random forest model, the weights of each feature can be adjusted to ensure that more important features have a greater impact on the model. Since the time domain and frequency domain features of the pulsating pressure signal describe the same signal from different dimensions, they need to be merged in an appropriate manner during the fusion process.
[0050] The fused real-time pulsating pressure signal feature vector is input into the pulsating pressure signal prediction model to predict pressure changes and generate the pressure value at the current moment. It should be noted that during the previous training process, the model had already learned the relationship between pulsating pressure and other relevant features based on historical data. Therefore, when a new feature vector is input into the model, it will predict the pressure value based on these learned relationships. Pulsating pressure signals typically have time-series properties, with trends changing continuously over time. The prediction model can identify features such as periodic fluctuations and sudden changes by learning from time-series data, thereby predicting the pressure value at the current moment and for a period of time in the future.
[0051] Noise is generated using a Gaussian noise model, and the noise intensity is calculated by analyzing the fluctuation of the pressure change prediction results to generate the final noise. It should be noted that a Gaussian noise model can be used to generate noise values at each time step. The noise intensity will be adjusted based on the volatility calculated in subsequent steps to ensure that the noise conforms to the changing characteristics of the actual environment. Volatility is typically expressed as the magnitude or fluctuation of pressure changes. By calculating the changes in the real-time pressure prediction results, the volatility characteristics of the pressure signal can be obtained.
[0052] By using an additive noise model, the final noise is added to the generated pressure value at the current moment to simulate a pulsating pressure signal. It should be noted that the generated noise is typically simulated using a Gaussian noise model, incorporating noise effects caused by the actual environment. The noise intensity is determined by factors such as fluctuation analysis, noise type, and noise frequency distribution. Through an additive noise model, the final noise signal is added to the predicted pressure value, forming a simulated pulsating pressure signal.
[0053] This embodiment also provides a device for simulating pulsating pressure sensor measurement signals, including: a data acquisition module, a noise identification and classification module, a noise cancellation module, a feature vector generation module, a prediction model generation module, and a simulated pulsating pressure signal module; the data acquisition module is used to acquire multi-source pulsating pressure sensor data and perform preprocessing; the noise identification and classification module is used to identify and classify noise based on the preprocessed pulsating pressure data using a convolutional neural network, obtain the noise type of the pulsating pressure signal, and calculate the noise intensity; the noise cancellation module is used to adjust the Kalma in real time based on the noise type and intensity of the pulsating pressure signal. The n-filter parameters are used to eliminate noise from the pulsating pressure signal; the feature vector generation module is used to calculate the time-domain features based on the denoised pulsating pressure signal, and then use the Fourier transform method to calculate the frequency-domain features to generate the pulsating pressure signal feature vector; the prediction model generation module is used to fuse the pulsating pressure signal feature vector through a random forest model and optimize it using cross-validation to generate a pulsating pressure signal prediction model; the simulated pulsating pressure signal module is used to predict the pressure value of real-time data through the pulsating pressure signal prediction model, and add noise to the prediction result to simulate the pulsating pressure signal.
[0054] This embodiment also provides a computer device applicable to the pulsating pressure sensor measurement signal simulation method, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the pulsating pressure sensor measurement signal simulation method proposed in the above embodiment.
[0055] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.
[0056] This embodiment also provides a storage medium storing a computer program that, when executed by a processor, implements the method for simulating the measurement signal of a pulsating pressure sensor as described in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.
[0057] In summary, this invention significantly improves the simulation accuracy and reliability of pulsating pressure signals through innovative noise identification and classification methods and a noise cancellation strategy that adjusts Kalman filter parameters in real time. It is applicable to multiple fields such as industrial monitoring, healthcare, and environmental monitoring. These innovations not only address the shortcomings of existing technologies but also provide new ideas and technical means for research and applications in related fields.
[0058] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.
Claims
1. A method for simulating the measurement signal of a pulsating pressure sensor, characterized in that: include, Data from multi-source pulsating pressure sensors is acquired and preprocessed. Based on the preprocessed pulsating pressure data, a convolutional neural network is used for noise identification and classification to obtain the noise type of the pulsating pressure signal and calculate the noise intensity. Based on the noise type and intensity of the pulsating pressure signal, the parameters of the Kalman filter are adjusted in real time to eliminate noise in the pulsating pressure signal. Based on the denoised pulsating pressure signal, the time domain features are calculated, and the frequency domain features are calculated using the Fourier transform method to generate the pulsating pressure signal feature vector. The feature vector of the pulsating pressure signal is fused using a random forest model and optimized using cross-validation to generate a pulsating pressure signal prediction model. The pressure value is predicted from real-time data using a pulsating pressure signal prediction model, and noise is added to the prediction results to simulate the pulsating pressure signal.
2. The method for simulating the measurement signal of a pulsating pressure sensor as described in claim 1, characterized in that: The specific steps for acquiring and preprocessing data from multi-source pulsating pressure sensors are as follows. Multi-source pulsating pressure sensor data includes pulsating pressure data, temperature data, humidity data, vibration data, flow velocity data, timestamps, and location information; Check the data, correct and remove errors in the dataset, and complete the data cleaning process; Time synchronization is achieved by adjusting the timestamps of data from different sensors using interpolation methods. Low-pass filtering algorithms are used to remove noise components from signals, thus completing the filtering of data. Scaling the data to a specific range creates normalization.
