Method and device for collecting pressure distribution data

Abstract

The present application relates to the field of pressure sensing technology, and discloses a method and device for collecting pressure distribution data, the method comprising: calibrating a pressure sensor array and performing signal preprocessing to obtain an initial state matrix, a preprocessing parameter set and a response curve database; calculating the pressure change rate to obtain a pressure sampling dynamic trigger threshold and a pressure sampling control matrix; performing interval control on the pressure data collection process to obtain a pressure collection partition control model; collecting target pressure data and inputting it into a time-weighted processing network model for feature extraction to obtain multi-modal pressure distribution feature data; inputting the multi-modal pressure distribution feature data into a closed-loop control system, updating pressure sampling parameters and pressure processing parameters to obtain a pressure collection control parameter set, thereby realizing high-precision pressure distribution collection and solving the limitations of traditional methods in processing nonlinear and time-varying pressure signals.

Description

Technical Field

[0001] The present application relates to the field of pressure sensing technology, and in particular to a method and device for collecting pressure distribution data. Background Art

[0002] Traditional pressure distribution acquisition methods mainly use fixed frequency sampling strategies, which cannot effectively cope with the dynamic characteristics of pressure signals, resulting in insufficient sampling in rapidly changing areas and data redundancy in stable areas, which seriously affects the efficiency of system resource utilization. Traditional pressure distribution acquisition technologies generally have problems such as imperfect signal preprocessing and single sampling strategies, making it difficult to accurately capture the transient characteristics of pressure signals.

[0003] In addition, traditional feature extraction methods often only focus on a single feature of the pressure signal, ignoring the correlation characteristics of pressure distribution in the time and space dimensions, resulting in incomplete extracted feature information and failure to fully reflect the dynamic changes in pressure distribution. At the same time, the lack of an effective closed-loop feedback mechanism makes it difficult for the system to dynamically adjust the sampling strategy and processing parameters according to the actual application scenario, limiting the adaptability and reliability of the pressure sensing system. Summary of the invention

[0004] The present application provides a method and device for collecting pressure distribution data, thereby achieving high-precision pressure distribution collection and overcoming the limitations of traditional methods in processing nonlinear and time-varying pressure signals.

[0005] In a first aspect, the present application provides a method for collecting pressure distribution data, the method for collecting pressure distribution data comprising:

[0006] Calibrate and preprocess the pressure sensor array to obtain an initial state matrix, a preprocessing parameter set, and a response curve database;

[0007] Calculating the pressure change rate based on the initial state matrix and the preprocessing parameter set, and correcting based on the response curve database to obtain a pressure sampling dynamic trigger threshold and a pressure sampling control matrix;

[0008] According to the pressure sampling dynamic trigger threshold and the pressure sampling control matrix, the pressure data acquisition process is controlled between partitions to obtain a pressure acquisition partition control model;

[0009] The target pressure data is collected through the pressure collection partition control model, and the target pressure data is input into the time-weighted processing network model for feature extraction to obtain multi-modal pressure distribution feature data;

[0010] The multi-modal pressure distribution characteristic data is input into a closed-loop control system, and the pressure sampling parameters and pressure processing parameters are updated to obtain a pressure acquisition control parameter set.

[0011] A second aspect of the present application provides a pressure distribution data acquisition device, the pressure distribution data acquisition device comprising:

[0012] A preprocessing module is used to calibrate the pressure sensor array and preprocess the signal to obtain an initial state matrix, a preprocessing parameter set and a response curve database;

[0013] A calculation module, used to calculate the pressure change rate based on the initial state matrix and the preprocessing parameter set, and to perform correction based on the response curve database to obtain a pressure sampling dynamic trigger threshold and a pressure sampling control matrix;

[0014] A control module, used to perform partition-based control on the pressure data acquisition process according to the pressure sampling dynamic trigger threshold and the pressure sampling control matrix to obtain a pressure acquisition partition control model;

[0015] An extraction module, used for collecting target pressure data through the pressure collection partition control model, and inputting the target pressure data into the time-weighted processing network model for feature extraction to obtain multi-modal pressure distribution feature data;

[0016] The updating module is used to input the multi-modal pressure distribution characteristic data into a closed-loop control system, update the pressure sampling parameters and the pressure processing parameters, and obtain a pressure acquisition control parameter set.

[0017] The third aspect of the present application provides an electronic device, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor calls the instructions in the memory so that the electronic device executes the above-mentioned pressure distribution data collection method.

[0018] A fourth aspect of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores instructions, which, when executed on a computer, enable the computer to execute the above-mentioned method for collecting pressure distribution data.

[0019] Compared with the prior art, the present application has the following beneficial effects: by introducing an adaptive event trigger mechanism and a partitioned sampling control strategy, data redundancy is significantly reduced and the efficiency of system resource utilization is optimized; a time-weighted processing network model is combined with a deep separation convolution unit for feature extraction, thereby enhancing the ability to capture the spatiotemporal characteristics of pressure distribution; an innovative multimodal feature analysis framework is constructed to achieve a comprehensive characterization of pressure distribution characteristics; a closed-loop control system based on performance evaluation is established, so that sampling parameters and processing parameters can be dynamically optimized and adjusted according to actual application requirements, effectively improving the robustness and adaptability of the system; through multi-level signal preprocessing and feature extraction strategies, the limitations of traditional methods in processing nonlinear and time-varying pressure signals are solved. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0021] The structures, proportions, sizes, etc. illustrated in the drawings of this specification are only used to match the contents disclosed in the specification so as to facilitate understanding and reading by persons familiar with this technology. They are not used to limit the conditions under which the present invention can be implemented, and therefore have no substantive technical significance. Any structural modification, change in proportion or adjustment of size, without affecting the effects and purposes that can be achieved by the present invention, should still fall within the scope of the technical contents disclosed by the present invention.

[0022] Figure 1 is a flow chart of a method for collecting pressure distribution data provided by an embodiment of the present invention;

[0023] Figure 2 is a schematic block diagram of the structure of a pressure distribution data acquisition device provided by an embodiment of the present invention;

[0024] Figure 3 It is a schematic block diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0025] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0026] The flowcharts shown in the accompanying drawings are only examples and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may also be decomposed, combined or partially merged, so the actual execution order may change according to actual conditions.

[0027] It should also be understood that the terms used in this application specification are only for the purpose of describing specific embodiments and are not intended to limit the application. As used in this application specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0028] It should be further understood that the term "and / or" used in the specification and appended claims of this application refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations. Figure 1 , an embodiment of the method for collecting pressure distribution data in the embodiment of the present application includes:

[0029] Step 100, calibrating and preprocessing the pressure sensor array to obtain an initial state matrix, a preprocessing parameter set and a response curve database;

[0030] It is understandable that the execution subject of the present application may be a pressure distribution data collection device, or a terminal or a server, which is not limited here. The present application embodiment is described by taking the server as the execution subject as an example.

[0031] It should be noted that the pressure distribution acquisition method in the embodiment of the present application is applicable to flexible or rigid pressure sensor arrays of different accuracy levels. In terms of physical layout, the pressure sensor array usually adopts an m×n matrix arrangement structure, where m and n are both integers greater than or equal to 2, and can be flexibly configured according to the actual application scenario requirements. The array can be a regular square or rectangular layout, or an irregular arrangement of a specific shape. In terms of performance characteristics, the pressure sensor array must have the following basic characteristics: it can achieve pressure measurement in the range of 0-100kPa; it has good linearity and repeatability; the sampling frequency can be adjusted in the range of 100Hz to 1000Hz; it has certain anti-interference ability and environmental adaptability. The method proposed in this application is not limited to the pressure sensor array, but improves the overall performance of pressure distribution acquisition through technical means such as adaptive event triggering mechanism, partitioned interval sampling control strategy and multimodal feature analysis.

[0032] In this embodiment, the pressure sensing unit is the basic component unit of the pressure sensor array, and each unit is an independent pressure sensing element that can convert the pressure into an electrical signal output. The pressure sensing unit usually includes a pressure sensing element (such as capacitive, resistive, piezoelectric, etc.), a signal conditioning circuit and a data acquisition interface. The pressure sensor array is a sensor network system composed of multiple pressure sensing units organized according to a specific topological structure, and has spatially distributed measurement capabilities. The relationship between the pressure sensing unit and the pressure sensor array can be compared to the relationship between pixels and display screens: a single pressure sensing unit is responsible for the collection of local pressure information; and the pressure sensor array can achieve overall perception and dynamic monitoring of pressure distribution. Each sensing unit in the array has a unique row and column coordinate identifier, and works together through a unified data acquisition and processing system.

[0033] Specifically, by inputting a standard voltage signal to each pressure sensing unit in the pressure sensor array, the ratio between the output voltage value and the input voltage value of each sensing unit is calculated to obtain an initial response coefficient matrix. The basic response relationship of each pressure sensing unit under standard conditions is established to provide the preliminary response characteristics of the sensor to the standard input signal. According to the initial response coefficient matrix, the difference between the maximum response value and the minimum response value of each pressure sensing unit is calculated to obtain the range normalization parameter of each pressure sensing unit, unify the output range of each sensing unit, and make different units comparable under the same input pressure, thereby effectively reducing the differences between the units. The range normalization parameter is linearly operated with the initial response coefficient matrix to obtain the initial state matrix. The initial state matrix is ​​the stable state of the pressure sensor array without external force or reference pressure. This matrix provides a reference benchmark for subsequent pressure measurements to ensure that the output between each pressure sensing unit is carried out under a unified calibration framework.

