Real-time correction method for crosstalk phenomenon of fabric pressure distribution sensing system

Abstract

The invention relates to a real-time correction method for a crosstalk phenomenon of a fabric pressure distribution sensing system, which relates to the technical field of flexible sensors and artificial intelligence, and comprises the following steps: receiving data corresponding to a pressure distribution data matrix; converting the data into a two-dimensional matrix format, displaying a pressure distribution cloud picture in real time, and storing the pressure distribution cloud picture; constructing a convolutional neural network model, and modifying the convolutional neural network model into a regression prediction model; receiving a pressure distribution data matrix without a crosstalk phenomenon, and storing a pressure distribution cloud picture; constructing a training set by taking the pre-stored pressure distribution cloud picture as an input image and taking the post-stored pressure distribution cloud picture as an output image; training by using the training data set until convergence is achieved; the image is corrected; and displaying the image. The method is simple and high in real-time performance, the output pressure distribution cloud picture result is accurate, and the influence of the crosstalk phenomenon on the pressure distribution cloud picture display result obtained by the fabric pressure sensing array can be effectively improved in real time.

Description

Technical Field

[0001] The present invention relates to the field of flexible sensors and artificial intelligence technology, and in particular to a method for real-time correction of crosstalk phenomenon in a fabric pressure distribution sensing system. Background Art

[0002] In the field of modern health monitoring and medical diagnosis, flexible fabric pressure sensor arrays and their sensing systems are gaining increasing attention. Due to their excellent flexibility and wearability, these sensor arrays can capture the pressure distribution of the human body in real time, thereby providing users with important physiological status information. The design concept of flexible fabric pressure sensor arrays is to better adapt to the curves of the human body and provide a more comfortable and natural monitoring experience. They can be embedded in clothing, insoles, and even bandages, making the monitoring process almost unnoticeable while continuously collecting data.

[0003] However, in practical applications, due to the flexibility of the fabric itself, when one or more sensing units in the fabric pressure sensing array are pressed and turned on, the surrounding unpressurized sensing units will also be turned on, causing the fabric pressure sensing array to often produce "artifacts", that is, crosstalk. This phenomenon is similar to the situation in electronic devices where the signal from one signal source interferes with another signal source, resulting in inaccurate data reading. In the field of medical monitoring, this may lead to misjudgment of the patient's status, and crosstalk may cause erroneous readings, thereby affecting diagnosis and treatment decisions.

[0004] This not only affects the accuracy of the data, but also limits its application in dynamic monitoring. Dynamic monitoring refers to the ability to continuously monitor the patient's physiological state while they perform daily activities such as walking, running or doing other exercises. The existence of crosstalk makes the data obtained under these dynamic conditions unreliable, limiting the scope of application of flexible fabric pressure sensing arrays in actual medical monitoring. Traditional crosstalk elimination methods are mostly offline processing, lack real-time performance, and cannot respond to rapidly changing pressure environments in a timely manner. Summary of the invention

[0005] In view of the above problems, a method for real-time correction of the crosstalk phenomenon of the fabric pressure distribution sensing system is proposed, so as to eliminate the problem of inaccurate pressure distribution cloud map display results output by the fabric pressure sensing array caused by the crosstalk phenomenon in real time, and the result accuracy is high.

[0006] In order to achieve the above object, the technical solution of the present invention provides a real-time correction method for crosstalk phenomenon in a fabric pressure distribution sensing system, comprising the following steps:

[0007] Receive data corresponding to the pressure distribution data matrix;

[0008] After the data format of the received data is converted into a two-dimensional matrix format, a pressure distribution cloud map is displayed in real time, and then the image is stored;

[0009] Construct a convolutional neural network model and modify the convolutional neural network model into an image-to-image regression prediction model;

[0010] Receive the corresponding data of the pressure distribution data matrix without crosstalk phenomenon / with accurate detection results, and store the corresponding pressure distribution cloud map;

[0011] The pressure distribution cloud map stored first is used as the input image of the convolutional neural network model, and the pressure distribution cloud map stored later is used as the output image of the convolutional neural network model to construct a training set for improving the training of the convolutional neural network model;

[0012] Using the training data set to train the improved convolutional neural network model until convergence;

[0013] The previously stored image to be corrected is transferred to the model for correction;

[0014] The corrected image is displayed.

[0015] Preferably, a U-Net convolutional neural network model is used.

[0016] Preferably, a U-Net convolutional neural network model is constructed and programmed in Matlab software.

[0017] Preferably, the fabric pressure sensing array is placed directly above a pressure sensing array with no crosstalk and accurate detection results to collect a pressure distribution data matrix with no crosstalk / accurate detection results, and simultaneously collect pressure distribution cloud maps under the same pressure multiple times.