3. The method for simulating the measurement signal of a pulsating pressure sensor as described in claim 2, characterized in that: Based on the preprocessed pulsating pressure data, a convolutional neural network is used for noise identification and classification to obtain the noise type of the pulsating pressure signal and calculate the noise intensity. The specific steps are as follows. Build a CNN model, utilize a convolutional neural network architecture, and train the model using the Adam optimizer and cross-entropy loss function; The preprocessed pulsating pressure data is input into a CNN model for feature extraction. In CNN models, features output from convolutional layers are synthesized. Using a CNN model, noise is classified from the signal to determine the type of background noise. The intensity of background noise is calculated by using a weighted average method on the feature maps output by the CNN model.
4. The method for simulating the measurement signal of a pulsating pressure sensor as described in claim 3, characterized in that: The noise cancellation of the pulsating pressure signal is achieved by adjusting the parameters of the Kalman filter in real time based on the noise type and intensity of the pulsating pressure signal. The specific steps are as follows. Based on the noise type and intensity of the pulsating pressure signal, the noise type influence factor is calculated through spectrum analysis, and the measured noise intensity is used as the influence factor. Based on the obtained influencing factors and noise intensity, a comprehensive analysis of the noise environment is conducted through statistical analysis, frequency domain analysis, time series analysis and machine learning methods. This analysis reveals the factors that have the greatest impact on the noise environment, the distribution of noise at different frequencies, and the variation of noise over time. Based on the results of noise analysis, an adjustment strategy is formulated to dynamically adjust the parameters of the Kalman filter; Based on the adjusted Kalman filter parameters, noise is eliminated from the pulsating pressure signal using a denoising algorithm.
5. The method for simulating the measurement signal of a pulsating pressure sensor as described in claim 4, characterized in that: The denoised pulsating pressure signal is used to calculate its time-domain features and then its frequency-domain features using Fourier transform to generate a pulsating pressure signal feature vector. The specific steps are as follows: The mean of the denoised pulsating pressure signal is calculated using the arithmetic mean. The standard deviation is calculated by squaring the population standard deviation. The maximum and minimum peak values of the pulsating pressure signal are calculated using the peak value calculation method. Kurtosis is obtained by calculating the sharpness of the distribution of the pulsating pressure signal using excess kurtosis. The mean, standard deviation, peak value, and kurtosis are integrated to form the time-domain characteristics of the pulsating pressure signal; Based on the denoised pulsating pressure signal, the time-frequency domain characteristics of the pulsating pressure signal are calculated using the Fourier transform method. By using a feature concatenation method, the time-domain features and frequency-domain features of the pulsating pressure signal are combined to generate a feature vector of the pulsating pressure signal.
6. The method for simulating the measurement signal of a pulsating pressure sensor as described in claim 5, characterized in that: The process involves fusing the feature vectors of the pulsating pressure signal using a random forest model and optimizing them with cross-validation to generate a pulsating pressure signal prediction model. The specific steps are as follows: Initialize the parameters of the random forest model, obtain raw data from the pulsating pressure sensor, preprocess and divide the data into training sets, and train the random forest model using the training set data. Based on the trained random forest model, feature concatenation and weighted fusion are performed on the feature vector of pulsating pressure signal. The hyperparameter configuration is obtained by combining grid search with cross-validation, and the trained random forest model is then optimized. By combining the fused pulsating pressure signal feature vector with the optimized random forest model, a machine learning algorithm is used to obtain a pulsating pressure signal prediction model.
7. The method for simulating the measurement signal of a pulsating pressure sensor as described in claim 6, characterized in that: The method involves using a pulsating pressure signal prediction model to predict pressure values from real-time data and adding noise to the prediction results to simulate a pulsating pressure signal. The specific steps are as follows. Real-time acquisition of multi-source pulsating pressure sensor data and generation of real-time pulsating pressure signal feature vectors; Based on the trained random forest model, the feature vectors of real-time pulsating pressure signals are fused. The fused real-time pulsating pressure signal feature vector is input into the pulsating pressure signal prediction model to predict pressure changes and generate the pressure value at the current moment. Noise is generated using a Gaussian noise model, and the noise intensity is calculated by analyzing the fluctuation of the pressure change prediction results to generate the final noise. By using an additive noise model, the final noise is added to the generated current pressure value to simulate a pulsating pressure signal.
8. A device for simulating the measurement signal of a pulsating pressure sensor, based on the method for simulating the measurement signal of a pulsating pressure sensor according to any one of claims 1 to 7, characterized in that: It includes a data acquisition module, a noise identification and classification module, a noise cancellation module, a feature vector generation module, a prediction model generation module, and a simulated pulsating pressure signal module; The data acquisition module is used to acquire data from multi-source pulsating pressure sensors and perform preprocessing. The noise identification and classification module is used to identify and classify noise based on preprocessed pulsating pressure data using a convolutional neural network, obtain the noise type of the pulsating pressure signal, and calculate the noise intensity. The noise cancellation module is used to adjust the parameters of the Kalman filter in real time based on the noise type and intensity of the pulsating pressure signal to cancel the noise in the pulsating pressure signal. The feature vector generation module is used to calculate the time-domain features based on the denoised pulsating pressure signal, and then calculate the frequency-domain features using the Fourier transform method to generate the pulsating pressure signal feature vector. The prediction model generation module is used to fuse the feature vector of the pulsating pressure signal through a random forest model and optimize it using cross-validation to generate a prediction model for the pulsating pressure signal. The simulated pulsating pressure signal module is used to predict pressure values from real-time data using a pulsating pressure signal prediction model, and to add noise to the prediction results to simulate a pulsating pressure signal.
9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the pulsating pressure sensor measurement signal simulation method according to any one of claims 1 to 7.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the method for simulating the pulsating pressure sensor measurement signal according to any one of claims 1 to 7.