[0034] A multi-order Butterworth filter is performed on the pressure signal collected by the pressure sensor array. The Butterworth filter has a smooth frequency response and is suitable for removing noise and interference from the pressure signal. After the multi-order Butterworth filter operation, a filter coefficient set is obtained to describe the characteristics of the filter to ensure that the filtered signal has good frequency characteristics and a high signal-to-noise ratio. Based on the filtered signal, the root mean square value and signal-to-noise ratio of the pressure signal are calculated. These parameters are used to describe the strength and quality of the signal to form a preprocessing parameter set. An increasing pressure load is applied to the pressure sensor array. By gradually increasing the applied pressure value and recording the output response of each pressure sensing unit at each pressure value, a pressure response data set is obtained.

[0035] The pressure response data set is fitted with a piecewise polynomial. The nonlinear response characteristics of the sensor are effectively captured through polynomial fitting, and the pressure-output relationship has a better analytical expression. The piecewise polynomial fitting ensures that the most appropriate fitting method is used in different pressure ranges to improve the accuracy of the fitting and minimize the error. The pressure response characteristic curve is stored in a matrix according to the position of the pressure sensing unit to form a response curve database of the pressure sensor. The pressure response characteristic curve is obtained through previous calibration and pressure loading experiments, and can accurately describe the response characteristics of the sensor in different pressure ranges. Among them, the calibration and pressure loading process of the pressure sensor is as follows: First, calibration is performed: a standard voltage signal is applied to each pressure sensing unit in the array, and the output / input voltage ratio is obtained to establish a reference response coefficient matrix; the difference between the maximum and minimum response values ​​of each unit is calculated to obtain the range normalization parameter; the normalization parameter is linearly operated with the response coefficient matrix, and the calibration correction equation is constructed to obtain the initial state matrix. Then, pressure loading is performed: a standard loading device is used to apply an increasing pressure load from 0 to 100 kPa to the array, and the output response of all sensing units is recorded at each pressure point to establish a pressure-output correspondence; the pressure response data of each sensing unit is fitted with a piecewise polynomial to obtain the fitting coefficient and error; finally, the fitted characteristic curve is stored in a matrix according to the unit position, and a query index is established to form a complete response curve database. This standardized process ensures that the nonlinear characteristics of the sensor can be accurately compensated during subsequent pressure acquisition, and the pressure signal can be accurately corrected through the response characteristic curve, providing a reliable data basis for the measurement of the entire system. The response curve database is a collection of the response behaviors of the pressure sensor under different pressure states, and stores the output characteristics of each sensing unit under different pressure conditions. By storing these response curves in a matrix, the query and use of sensor data are simplified, and in the subsequent pressure distribution acquisition process, dynamic correction and compensation are performed by comparing the actual collected data with the reference data in the response curve database, ensuring that the results of pressure distribution acquisition have higher accuracy and reliability, and effectively improving the measurement capability and anti-interference ability of the entire pressure sensing system.

[0036] Step 200, calculating the pressure change rate based on the initial state matrix and the preprocessing parameter set, and correcting it based on the response curve database to obtain the pressure sampling dynamic trigger threshold and the pressure sampling control matrix;

[0037] Specifically, the data of adjacent sampling points are extracted from the initial state matrix, and the real-time change rate matrix is ​​obtained by dividing the pressure difference of all pressure sensing units at adjacent moments by the sampling time interval, which reflects the change speed of the pressure signal of each sensing unit in a short time. The real-time change rate matrix is ​​convolved with the filter coefficient set in the preprocessing parameter set. The high-frequency noise components in the signal are filtered out by the convolution operation, and the real pressure change trend is retained to obtain the filtered pressure change rate data. Based on the filtered pressure change rate data, the pressure response characteristic curve of the corresponding pressure range is retrieved from the response curve database. The pressure response characteristic curve is obtained by the previous calibration and pressure loading experiment, and can accurately describe the response characteristics of the sensor in different pressure ranges. The corresponding pressure response correction coefficient is obtained by matching the corresponding pressure range in the response curve database. The point multiplication operation is performed on the filtered pressure change rate data and the pressure response correction coefficient to obtain the corrected pressure change rate data, which more accurately reflects the real change speed of each sensing unit under the current pressure state and eliminates the influence of the nonlinear and uneven response of the sensor. Based on the corrected pressure change rate data, the ratio of the standard deviation of the pressure change rate to the mean is calculated to obtain the dynamic adjustment factor. The dynamic adjustment factor reflects the degree of fluctuation and stability in the current pressure change process. Through the calculation of this factor, the trigger threshold of pressure sampling is dynamically adjusted to adapt to different pressure change conditions. The dynamic adjustment factor is multiplied by the reference trigger threshold to obtain the dynamic trigger threshold of pressure sampling. The dynamic trigger threshold determines when the pressure signal needs to be sampled, thereby ensuring that the system can respond quickly when the pressure changes drastically, and reducing unnecessary sampling when the pressure is relatively stable. The corrected pressure change speed data is compared with the dynamic trigger threshold to determine whether each pressure sensing unit meets the sampling trigger condition. Through this step, the sampling trigger flag matrix is ​​obtained. The sampling trigger flag matrix is ​​a Boolean matrix that identifies whether each sensing unit needs to collect data at the current moment. Only when the change speed exceeds the trigger threshold, the sensing unit will be marked as needing sampling, thereby realizing dynamic sampling control based on the change rate. The sampling trigger flag matrix is ​​encoded and reconstructed according to the position of the pressure sensing unit to obtain the pressure sampling control matrix.

[0038] Step 300, performing partition control on the pressure data acquisition process according to the pressure sampling dynamic trigger threshold and the pressure sampling control matrix to obtain a pressure acquisition partition control model;

[0039] It should be noted that the pressure sampling control matrix is ​​scanned in time sequence, and the initial sampling state of the pressure preparation acquisition interval is obtained by analyzing the state of the pressure sensing unit at each moment. The initial sampling state is used to determine which sensing units are at the boundary of pressure change and which units have not yet triggered the sampling conditions, so as to effectively distinguish the sampling needs of different areas. Based on the initial sampling state of the pressure preparation acquisition interval, the dynamic trigger threshold of pressure sampling is monitored to obtain the interval switching trigger signal. The state of the pressure sensing unit is evaluated in real time to determine whether the current interval should be switched to the continuous acquisition state. The interval switching trigger signal is a key signal in the entire sampling control process. By comparing the dynamic trigger threshold in real time, it is determined whether it is necessary to collect the pressure data in a certain interval more intensively in order to capture more subtle pressure changes.

[0040] The sampling frequency is adjusted based on the interval switching trigger signal to obtain the sampling control signal of the pressure continuous acquisition interval. The pressure continuous acquisition interval refers to the interval in which the system needs to maintain a high sampling frequency to continuously track pressure changes when the sampling condition is triggered by some sensing units. Based on the pressure change rate data in this interval, trend analysis is performed to predict the future direction of the pressure signal by judging the trend of data changes. The results of trend analysis are used to evaluate whether the current pressure signal changes are gradually becoming stable, and then optimize and adjust the sampling frequency.

[0041] Based on the judgment result of the change trend, if it is detected that the pressure change is gradually becoming stable, the sampling frequency is reduced to enter the pressure residual acquisition interval. In this interval, the main goal of sampling is to monitor the remaining small changes. Therefore, reducing the sampling frequency not only reduces the consumption of system resources, but also effectively avoids the collection of redundant data and improves the collection efficiency. At this stage, the frequency control parameters obtained by the frequency reduction process are used to guide the system on how to reasonably allocate the sampling frequency when the pressure changes steadily, so as to achieve a balance between acquisition accuracy and resource usage.

[0042] In the pressure residual acquisition interval, the frequency control parameters are adjusted autonomously to adapt to the current pressure change state and obtain the dynamic frequency range of the pressure active acquisition interval. In the pressure active acquisition interval, the sampling frequency is optimized according to the frequency adjustment results in the residual acquisition interval to achieve a rapid response to sudden changes that may occur in the future. The dynamic adjustment capability of this process ensures that the system can respond quickly when the pressure change increases again, thereby avoiding the loss of important pressure information due to too low sampling frequency. The initial sampling state of the pressure preparation acquisition interval, the sampling control signal of the pressure continuous acquisition interval, the frequency control parameters of the pressure residual acquisition interval, and the dynamic frequency range of the pressure active acquisition interval are combined and modeled by a state machine to obtain a pressure acquisition partition control model.

[0043] Step 400: collecting target pressure data through the pressure collection partition control model, and inputting the target pressure data into the time-weighted processing network model for feature extraction to obtain multi-modal pressure distribution feature data;

[0044] Specifically, the target pressure data is collected through the pressure collection partition control model. Through the partition control of the system, the collection process can achieve the optimal sampling strategy in different pressure intervals. The target pressure data is segmented into time series, and it is divided into multiple small segments according to the time dimension to obtain a pressure time series data group. The pressure time series data group is input into the time compression layer of the time weighted processing network model. In the time compression layer, the pressure time series data group is channel compressed by using a 1×1 convolution kernel to obtain reduced-dimensional pressure feature data. The purpose of channel compression is to reduce the redundancy of the data while maintaining important information and improving the efficiency of subsequent calculations. The time compression layer uses the ReLU activation function and batch normalization processing in the processing process to increase the nonlinear expression ability of the model, and standardizes the data to prevent the model training process from converging due to the scale difference of the input data.