[0018] Preferably, the method further includes the steps of: building a fabric pressure distribution sensing system for collecting a pressure distribution data matrix, wherein the fabric pressure distribution sensing system includes a fabric pressure sensing array, a data processing module, a wireless transmission module, a power supply module and a computer.

[0019] Preferably, the fabric pressure distribution sensing system collects the pressure distribution data matrix and converts it into a corresponding electrical signal data matrix, converts the analog signal into a digital signal, and then sends the data to the server.

[0020] Preferably, before correcting the image, the trained improved U-Net convolutional neural network model is exported as an M file in Matlab software, and the M file is loaded and the Matlab parser is called using the Matlab Script node in LabVIEW.

[0021] Preferably, the corrected image is transferred from Matlab back to LabVIEW before displaying the corrected image.

[0022] The beneficial effects of the present invention are as follows: a method for real-time correction of the crosstalk phenomenon of a fabric pressure distribution sensor system of the present invention utilizes a convolutional neural network method to correct the crosstalk phenomenon of a fabric pressure sensor array, the method of the present invention can simplify the structure of the circuit, and the real-time correction of the crosstalk phenomenon is achieved by hybrid programming of LabVIEW software and Matlab software. The method is simple and has high real-time performance. The output pressure distribution cloud map result is accurate, and the influence of the crosstalk phenomenon on the pressure distribution cloud map display result obtained by the fabric pressure sensor array can be effectively improved in real time. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 It is a schematic diagram before correction of the method for real-time correction of the crosstalk phenomenon of the fabric pressure distribution sensing system of the present invention;

[0024] Figure 2 It is a corrected schematic diagram of the method for real-time correction of the crosstalk phenomenon of the fabric pressure distribution sensing system according to 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 only 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 present invention discloses a method for real-time correction of crosstalk phenomenon in a fabric pressure distribution sensing system, which comprises the following steps:

[0027] (1) Build a fabric pressure distribution sensing system, which consists of several key parts: a fabric pressure sensing array, a data processing module, a wireless transmission module, a power module and a computer. The structural design and preparation method of the fabric pressure sensing array can refer to the Chinese invention patent application with publication number CN 116839767 A. In this embodiment, the fabric pressure sensing array includes 32×32 sensing units, each of which can sense pressure changes. The design of the data processing module, the wireless transmission module and the power module can refer to the Chinese invention patent application with publication number CN 116999056 A, which work together to ensure accurate data collection and transmission.

[0028] (2) When a human body stands on the fabric pressure sensor array, the pressure on the sole of the foot will act on the sensor unit. The scanning circuit of the data processing module will collect the pressure distribution data matrix on the entire fabric pressure sensor array at one time. The microcontroller of the data processing module converts the collected pressure distribution data matrix into a corresponding electrical signal data matrix according to the pressure-electrical signal response characteristics of the fabric pressure sensor array, and converts the analog signal into a digital signal. Then, these data are sent to the server through the wireless transmission module. In this embodiment, the data transmission protocol adopted is the message queue telemetry transmission (MQTT) protocol based on the TCP / IP protocol cluster to ensure efficient data transmission.

[0029] (3) After receiving the data distributed by the server, the computer converts the data format into a two-dimensional matrix format and displays the pressure distribution cloud map in real time, such as Figure 1 This cloud map can visually display the pressure distribution, and the image data will be stored for subsequent processing and analysis.

[0030] (4) Construct a U-Net convolutional neural network model in Matlab software and write the corresponding program code. U-Net is a commonly used image segmentation network. It is modified into an image-to-image regression prediction model to meet the needs and obtain an improved U-Net convolutional neural network model.

[0031] (5) In order to obtain accurate training data, the fabric pressure sensor array is placed directly above a pressure sensor array without crosstalk and with accurate detection results, and the pressure distribution cloud map under the same pressure is collected multiple times. Specifically, this embodiment adopts the RX-M3232L piezoresistive flexible film pressure array sensor pad, which is based on a polyester film substrate and has 32×32 sensor units. The sensor units are composed of a variety of conductive carbon-based materials and a variety of polymer resins mixed in a certain proportion, and each sensor point is independently designed with little interference between each other. Referring to the methods in steps (2) and (3), the pressure distribution image is collected and stored.