[0045] The reduced-dimensional pressure feature data is input into the deep separation convolution unit in the time-weighted processing network model for deep separation convolution processing. Deep separation convolution is an effective convolution method that can process spatial and channel information separately, reduce computational complexity and improve model efficiency. In the deep separation convolution unit, it includes point-by-point convolution layers and deep convolution layers. The point-by-point convolution layer is used for cross-channel information fusion, while the deep convolution layer is used to extract spatial information. Through this series of operations, the spatial feature mapping matrix of the pressure data is obtained, which contains the spatial correlation information between each sensing unit.

[0046] The spatial feature mapping matrix is ​​input into the time attention module of the time weighted processing network model to calculate the time series correlation weights to obtain the time weighting coefficient matrix. In the time attention module, the self-attention mechanism is adopted. The self-attention mechanism can adaptively assign weights to different time points according to the importance of the data of each time step, thereby improving the model's sensitivity and expression ability to time information. In this way, the mutual influence of pressure data at different time points can be effectively captured, especially in complex pressure scenarios, which can highlight important time series features. Based on the time weighting coefficient matrix, the spatial feature mapping matrix is ​​weighted to obtain the time weighted feature matrix. The time weighted feature matrix enhances the representation ability of important time series features through weighted operations, while suppressing redundant or unimportant features.

[0047] The time-weighted feature matrix is ​​further input into the feature fusion module of the time-weighted processing network model for multi-scale feature combination. The feature fusion module contains three parallel feature extraction branches, which extract features of different scales on the input data respectively to obtain multi-scale feature fusion data. Channel attention processing is performed on the multi-scale feature fusion data to obtain a channel-weighted feature matrix. Channel attention processing strengthens the focus on specific channels by assigning different weights to different feature channels, so that the model can adaptively focus on important feature information and ignore irrelevant channel information.

[0048] The channel weighted feature matrix is ​​input into the feature reconstruction layer of the time-weighted processing network model, and bilinear interpolation upsampling is performed to obtain the pressure distribution feature set. Bilinear interpolation is an upsampling method that can effectively restore the resolution of the original features and retain the spatial information of the feature matrix. Multimodal feature analysis is performed on the pressure distribution feature set to obtain the final multimodal pressure distribution feature data. The relationship between different modes is comprehensively considered, and the distribution characteristics of pressure in space and time are described more comprehensively and accurately by fusing the information of different modes.

[0049] Specifically, the pressure distribution feature set is feature decomposed in the time dimension, space dimension and pressure amplitude dimension to obtain a multidimensional feature subset, including a time series feature subset, a space feature subset and a pressure feature subset. The time dimension feature subset is used to capture the changing law of the pressure signal at different time points, the space dimension feature subset is used to reflect the pressure distribution of the sensor element in space, and the pressure amplitude feature subset describes the distribution characteristics of pressure at different amplitudes.

[0050] Perform feature enhancement on the multidimensional feature subset to obtain an enhanced feature group. By performing operations such as amplification, smoothing and standardization on the features, each feature has stronger representation ability and recognition in subsequent processing. Perform multi-scale pyramid decomposition on the enhanced feature group to obtain a pyramid feature group. Pyramid decomposition is an effective multi-scale feature extraction method. By performing multi-level decomposition on the features, the performance of the features at different scales is captured, ensuring that not only the macroscopic overall features are considered during data analysis, but also the microscopic detailed features can be paid attention to. Perform multi-channel feature extraction on the pyramid feature group to obtain a multi-channel feature descriptor. The diversity and richness of features extracted through multiple independent feature channels enable the final feature descriptor to cover all important information of pressure distribution.

[0051] Based on the multi-channel feature descriptor, feature correlation analysis is performed to obtain the feature correlation matrix. By calculating the interdependence between features, we can understand which features have a greater impact on the pressure distribution and which features are redundant or strongly correlated. The feature sorting algorithm based on information gain performs feature screening on the feature correlation matrix to obtain the target feature set. Information gain is a criterion for measuring the importance of features, which can quantify the contribution of each feature in characterizing the target data. During the feature screening process, the feature sorting algorithm calculates and sorts the contribution of each feature, and selects features with higher information gain to enter the target feature set. Multimodal feature fusion is performed on the target feature set to obtain multimodal pressure distribution feature data, so that the fused feature data can fully reflect the pressure distribution.

[0052] Step 500: Input the multi-modal pressure distribution characteristic data into the closed-loop control system, update the pressure sampling parameters and the pressure processing parameters, and obtain a pressure acquisition control parameter set.

[0053] Specifically, the performance indicators of the multimodal pressure distribution characteristic data are calculated to obtain the performance evaluation data of the system. By analyzing the pressure distribution characteristics, various performance indicators of the system in the current state are extracted, including the accuracy, stability, response speed, etc. of the pressure response, so as to comprehensively evaluate the overall performance of the system. Parameter sensitivity analysis is performed based on the system performance evaluation data to obtain a parameter optimization weight sequence. Parameter sensitivity analysis quantifies the degree of influence of changes in various control parameters in the system on system performance. Through analysis, it is determined which parameters have a high sensitivity to the performance of the system and which parameter changes have a small impact on the system. Through analysis, the optimization priority of each parameter is sorted to obtain a parameter optimization weight sequence.

[0054] The parameter optimization weight sequence is weightedly calculated with the current pressure sampling parameters to obtain the updated sampling parameters. By multiplying and accumulating each sampling parameter with its corresponding optimization weight, the value of each parameter is effectively adjusted so that the updated sampling parameters are more in line with the current needs of the system. The updated sampling parameters are constrained to ensure that they are valid both physically and in system operation. The purpose of constraint processing is to prevent out-of-bounds or unreasonable values ​​from appearing after the parameters are updated, such as the sampling frequency is too high so that the hardware cannot process it, or some filter parameters are set unreasonably, resulting in signal distortion. Through constraint processing, an effective sampling parameter group is obtained, which can ensure the stability and effectiveness of the system in actual operation.

[0055] The pressure processing parameters of the effective sampling parameter group are updated to obtain the updated amount of the processing parameters. Based on the updated amount of the processing parameters, the filtering parameters and feature extraction parameters in the system are adjusted accordingly to obtain the updated processing parameter group. The adjustment of the filtering parameters can optimize the noise suppression ability of the signal, ensuring that useful information is retained in the collected data while removing unnecessary interference; while the update of the feature extraction parameters improves the accuracy of the system in extracting features in different scenarios, ensuring that the extracted features can fully reflect the actual situation of the pressure distribution.

[0056] The updated processing parameter group is partitioned and parameterized to obtain a partition parameter control table. According to the specific needs of different spatial and temporal regions, the processing parameters are reasonably allocated to each region to achieve targeted processing of pressure data in different regions. By establishing a partition parameter control table, the sampling and processing requirements of different regions are distinguished, so that the system can better adapt to the changes in pressure distribution in complex environments. For example, for areas with drastic pressure changes, higher sampling frequencies and stronger filtering capabilities are allocated, while for areas with relatively stable changes, lower sampling frequencies are allocated to save system resources. The pressure sampling parameters and pressure processing parameters in the partition parameter control table are integrated to obtain a pressure acquisition control parameter set. This control parameter set is the result of continuous iteration and optimization of the system in closed-loop control. By integrating the updated results of the sampling parameters and processing parameters, it is possible to realize partitioned acquisition and processing according to the specific needs of different spatial and temporal regions of the pressure sensor array during the pressure distribution data acquisition process, ensuring the high consistency and coordination of the system during the pressure acquisition and processing process.

[0057] In the embodiments of the present application, by introducing an adaptive event trigger mechanism and a partitioned sampling control strategy, data redundancy is significantly reduced and the efficiency of system resource utilization is optimized; a time-weighted processing network model is combined with a deep separation convolution unit for feature extraction, thereby enhancing the ability to capture the spatiotemporal characteristics of pressure distribution; an innovative multimodal feature analysis framework is constructed to achieve a comprehensive characterization of pressure distribution characteristics; a closed-loop control system based on performance evaluation is established, so that sampling parameters and processing parameters can be dynamically optimized and adjusted according to actual application requirements, effectively improving the robustness and adaptability of the system; through multi-level signal preprocessing and feature extraction strategies, the limitations of traditional methods in processing nonlinear and time-varying pressure signals are solved.