[0032] (6) The pressure distribution cloud map obtained in step (3) is used as the input image of the improved U-Net convolutional neural network model established in step (4), and the pressure distribution cloud map obtained in step (5) under the same pressure collected by the pressure sensor array without crosstalk and with accurate detection results is used as the output image of the improved U-Net convolutional neural network model, and a training data set for training the improved U-Net convolutional neural network model is constructed. Among them, the input and output data sets respectively collect 2160 left foot standing pressure images, 2160 right foot standing pressure images, and 2400 two feet standing pressure images, and the input and output data sets each contain 6720 images.

[0033] (7) The input and output image data are divided into training set, validation set and test set in a ratio of 8:1:1, and the improved U-Net convolutional neural network model is trained using the training data set until convergence is achieved. During the training process, in order to ensure effective learning and convergence of the model, the network model setting parameters are set as follows: the solver is Adam, the learning rate is 0.001, the maximum number of iterations is 240, the Mini-batch size is 128, and the L2 Regularization is 0.0001.

[0034] (8) In Matlab software, export the trained improved U-Net convolutional neural network model as an M file. In LabVIEW, use the Matlab Script node to load this M file and call the Matlab parser to pass the real-time image to be corrected obtained in step (3) to the loaded model for correction.

[0035] (9) The corrected image obtained in step (8) is transferred from Matlab back to LabVIEW, and the corrected image is displayed on the computer interface, such as Figure 2 In this way, the real-time correction of the crosstalk phenomenon in the fabric pressure distribution sensing system is completed.

[0036] The present invention relates to a method for real-time correction of the crosstalk phenomenon of a fabric pressure distribution sensor system, constructing an improved U-Net convolutional neural network model, and adopting a mixed programming method of Matlab software and LabVIEW software to perform real-time correction on the obtained pressure distribution image. The method of the present invention is simple, has high real-time performance, and outputs accurate pressure distribution cloud map results, and can effectively and effectively improve the influence of the crosstalk phenomenon on the pressure distribution cloud map display results obtained by the fabric pressure sensor array in real time.

[0037] Finally, it should be noted that the above is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A real-time correction method for crosstalk phenomenon in a fabric pressure distribution sensing system, characterized in that: The following steps are involved: Receive data corresponding to the pressure distribution data matrix; After the data format of the received data is converted into a two-dimensional matrix format, a pressure distribution cloud map is displayed in real time, and then the image is stored; Construct a convolutional neural network model and modify the convolutional neural network model into an image-to-image regression prediction model; Receive the corresponding data of the pressure distribution data matrix without crosstalk phenomenon / with accurate detection results, and store the corresponding pressure distribution cloud map; The pressure distribution cloud map stored first is used as the input image of the convolutional neural network model, and the pressure distribution cloud map stored later is used as the output image of the convolutional neural network model to construct a training set for improving the training of the convolutional neural network model; Using the training data set to train the improved convolutional neural network model until convergence; The previously stored image to be corrected is transferred to the model for correction; The corrected image is displayed.

2. A real-time correction method for crosstalk phenomenon in a fabric pressure distribution sensing system according to claim 1, characterized in that: The U-Net convolutional neural network model is used.

3. A real-time correction method for crosstalk phenomenon in a fabric pressure distribution sensing system according to claim 2, characterized in that: Build the U-Net convolutional neural network model in Matlab software and write its program.

4. A real-time correction method for crosstalk phenomenon in a fabric pressure distribution sensing system according to claim 3, characterized in that: The fabric pressure sensing array is placed directly above the pressure sensing array with no crosstalk and accurate detection results to collect the pressure distribution data matrix with no crosstalk / accurate detection results, and at the same time, the pressure distribution cloud diagram under the same pressure is collected multiple times.

5. A real-time correction method for crosstalk phenomenon in a fabric pressure distribution sensing system according to claim 4, characterized in that: The method also includes the steps of: building a fabric pressure distribution sensing system for collecting a pressure distribution data matrix, wherein the fabric pressure distribution sensing system includes a fabric pressure sensing array, a data processing module, a wireless transmission module, a power supply module and a computer.

6. A real-time correction method for crosstalk phenomenon in a fabric pressure distribution sensing system according to claim 5, characterized in that: The fabric pressure distribution sensing system collects the pressure distribution data matrix and converts it into the corresponding electrical signal data matrix. After converting the analog signal into a digital signal, the data is sent to the server.

7. A real-time correction method for crosstalk phenomenon in a fabric pressure distribution sensing system according to claim 6, characterized in that: Before correcting the image, the trained improved U-Net convolutional neural network model was exported as an M file in Matlab software, and the Matlab Script node was used in LabVIEW to load the M file and call the Matlab parser.

8. A real-time correction method for crosstalk phenomenon in a fabric pressure distribution sensing system according to claim 7, characterized in that: Before displaying the corrected image, pass the corrected image back from Matlab to LabVIEW.

Images