[0058] In a specific embodiment, the process of executing step 100 may specifically include the following steps:

[0059] Input a standard voltage signal to each pressure sensing unit in the pressure sensor array, and calculate the ratio of the output voltage value and the input voltage value of each pressure sensing unit to obtain an initial response coefficient matrix of each pressure sensing unit;

[0060] According to the initial response coefficient matrix, the difference between the maximum response value and the minimum response value of each pressure sensing unit is calculated to obtain the range normalization parameter of each pressure sensing unit, and the range normalization parameter of the pressure sensing unit is linearly operated with the initial response coefficient matrix to obtain the initial state matrix;

[0061] Performing a multi-order Butterworth filtering operation on the pressure signal collected by the pressure sensor array to obtain a filter coefficient set, and calculating the root mean square value and signal-to-noise ratio of the pressure signal based on the filter coefficient set to obtain a preprocessing parameter set;

[0062] Applying an increasing pressure load to the pressure sensor array, obtaining a pressure response data set by recording the output response of each pressure sensing unit under different pressure values, and performing piecewise polynomial fitting on the pressure response data set to obtain a pressure response characteristic curve;

[0063] The pressure response characteristic curve is stored in a matrix according to the position of the pressure sensing unit to obtain a response curve database of the pressure sensor.

[0064] Specifically, a standard voltage signal is input to each pressure sensing unit in the pressure sensor array, and the ratio of the output voltage value to the input voltage value of each pressure sensing unit is calculated to obtain the initial response coefficient matrix of each sensing unit. Assume that for a certain pressure sensing unit, the input voltage value is The output voltage is , then the initial response coefficient It is expressed as:

[0065] ;

[0066] in Indicates The initial response coefficient matrix is ​​obtained by performing the same operation on all the sensing units of the entire pressure sensor array. , where each element Indicates Line The initial response coefficients of the sensor units are listed. This matrix can reflect the reference response characteristics of the sensor under standard voltage conditions and is used for subsequent calibration and data processing. , calculate the difference between the maximum response value and the minimum response value of each pressure sensing unit to obtain the range normalization parameter. , assuming that its maximum response value is , the minimum response value is , then the range normalization parameter It is expressed as:

[0067] ;

[0068] Range normalization parameters The function of is to measure the response range of each sensor unit under different input conditions, so as to unify the output scale of each sensor unit. and the initial response coefficient matrix Perform linear operations to obtain the initial state matrix The process of linear operation is expressed as:

[0069] ;

[0070] in represents the first Line The elements of the columns are normalized to ensure that the outputs of all sensing units are compared within the same range, thereby reducing the deviation between different sensing units and making the entire pressure sensor array consistent in the initial state. After the initial state matrix is ​​established, a multi-order Butterworth filter operation is performed on the pressure signal collected by the pressure sensor array to obtain a filter coefficient set. The Butterworth filter is suitable for removing noise components in the signal to ensure that the obtained pressure data has a high signal-to-noise ratio. The transfer function of the Butterworth filter is expressed as:

[0071] ;

[0072] in represents the gain of the filter, is the angular frequency of the signal, is the cutoff frequency of the filter, is the order of the filter. By adjusting and , control the frequency characteristics of the filter to adapt to different pressure signals. The filtered signal is used to calculate the root mean square value and signal-to-noise ratio of the pressure signal. The root mean square value is expressed as:

[0073] ;

[0074] in represents the filtered signal, is the sampling period. The RMS value is used to measure the strength of the signal, while the signal-to-noise ratio is expressed as follows:

[0075] ;

[0076] in represents the signal power, Represents the noise power. By calculating the RMS value and the signal-to-noise ratio, a preprocessing parameter set for describing the signal quality is obtained. An increasing pressure load is applied to the pressure sensor array, and the output response of each sensor unit at each pressure value is recorded to obtain a pressure response data set. Assume that the pressure value applied to each sensor unit is , the corresponding output is , then the pressure response data set is expressed as In order to model these data more effectively, the pressure response data set is fitted using the piecewise polynomial fitting method to obtain the pressure response characteristic curve. Polynomial fitting is to establish an analytical relationship between pressure and sensor output, which is expressed in the following form:

[0077] ;

[0078] in are the fitted polynomial coefficients, is the applied pressure. Through segmented fitting, the fitting curve is better adapted to the nonlinear response characteristics of the sensor, thereby reducing the fitting error and improving the accuracy of the model. The pressure response characteristic curve is stored in a matrix according to the position of the pressure sensing unit to form a response curve database of the pressure sensor. The response curve database stores the output characteristics of each sensing unit under different pressure conditions in the form of a matrix. The matrix storage form enables the response characteristics of each sensing unit to be quickly searched and called in the subsequent system use to realize the correction and compensation of the pressure data. For example, each element in the response curve database Indicates Line The sensing unit of the column is under pressure Output response characteristics under .

[0079] In a specific embodiment, the process of executing step 200 may specifically include the following steps:

[0080] Extract the adjacent sampling point data of the initial state matrix, and divide the pressure difference of all pressure sensing units at adjacent moments by the sampling time interval to obtain a real-time change rate matrix;

[0081] Performing a convolution operation on the real-time change rate matrix and the filter coefficient set in the preprocessing parameter set to obtain filtered pressure change rate data;

[0082] Based on the filtered pressure change rate data, a pressure response characteristic curve corresponding to the pressure range is retrieved in a response curve database to obtain a pressure response correction coefficient, and a point multiplication operation is performed on the filtered pressure change rate data and the pressure response correction coefficient to obtain corrected pressure change rate data;

[0083] According to the corrected pressure change rate data, the ratio of the standard deviation of the pressure change rate to the mean is calculated to obtain a dynamic adjustment factor, and the dynamic adjustment factor is multiplied by the reference trigger threshold to obtain a pressure sampling dynamic trigger threshold;

[0084] The corrected pressure change rate data is compared with the pressure sampling dynamic trigger threshold. By judging whether each sensing unit meets the trigger condition, a sampling trigger flag matrix is ​​obtained. The sampling trigger flag matrix is ​​encoded and reconstructed according to the position of the pressure sensing unit to obtain a pressure sampling control matrix.

[0085] Specifically, the adjacent sampling point data of the initial state matrix are extracted and the pressure difference of each pressure sensing unit at adjacent moments is divided by the sampling time interval to obtain the real-time change rate matrix. Assume that the initial state matrix is ,in Indicates Row, No. The sensor unit is in time For two adjacent moments and , calculate the pressure difference matrix , whose elements It is expressed as:

[0086] ;

[0087] Divide the pressure difference of all sensing units by the sampling time interval , to obtain the real-time rate matrix ,in Indicates Row, No. Pressure change rate of the sensor unit:

[0088] ;

[0089] in, Reflects the sampling time interval of each sensor unit The real-time rate of change matrix can describe the dynamic change characteristics of pressure in a short period of time. The real-time rate of change matrix is ​​filtered to eliminate high-frequency noise and interference. The filter coefficient set in the preprocessing parameter set Perform convolution operation to obtain the filtered pressure change rate data matrix The convolution operation is expressed by the following formula:

[0090] ;

[0091] Among them, * represents the convolution operation, It is the coefficient set of the filter, which is used to smooth the pressure change rate data and eliminate unnecessary noise components. The filter selects a multi-order Butterworth filter, which has a good smoothing effect and can retain the main characteristics of the signal while suppressing high-frequency noise. The filtered pressure change rate data obtained by convolution , which is used to more accurately describe the pressure changes of the sensing unit at different time points. Based on the filtered pressure change rate data, the pressure response characteristic curve of the corresponding pressure range is retrieved in the response curve database to obtain the pressure response correction coefficient. Assume that the filtered pressure change rate data is For each sensing unit, find the corresponding pressure range in the response curve database according to the current pressure value and extract the pressure response characteristic curve. Let the response curve be ,in is the current pressure value, Indicates Row, No. The pressure sensor unit Through this response curve, we can get the correction coefficient matrix ,in Indicates Row, No. The correction coefficient of the sensor unit is calculated. and the pressure response correction coefficient matrix Perform dot multiplication to obtain the corrected pressure change velocity data matrix :

[0092] ;

[0093] in, Represents the multiplication of the elements of the matrix one by one (i.e. dot product). Represents the corrected pressure change rate, which can more accurately reflect the changes of each sensor unit under the current pressure state and eliminate the nonlinear response error of the sensor unit under different pressure conditions. According to the corrected pressure change rate data, the ratio of the standard deviation of the pressure change rate to the mean is calculated to obtain the dynamic adjustment factor. for The mean of is its standard deviation, then the dynamic adjustment factor It is expressed as:

[0094] ;

[0095] in, represents the average change rate of all sensor units, Indicates the degree of fluctuation of the change speed. Dynamic adjustment factor Reflects the stability of pressure changes. When it is larger, it means that there is a large fluctuation in pressure change, and the sampling frequency needs to be increased to capture more details; When it is small, it means that the pressure change is relatively stable, and the sampling frequency should be appropriately reduced. Based on the dynamic adjustment factor , and compare it with the reference trigger threshold Perform product operation to obtain the dynamic trigger threshold of pressure sampling :

[0096] ;

[0097] in, It is the sampling trigger threshold initially set by the system. By multiplying it with the dynamic adjustment factor, the sampling trigger condition is adaptively adjusted according to the actual pressure change. When it is larger, the dynamic trigger threshold will increase accordingly, thereby improving the sampling sensitivity. When smaller, It will also reduce unnecessary sampling operations. Dynamic trigger threshold with pressure sampling Perform comparison operations to determine whether each sensor unit meets the trigger conditions and obtain the sampling trigger flag matrix . Sample trigger flag matrix Elements It is expressed as:

[0098] ;

[0099] in, Indicates Row, No. The corrected change speed of the column sensor unit, A value of 1 indicates that the sensor unit meets the sampling trigger condition and needs to perform data sampling. A value of 0 indicates that the sensor unit does not need to perform sampling. According to the position of the pressure sensing unit, the encoding is reconstructed to obtain the pressure sampling control matrix The pressure sampling control matrix is ​​used to describe whether each sensor unit needs to be sampled at the current time, and according to the matrix The structure is used to encode the position so that the control system can quickly identify and execute the corresponding sampling operations.

[0100] In a specific embodiment, the process of executing step 300 may specifically include the following steps:

[0101] Performing a time-series state scan on the pressure sampling control matrix to obtain the initial sampling state of the pressure preparation acquisition interval;

[0102] Based on the initial sampling state of the pressure preparation acquisition interval, the pressure sampling dynamic trigger threshold is monitored to obtain an interval switching trigger signal;

[0103] Based on the interval switching trigger signal, the frequency is adjusted to obtain the sampling control signal of the pressure continuous acquisition interval, and the pressure change rate data in the pressure continuous acquisition interval is trend analyzed to obtain the change trend judgment result;

[0104] Based on the change trend judgment result, the sampling frequency is reduced to obtain the frequency control parameter of the pressure residual acquisition interval;

[0105] The frequency control parameters of the pressure residual acquisition interval are adjusted autonomously to obtain the dynamic frequency range of the pressure active acquisition interval;

[0106] The initial sampling state of the pressure preparation acquisition interval, the sampling control signal of the pressure continuous acquisition interval, the frequency control parameters of the pressure residual acquisition interval and the dynamic frequency range of the pressure active acquisition interval are modeled by state machines to obtain the pressure acquisition partition control model.

[0107] Specifically, the pressure sampling control matrix is ​​scanned in time sequence to determine the sampling state of each sensing unit at different time points and obtain the initial sampling state of the pressure preparation acquisition interval. Assume that the pressure sampling control matrix is ,in Indicates Row, No. The sensor unit is in time The sampling state of Scan time by time to determine which sensor units need to be sampled at different time points and which sensor units are ready to sample. Initial sampling state matrix It is a static snapshot extracted from the control matrix, indicating the sampling preparation status of each sensor unit in the pressure preparation acquisition interval. Based on the initial sampling state of the pressure preparation acquisition interval, the pressure sampling dynamic trigger threshold Monitor to determine whether the conditions for switching to the next acquisition interval are met. The pressure sampling dynamic trigger threshold represents the pressure change speed threshold for each sensor unit to sample. Assuming the current pressure change speed is , then the sampling state of each sensor unit is compared and Specifically, let Row, No. Switching trigger signal of the column sensor unit It is expressed as:

[0108] ;

[0109] in, Indicates that the current sensor unit meets the switching conditions and needs to enter the next acquisition interval, while This means that the current sampling state remains unchanged. By summarizing the switching trigger signals of all sensor units, the interval switching trigger signal matrix of the entire pressure sensor array is obtained. . Based on the interval switching trigger signal matrix , frequency adjustment is performed to obtain the sampling control signal of the continuous pressure acquisition interval. The continuous acquisition interval means that the sensor needs to increase the sampling frequency in order to accurately capture the rapidly changing pressure situation. Assume that the sampling frequency is , within the continuous acquisition interval, the frequency is adjusted so that it can be dynamically adjusted according to the speed of pressure change. The adjustment formula of the sampling frequency is expressed as:

[0110] ;

[0111] in, is the base sampling frequency, is the adjustment coefficient, is the average value of the interval switching trigger signal matrix, which is used to reflect the severity of the overall pressure change. Through this formula, the sampling frequency is dynamically adjusted according to the actual situation to achieve fine control of the pressure continuous acquisition interval. In the pressure continuous acquisition interval, the pressure change rate data is trend analyzed to obtain the pressure change trend judgment result. Assume that the pressure change rate data is The goal of trend analysis is to determine whether the current pressure change is gradually stabilizing or continuing to increase. To achieve this goal, the time derivative of the pressure change rate is calculated. Describe its changing trend:

[0112] ;

[0113] if If the value of tends to zero or is negative, it means that the pressure change tends to be stable or decrease; if If the value of is positive and large, it means that the pressure change is still increasing. Based on the trend judgment result, the next sampling strategy is determined. For example, if the pressure change gradually stabilizes, consider reducing the sampling frequency to reduce unnecessary data collection and enter the pressure residual collection interval. In the pressure residual collection interval, the sampling frequency is reduced to obtain the frequency control parameter . Reduce the burden on the system while ensuring that sufficient information is collected. Suppose the current frequency is , then the frequency after frequency reduction It is expressed as:

[0114] ;

[0115] in, is the frequency reduction factor, which is selected according to the smoothness of the pressure change. If the pressure change is very smooth, then The value of is small, which significantly reduces the sampling frequency; if there is still a certain fluctuation in the pressure change, then The value of should be relatively large to ensure that the sampling frequency is high enough. In the pressure residual acquisition interval, the frequency control parameter is adjusted autonomously to obtain the dynamic frequency range of the pressure active acquisition interval. According to different pressure changes, the upper and lower limits of the sampling frequency are flexibly adjusted. Assume that the dynamic frequency range is ,in and Respectively represent the lower and upper limits of the sampling frequency. By applying upper and lower limit constraints, we get:

[0116] ;

[0117] ;

[0118] in, and is the adjustment coefficient, is the minimum frequency at which the system can operate stably. Through adjustment, it is ensured that the sampling frequency is neither lower than the minimum requirement of the system nor too high to waste system resources. The initial sampling state of the pressure preparation acquisition interval, the sampling control signal of the pressure continuous acquisition interval, the frequency control parameter of the pressure residual acquisition interval, and the dynamic frequency range of the pressure active acquisition interval are modeled by a state machine to obtain a pressure acquisition partition control model. The construction of the state machine model can effectively describe the state transition relationship between the system in different pressure acquisition intervals. Assume that the state set of the state machine is , respectively represent the preparatory acquisition state, continuous acquisition state, residual acquisition state and active acquisition state. The state transition condition depends on the trigger signal, trend analysis results and frequency adjustment parameters mentioned above. State transition function It is expressed as:

[0119] ;

[0120] in, Indicates that the system is at time status, For time status, , and It is the main parameter that determines the state transition. In this way, the state machine performs orderly transitions between different acquisition intervals according to the real-time data and state information of the pressure sensor, thereby realizing the partition control of the entire pressure acquisition process.

[0121] In a specific embodiment, the process of executing step 400 may specifically include the following steps:

[0122] The target pressure data is collected through the pressure collection partition control model, and the target pressure data is segmented into time series to obtain a pressure time series data group;

[0123] The pressure time series data group is input into the time compression layer of the time weighted processing network model, and the pressure time series data group is channel compressed by the 1×1 convolution kernel of the time compression layer to obtain the reduced-dimensional pressure feature data. The time compression layer adopts the ReLU activation function and batch normalization processing;

[0124] The reduced-dimensional pressure feature data is input into the depth separation convolution unit of the time-weighted processing network model for depth separation convolution processing to obtain a spatial feature mapping matrix. The depth separation convolution unit includes a point-by-point convolution layer and a depth convolution layer.

[0125] The spatial feature mapping matrix is ​​input into the time attention module of the time weighted processing network model to calculate the time correlation weight, and the time weight coefficient matrix is ​​obtained. The time attention module adopts the self-attention mechanism;

[0126] Performing weighted operations on the spatial feature mapping matrix based on the time weighting coefficient matrix to obtain a time weighted feature matrix;

[0127] The time-weighted feature matrix is ​​input into the feature fusion module of the time-weighted processing network model for multi-scale feature combination to obtain multi-scale feature fusion data. The feature fusion module includes three parallel feature extraction branches.

[0128] Perform channel attention processing on the multi-scale feature fusion data to obtain a channel weighted feature matrix, and then input the channel weighted feature matrix into the feature reconstruction layer of the time weighted processing network model for bilinear interpolation upsampling to obtain a pressure distribution feature set;

[0129] A multimodal feature analysis is performed on the pressure distribution feature set to obtain multimodal pressure distribution feature data.

[0130] Specifically, the pressure data of the target area is obtained from the pressure sensor array according to the partition control model. The collected target pressure data changes with time, and these data are segmented into time series to obtain a pressure time series data group. Suppose the target pressure data is , Indicates time, Segment according to a certain time window, for example, using a length of The entire sequence is sliced ​​by the time window of , and the obtained segmented data is represented as ,in Indicates In this way, multiple pressure time series data sets are formed, each of which represents the pressure change characteristics in a specific time period. Input to the time compression layer of the time weighted processing network model. The purpose of the time compression layer is to reduce the number of channels of the data while retaining the main information features. In this process, a 1×1 convolution kernel is used to perform channel compression on the pressure time series data set. Suppose the input data is , whose dimensions are ,in is the number of channels, and are the height and width of the data. The convolution kernel changes the number of channels from Compress to a smaller value , to obtain the pressure characteristic data after dimension reduction This process is expressed as:

[0131] ;

[0132] in, is the weight matrix of the 1×1 convolution kernel, and * represents the convolution operation. The time compression layer uses the ReLU activation function and batch normalization. The expression of the ReLU activation function is:

[0133] ;

[0134] It is used to increase the nonlinear expression ability of the network, while batch normalization standardizes the data of each mini-batch to improve the training stability and convergence speed of the model. The input is sent to the deep separation convolution unit of the time-weighted processing network model for deep separation convolution processing. The deep separation convolution consists of two main parts: point-by-point convolution and depth convolution. The depth convolution performs spatial convolution on each channel of the input data. Assuming that the dimension of the input reduced-dimensional data is , the depthwise convolution will be applied to each channel The convolution kernel is used to obtain the feature representation of each channel. Point-by-point convolution uses a 1x1 convolution kernel to fuse all channels, combining the information of different channels to obtain a spatial feature mapping matrix. ,in The dimension is , is the number of new channels after point-by-point convolution. The input is sent to the time attention module of the time weighted processing network model to calculate the time series correlation weight. The function of the time attention module is to assign different weights to data at different time points to highlight important time series features, using the self-attention mechanism. Suppose the spatial feature mapping matrix is , the self-attention mechanism calculates the correlation between each pair of time points and calculates the similarity between time points through dot products:

[0135] ;

[0136] in, Indicates time point and The similarities between is a scaling factor used to prevent the dot product result from being too large. The similarity is normalized through the Softmax operation to obtain the time weighting coefficient matrix :

[0137] ;

[0138] Based on the time weighting coefficient matrix , for the spatial feature mapping matrix Perform weighted operations to obtain the time-weighted feature matrix :

[0139] ;

[0140] in, Represents matrix multiplication. Time-weighted feature matrix By weighting the importance of each time point, the ability to capture key time series features is enhanced. The feature fusion module is input into the time-weighted processing network model to perform multi-scale feature combination. The feature fusion module contains three parallel feature extraction branches, which are used to extract features at different scales. The convolution kernel sizes of the three branches are , then the features extracted by each branch are expressed as:

[0141] ;

[0142] in, They are the convolution kernel weight matrices of different branches. By combining the output features of the three branches, multi-scale feature fusion data is obtained. :

[0143] ;

[0144] in, The splicing operation of the representation feature. Combine the feature information of different scales to better capture the diverse characteristics of pressure distribution. Perform channel attention processing to obtain the channel weighted feature matrix Different weights are assigned to different characteristic channels, thereby highlighting the characteristic channels that have a greater contribution to the pressure distribution. Suppose the channel weight coefficient is , then the channel weighted feature matrix is ​​expressed as:

[0145] ;

[0146] in, The weight coefficient is obtained through the channel attention mechanism. The channel weighted feature matrix is ​​input into the feature reconstruction layer of the time weighted processing network model, and bilinear interpolation upsampling is performed to restore the spatial resolution of the feature to obtain the pressure distribution feature set. The process of bilinear interpolation upsampling is expressed as:

[0147] ;

[0148] Among them, UpSample( represents the bilinear interpolation upsampling operation, which is used to enlarge the spatial size of the feature matrix to match the size of the original input data. Multimodal feature analysis is performed to obtain multimodal pressure distribution feature data. The information of different characteristic modes, such as time features, spatial features, and pressure amplitude features, is integrated to fully describe the characteristics of pressure distribution. Suppose the result of multimodal feature fusion is ,but It is represented by weighted combination of different modal features:

[0149] ;

[0150] in, Represent the weight coefficients of time mode, space mode and pressure amplitude mode respectively, They represent the characteristic data of different modes respectively. By fusing these features, the multi-modal pressure distribution characteristic data obtained can more comprehensively reflect the dynamic change characteristics of pressure distribution.

[0151] In a specific embodiment, the process of performing a multi-modal feature analysis on the pressure distribution feature set to obtain multi-modal pressure distribution feature data may specifically include the following steps:

[0152] The pressure distribution feature set is feature decomposed in the time dimension, space dimension and pressure amplitude dimension to obtain a multidimensional feature subset, which includes a time series feature subset, a space feature subset and a pressure feature subset;

[0153] Perform feature enhancement on the multidimensional feature subset to obtain an enhanced feature group, and perform multi-scale pyramid decomposition on the enhanced feature group to obtain a pyramid feature group;

[0154] Perform multi-channel feature extraction on the pyramid feature group to obtain a multi-channel feature descriptor, and perform feature correlation analysis based on the multi-channel feature descriptor to obtain a feature correlation matrix;

[0155] A feature sorting algorithm based on information gain is used to screen the feature correlation matrix to obtain the target feature set, and multimodal feature fusion is performed on the target feature set to obtain multimodal pressure distribution feature data.

[0156] Specifically, the pressure distribution feature set is decomposed in the time dimension, space dimension and pressure amplitude dimension to obtain multidimensional feature subsets, including time series feature subsets, space feature subsets and pressure feature subsets, which can respectively describe the change characteristics of pressure signals in different aspects. Suppose the pressure distribution feature set is ,in is the time dimension, is the spatial dimension, Indicates the pressure amplitude. Perform feature decomposition to obtain a subset of time series features , indicating the pressure change at different time points; in the spatial dimension Perform feature decomposition to obtain spatial feature subsets , describing the spatial pressure distribution of the sensor array; performing feature decomposition in the pressure amplitude dimension to obtain the pressure feature subset , which is used to describe the amplitude change of pressure. , spatial feature subset and pressure feature subset Perform feature enhancement to obtain an enhanced feature group. Increase the representation ability of features through mathematical operations and data transformation. For example, perform autocorrelation analysis on a subset of time series features to enhance the ability of features to capture time patterns by calculating the similarity of features at different time lags. Suppose the autocorrelation function is , then for the time series characteristics , the autocorrelation function is expressed as:

[0157] ;

[0158] in, is the time lag, is the time length. Through the autocorrelation function, the enhanced time series feature group is obtained , retaining the temporal similarity and periodicity information. For the spatial feature subset , enhance the spatial variation characteristics through gradient operation, and set the gradient operator as , then the enhanced spatial feature set It is expressed as:

[0159] ;

[0160] This enhancement process can highlight the changes in spatial pressure gradients, especially the mutations and boundary features in space. By normalizing the amplitude of the pressure signal, the enhanced pressure feature group It is expressed as:

[0161] ;

[0162] in, and are the minimum and maximum values ​​of the pressure feature, respectively. Through normalization, the pressure feature varies within the range of [0,1] for subsequent processing. Multiscale pyramid decomposition is performed on the enhanced feature group to obtain a pyramid feature group. Multiscale pyramid decomposition is an effective image and signal processing method that can decompose features at different scales and capture the changes in pressure distribution at the macro and micro levels. Assume that the scale of the pyramid decomposition is , then in The features at each scale are expressed as , the pyramid decomposition is expressed as:

[0163] ;

[0164] in, For the The pyramid generation function of the layer, is the enhanced feature group. Through pyramid decomposition, feature representations of different scales are obtained, such as global trends at a coarse scale and local changes at a fine scale. Perform multi-channel feature extraction to obtain a multi-channel feature descriptor. Through multiple independent feature channels, the diversity and complexity of the pressure signal are extracted. Represents the number of feature channels, then each pyramid feature Through different convolution kernels, feature representations of different channels are obtained. ,in . Multi-channel feature descriptor Represented as the collection of all channel features:

[0165] ;

[0166] Through multi-channel feature extraction, feature descriptors at different scales and channels can be obtained. , perform feature correlation analysis to obtain a feature correlation matrix. Evaluate the interdependence between different features, remove redundant features and highlight features that have an important impact on pressure distribution. Suppose the two feature descriptors are and , and their correlation is calculated using the Pearson correlation coefficient:

[0167] ;

[0168] in, Representation characteristics and The covariance between and Characteristics and By calculating the correlation between all feature descriptors, we can get the feature correlation matrix ,in In order to optimize feature selection, the feature ranking algorithm based on information gain is used to screen the feature correlation matrix to obtain the target feature set. Information gain is used to measure the importance of a feature to the target variable. Suppose the target variable is , characterized by , then the feature Information gain It is expressed as:

[0169] ;

[0170] in, is the entropy of the target variable, representing the overall uncertainty of the system, For a given feature The larger the information gain, the greater the conditional entropy of the target variable. The greater the impact on the target variable. Sort and select features with larger information gain to form the target feature set . For the target feature set Multimodal feature fusion is performed to obtain multimodal pressure distribution feature data. The information from different modes is integrated so that the fused feature data can fully reflect the pressure distribution. Suppose the time series feature in the target feature set is , the spatial characteristics are , the pressure amplitude characteristic is The multimodal feature fusion is performed by weighted summation:

[0171] ;

[0172] in, is the fusion weight, which represents the importance weight of the time mode, space mode and pressure mode respectively. By adjusting these weights, the influence of different modes on the final feature representation is controlled, and the multi-modal pressure distribution feature data is obtained. .

[0173] In a specific embodiment, the process of executing step 500 may specifically include the following steps:

[0174] Calculate the performance index of the multi-modal pressure distribution characteristic data to obtain the system performance evaluation data, and perform parameter sensitivity analysis based on the system performance evaluation data to obtain the parameter optimization weight sequence;

[0175] Performing weighted calculation on the parameter optimization weight sequence and the current pressure sampling parameters to obtain updated sampling parameters, and performing constraint processing on the updated sampling parameters to obtain a valid sampling parameter group;

[0176] Performing pressure processing parameter update on the effective sampling parameter group to obtain a processing parameter update amount, and adjusting the filtering parameters and feature extraction parameters based on the processing parameter update amount to obtain an updated processing parameter group;

[0177] The updated processing parameter group is partitioned and parameter allocation is performed to obtain a partition parameter control table, and the pressure sampling parameters and pressure processing parameters in the partition parameter control table are integrated to obtain a pressure acquisition control parameter set.

[0178] Specifically, the performance index of the multi-modal pressure distribution characteristic data is calculated to obtain the system performance evaluation data, which reflects the performance of the system in terms of accuracy, response speed, stability, etc. Assume that the multi-modal pressure distribution characteristic data is , performance evaluation indicators include mean square error (MSE), signal-to-noise ratio (SNR) and delay time (Delay). Mean square error MSE is used to measure the error between actual collected data and ideal data, and is defined as:

[0179] ;

[0180] in, and Respectively represent The actual measured pressure value and ideal pressure value of each sampling point, is the total number of sampling points. The signal-to-noise ratio (SNR) is used to evaluate the quality of the signal and is defined as:

[0181] ;

[0182] in, and Delay is used to evaluate the time difference between the system receiving the input signal and generating a response, which is defined as Delay. ,in is the system response time, is the time of the input signal. By calculating these performance indicators, we can obtain the system performance evaluation data, which reflects the performance of the system in different aspects. Based on the performance evaluation data, we can perform parameter sensitivity analysis to determine the influence of each system parameter on the performance and obtain the parameter optimization weight sequence. Assume that the parameters involved in the system are ,in is the number of parameters. The goal of parameter sensitivity analysis is to determine the sensitivity of each parameter to the performance index and calculate the impact of a small change in each parameter on system performance. Suppose the performance evaluation function is , such as mean square error or signal-to-noise ratio, then the parameter Sensitivity It is expressed as:

[0183] ;

[0184] By normalizing the sensitivity of all parameters, we can get the parameter optimization weight sequence ,in:

[0185] ;

[0186] Parameter Optimization Weight Sequence Indicates the importance of each parameter in the overall performance optimization. The larger the weight, the greater the impact of the parameter on the system performance, so it should be given priority in the optimization process. With the current pressure sampling parameters Perform weighted calculation to obtain updated sampling parameters The updated sampling parameters are expressed as:

[0187] ;

[0188] in, is the learning rate, which controls the amplitude of parameter update. Through the weighted update process, the parameters that have a greater impact on system performance are optimized first, thus improving the overall performance of the system. In order to ensure that the updated sampling parameters are within a reasonable range, Perform constraint processing to obtain a valid sampling parameter set . Let the upper and lower limits of the parameter be and , then the constraint processing is expressed as:

[0189] ;

[0190] In this way, we can ensure that the parameters do not exceed the physical or system allowed range and obtain a valid sampling parameter set. Update the pressure processing parameters to obtain the update amount of the processing parameters. The pressure processing parameters include filtering parameters and feature extraction parameters, which directly affect the pressure data processing effect of the system. Suppose the filtering parameters are , the feature extraction parameters are , the processing parameter update amount is , then the update of processing parameters is expressed as:

[0191] ;

[0192] in, and Represents the update amount of the filtering parameters and feature extraction parameters, respectively, and is calculated based on the change amount in the effective sampling parameter group. For example, if the sampling frequency increases, the corresponding filtering parameters need to be adjusted to adapt to the higher frequency, and the feature extraction parameters also need to be adjusted to enhance the ability to capture high-speed changing signals. Perform partition parameter allocation to obtain a partition parameter control table. According to the pressure characteristics of different regions, the processing parameters are regionally optimized so that the system can better adapt to complex pressure distribution scenarios. Assume that the system is divided into partitions, and the parameters of each partition are , then the partition parameter control table is expressed as:

[0193] ;

[0194] Among them, each The pressure sampling parameters and the pressure processing parameters in the partition parameter control table are integrated to obtain the final pressure acquisition control parameter set. . The parameter integration process in the partition parameter control table mainly includes the following steps: the sampling parameters (including sampling frequency, trigger threshold, sampling time window length) and processing parameters (including filtering parameters, feature extraction parameters) of each partition are organized into parameter vectors. Then, a unified parameter structure is constructed, which includes two main parts: sampling parameter domain and processing parameter domain. For each partition, its sampling parameters and processing parameters are filled into the corresponding parameter domain respectively, and a parameter index table is established to record the position mapping relationship of each partition parameter in the overall parameter set. Add partition identifiers and timestamps to the parameter set, and set parameter update flags to realize the dynamic update mechanism of parameters. This integration method not only maintains the independence of partition parameters, but also realizes the unification of parameter management, which is convenient for the system to perform real-time parameter scheduling and control. The parameter set contains all sampling and processing control information of the system in different partitions, which is used to guide the entire pressure acquisition and processing process. Through the pressure acquisition control parameter set, the sampling frequency, filter parameters and feature extraction parameters are adaptively adjusted according to the real-time pressure changes, so as to achieve efficient pressure acquisition and processing.

[0195] The above describes the method for collecting pressure distribution data in the embodiment of the present application. The following describes the pressure distribution data collection device 10 in the embodiment of the present application. Figure 2 In one embodiment of the present application, a pressure distribution data acquisition device 10 includes:

[0196] The preprocessing module 11 is used to calibrate the pressure sensor array and preprocess the signal to obtain an initial state matrix, a preprocessing parameter set and a response curve database;

[0197] A calculation module 12 is used to calculate the pressure change rate based on the initial state matrix and the preprocessing parameter set, and to perform correction based on the response curve database to obtain the pressure sampling dynamic trigger threshold and the pressure sampling control matrix;

[0198] A control module 13 is used to perform partition control on the pressure data acquisition process according to the pressure sampling dynamic trigger threshold and the pressure sampling control matrix to obtain a pressure acquisition partition control model;

[0199] An extraction module 14 is used to collect target pressure data through a pressure collection partition control model, and input the target pressure data into a time-weighted processing network model for feature extraction to obtain multi-modal pressure distribution feature data;

[0200] The updating module 15 is used to input the multi-modal pressure distribution characteristic data into the closed-loop control system, update the pressure sampling parameters and the pressure processing parameters, and obtain the pressure acquisition control parameter set.

[0201] Through the coordinated cooperation of the above-mentioned components, by introducing an adaptive event trigger mechanism and partitioned interval sampling control strategy, data redundancy is significantly reduced and the system resource utilization efficiency is optimized; a time-weighted processing network model is combined with a deep separation convolution unit for feature extraction, which enhances the ability to capture the spatiotemporal characteristics of pressure distribution; an innovative multimodal feature analysis framework is constructed to achieve a comprehensive characterization of pressure distribution characteristics; a closed-loop control system based on performance evaluation is established, so that sampling parameters and processing parameters can be dynamically optimized and adjusted according to actual application requirements, effectively improving the robustness and adaptability of the system; through multi-level signal preprocessing and feature extraction strategies, the limitations of traditional methods in processing nonlinear and time-varying pressure signals are solved.

[0202] See also Figure 3 , Figure 3 This is a schematic block diagram of the structure of an electronic device 300 provided in an embodiment of the present application. The electronic device 300 includes a processor 301 and a memory 302. The processor 301 and the memory 302 are connected via a device bus 303, wherein the memory 302 may include a non-volatile storage medium and an internal memory.

[0203] The non-volatile storage medium can store a computer program. The computer program includes program instructions, and when the program instructions are executed by the processor 301, the processor 301 can execute any of the above-mentioned pressure distribution data collection methods.

[0204] The processor 301 is used to provide computing and control capabilities to support the operation of the entire electronic device 300 .

[0205] The internal memory provides an environment for the operation of the computer program in the non-volatile storage medium. When the computer program is executed by the processor 301, the processor 301 can execute any of the above-mentioned pressure distribution data collection methods.

[0206] Those skilled in the art will understand that Figure 3 The structure shown in the figure is only a block diagram of a partial structure related to the present application scheme, and does not constitute a limitation on the electronic device 300 involved in the present application scheme. The specific electronic device 300 may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0207] It should be understood that the processor 301 may be a central processing unit (CPU), and the processor 301 may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0208] It should be noted that those skilled in the art can clearly understand that, for the convenience and simplicity of description, the specific working process of the electronic device 300 described above can refer to the corresponding process of the aforementioned pressure distribution data collection method, which will not be repeated here.

[0209] An embodiment of the present application also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by one or more processors, the one or more processors implement the pressure distribution data collection method provided in the embodiment of the present application.

[0210] The computer-readable storage medium may be an internal storage unit of the electronic device 300 in the aforementioned embodiment, such as a hard disk or memory of the electronic device 300. The computer-readable storage medium may also be an external storage device of the electronic device 300, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped with the electronic device 300.

[0211] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0212] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable an electronic device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc., various media that can store program codes.

[0213] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for collecting pressure distribution data, characterized in that: The method comprises: Calibrate and preprocess the pressure sensor array to obtain an initial state matrix, a preprocessing parameter set, and a response curve database; The pressure change rate is calculated based on the initial state matrix and the preprocessing parameter set, and is corrected based on the response curve database to obtain a pressure sampling dynamic trigger threshold and a pressure sampling control matrix; specifically comprising: extracting adjacent sampling point data of the initial state matrix, and dividing the pressure difference of all pressure sensing units at adjacent moments by the sampling time interval to obtain a real-time change rate matrix; performing a convolution operation on the real-time change rate matrix and the filter coefficient set in the preprocessing parameter set to obtain filtered pressure change rate data; based on the filtered pressure change rate data, retrieving a pressure response characteristic curve of a corresponding pressure range in the response curve database to obtain a pressure response correction coefficient, and performing a dot multiplication operation on the filtered pressure change rate data and the pressure response correction coefficient to obtain corrected pressure change rate data; calculating the ratio of the standard deviation of the pressure change rate to the mean value according to the corrected pressure change rate data to obtain a dynamic adjustment factor, and performing a product operation on the dynamic adjustment factor and a reference trigger threshold to obtain a pressure sampling dynamic trigger threshold; performing a comparison operation on the corrected pressure change rate data and the pressure sampling dynamic trigger threshold, and obtaining a sampling trigger flag matrix by judging whether each sensing unit meets the trigger condition, and encoding and reconstructing the sampling trigger flag matrix according to the position of the pressure sensing unit to obtain a pressure sampling control matrix; According to the pressure sampling dynamic trigger threshold and the pressure sampling control matrix, the pressure data acquisition process is controlled in intervals to obtain a pressure acquisition partition control model; specifically, the pressure sampling control matrix is ​​scanned in time sequence to obtain an initial sampling state of the pressure preparation acquisition interval; based on the initial sampling state of the pressure preparation acquisition interval, the pressure sampling dynamic trigger threshold is monitored to obtain an interval switching trigger signal; based on the interval switching trigger signal, the frequency is adjusted to obtain a sampling control signal of the pressure continuous acquisition interval, and the pressure change rate data in the pressure continuous acquisition interval is trend analyzed to obtain a change trend judgment result; based on the change trend judgment result, the sampling frequency is down-sampled to obtain a frequency control parameter of the pressure residual acquisition interval; the frequency control parameter of the pressure residual acquisition interval is autonomously adjusted to obtain a dynamic frequency range of the pressure active acquisition interval; state machine modeling is performed on the initial sampling state of the pressure preparation acquisition interval, the sampling control signal of the pressure continuous acquisition interval, the frequency control parameter of the pressure residual acquisition interval and the dynamic frequency range of the pressure active acquisition interval to obtain a pressure acquisition partition control model; The target pressure data is collected through the pressure collection partition control model, and the target pressure data is input into the time-weighted processing network model for feature extraction to obtain multi-modal pressure distribution feature data; The multi-modal pressure distribution characteristic data is input into a closed-loop control system, and the pressure sampling parameters and pressure processing parameters are updated to obtain a pressure acquisition control parameter set.

2. The method for collecting pressure distribution data according to claim 1, characterized in that: The pressure sensor array is calibrated and the signal is preprocessed to obtain an initial state matrix, a preprocessing parameter set and a response curve database, including: Input a standard voltage signal to each pressure sensing unit in the pressure sensor array, and calculate the ratio of the output voltage value and the input voltage value of each pressure sensing unit to obtain an initial response coefficient matrix of each pressure sensing unit; According to the initial response coefficient matrix, the difference between the maximum response value and the minimum response value of each pressure sensing unit is calculated to obtain the range normalization parameter of each pressure sensing unit, and the range normalization parameter of the pressure sensing unit is linearly operated with the initial response coefficient matrix to obtain the initial state matrix; Performing a multi-order Butterworth filtering operation on the pressure signal collected by the pressure sensor array to obtain a filter coefficient set, and calculating the root mean square value and signal-to-noise ratio of the pressure signal according to the filter coefficient set to obtain a preprocessing parameter set; Applying an increasing pressure load to the pressure sensor array, obtaining a pressure response data set by recording the output response of each pressure sensing unit under different pressure values, and performing piecewise polynomial fitting on the pressure response data set to obtain a pressure response characteristic curve; The pressure response characteristic curve is stored in a matrix according to the position of the pressure sensing unit to obtain a response curve database of the pressure sensor.

3. The method for collecting pressure distribution data according to claim 1, characterized in that: The target pressure data is collected through the pressure collection partition control model, and the target pressure data is input into the time weighted processing network model for feature extraction to obtain multi-modal pressure distribution feature data, including: The target pressure data is collected by the pressure collection partition control model, and the target pressure data is segmented into time series to obtain a pressure time series data group; Inputting the pressure time series data group into the time compression layer of the time weighted processing network model, performing channel compression on the pressure time series data group through the 1×1 convolution kernel of the time compression layer to obtain reduced-dimensional pressure feature data, wherein the time compression layer adopts ReLU activation function and batch normalization processing; Inputting the reduced-dimensional pressure feature data into the depth separation convolution unit of the time-weighted processing network model for depth separation convolution processing to obtain a spatial feature mapping matrix, wherein the depth separation convolution unit includes a point-by-point convolution layer and a depth convolution layer; Inputting the spatial feature mapping matrix into the time attention module of the time weighted processing network model to calculate the time series correlation weight to obtain a time weighting coefficient matrix, wherein the time attention module adopts a self-attention mechanism; Performing a weighted operation on the spatial feature mapping matrix based on the time weighting coefficient matrix to obtain a time weighted feature matrix; Inputting the time-weighted feature matrix into the feature fusion module of the time-weighted processing network model to perform multi-scale feature combination to obtain multi-scale feature fusion data, wherein the feature fusion module includes three parallel feature extraction branches; Performing channel attention processing on the multi-scale feature fusion data to obtain a channel weighted feature matrix, and inputting the channel weighted feature matrix into the feature reconstruction layer of the time weighted processing network model for bilinear interpolation upsampling to obtain a pressure distribution feature set; A multimodal feature analysis is performed on the pressure distribution feature set to obtain multimodal pressure distribution feature data.

4. The method for collecting pressure distribution data according to claim 3, characterized in that: The performing multimodal feature analysis on the pressure distribution feature set to obtain multimodal pressure distribution feature data includes: Decomposing the pressure distribution feature set in the time dimension, space dimension and pressure amplitude dimension to obtain a multidimensional feature subset, wherein the multidimensional feature subset includes a time series feature subset, a space feature subset and a pressure feature subset; Performing feature enhancement on the multidimensional feature subset to obtain an enhanced feature group, and performing multi-scale pyramid decomposition on the enhanced feature group to obtain a pyramid feature group; Performing multi-channel feature extraction on the pyramid feature group to obtain a multi-channel feature descriptor, and performing feature correlation analysis based on the multi-channel feature descriptor to obtain a feature correlation matrix; A feature sorting algorithm based on information gain is used to perform feature screening on the feature correlation matrix to obtain a target feature set, and multimodal feature fusion is performed on the target feature set to obtain multimodal pressure distribution feature data.

5. The method for collecting pressure distribution data according to claim 4, characterized in that: The multi-modal pressure distribution characteristic data is input into a closed-loop control system, and the pressure sampling parameters and pressure processing parameters are updated to obtain a pressure acquisition control parameter set, including: Performing performance index calculation on the multimodal pressure distribution characteristic data to obtain system performance evaluation data, and performing parameter sensitivity analysis based on the system performance evaluation data to obtain a parameter optimization weight sequence; Performing weighted calculation on the parameter optimization weight sequence and the current pressure sampling parameters to obtain updated sampling parameters, and performing constraint processing on the updated sampling parameters to obtain a valid sampling parameter group; Performing pressure processing parameter update on the effective sampling parameter group to obtain a processing parameter update amount, and adjusting filtering parameters and feature extraction parameters based on the processing parameter update amount to obtain an updated processing parameter group; Partition parameters are allocated to the updated processing parameter group to obtain a partition parameter control table, and the pressure sampling parameters and pressure processing parameters in the partition parameter control table are integrated to obtain a pressure acquisition control parameter set.

6. A pressure distribution data acquisition device, characterized in that: Used to perform the pressure distribution data collection method according to any one of claims 1 to 5, the pressure distribution data collection device comprising: A preprocessing module is used to calibrate the pressure sensor array and preprocess the signal to obtain an initial state matrix, a preprocessing parameter set and a response curve database; A calculation module, used to calculate the pressure change rate based on the initial state matrix and the preprocessing parameter set, and to perform correction based on the response curve database to obtain a pressure sampling dynamic trigger threshold and a pressure sampling control matrix; A control module, used to perform partition-based control on the pressure data acquisition process according to the pressure sampling dynamic trigger threshold and the pressure sampling control matrix to obtain a pressure acquisition partition control model; An extraction module, used for collecting target pressure data through the pressure collection partition control model, and inputting the target pressure data into the time-weighted processing network model for feature extraction to obtain multi-modal pressure distribution feature data; The updating module is used to input the multi-modal pressure distribution characteristic data into a closed-loop control system, update the pressure sampling parameters and the pressure processing parameters, and obtain a pressure acquisition control parameter set.

7. An electronic device, characterized in that: The electronic device comprises: a memory and at least one processor, wherein instructions are stored in the memory; The at least one processor calls the instructions in the memory to enable the electronic device to execute the method for collecting pressure distribution data as described in any one of claims 1-5.

8. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instructions are executed by the processor, the method for collecting pressure distribution data according to any one of claims 1 to 5 is implemented.

Images