UWB-based non-contact textile fabric humidity detection method

Through the UWB-based non-contact textile humidity detection method, UWB signal and electromagnetic wave propagation theory, combined with antenna array and wavelet denoising technology, the problems of traditional humidity sensors in terms of accuracy and interference are solved, and high-precision and non-invasiveness are achieved. Humidity detection, suitable for smart textiles.

CN119064384APending Publication Date: 2024-12-03DALIAN UNIV OF TECH

Patent Information

Application Number
CN202411187936.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-28
Publication Date
2024-12-03

AI Technical Summary

Technical Problem

Traditional contact humidity sensors have shortcomings in detection accuracy, environmental interference and equipment maintenance, and it is difficult to maintain the original characteristics of textiles, and lack contactless detection technology to efficiently perceive humidity changes.

Method used

Using UWB-based non-contact textile humidity detection method, we can construct a theoretical model of textile moisture content and wireless signal strength, and use electromagnetic wave propagation theory and Friis transmission formula, combined with antenna array and wavelet denoising technology to achieve high precision and non-invasiveness. humidity detection eliminates environmental interference and improves distance measurement accuracy.

Benefits of technology

Achieve non-contact, non-invasive and high-precision textile humidity detection, providing a new theoretical model between the moisture content of the bridged textile and wireless signal, eliminating environmental noise interference, improving equipment distance measurement accuracy and eliminating fixed distance limitations. In intelligent textile humidity sensing technology.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119064384A_ABST
    Figure CN119064384A_ABST
Patent Text Reader

Abstract

The invention provides a non-contact textile fabric humidity detection method based on UWB (Ultra Wideband). According to the system, UWB signals are used for sensing, and an innovative signal processing method is introduced to cope with the challenges that fabric reflection signals are weak and are prone to noise interference. In order to reduce dependence on a large number of data sets, a model-driven method is adopted. A bottom layer physical relation that the humidity of the textile fabric affects the radio-frequency signal is searched, and a theoretical model of bridging radio-frequency signal characteristics and the moisture content of the textile fabric is established; based on the model, through fine signal processing and parameter calibration, the system successfully eliminates the influence of interference factors such as a target distance and target attributes. The method realizes high-precision fabric moisture measurement, and can be widely applied to the fields of intelligent clothing, wound care, textile manufacturing and the like.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the field of wireless radio frequency signal perception, and in particular relates to a UWB-based non-contact textile humidity detection method. Background Art

[0002] With the rapid development of smart textiles, textile humidity sensors are increasingly being used in healthcare, sports monitoring, and industrial monitoring. However, traditional humidity sensors usually rely on changes in capacitance or resistance to detect humidity, but these contact methods have many deficiencies in detection accuracy, environmental interference, equipment maintenance, and are prone to sample damage. The current market lacks a detection technology that can both efficiently sense humidity changes and maintain the original properties of textiles.

[0003] Ultra-wideband (UWB) technology, as an emerging wireless communication technology, has the advantages of high frequency bandwidth, low power consumption and high precision. It has been widely used in indoor positioning and wireless sensing and other fields (patent publication number: CN113038597A, CN113453147A). However, its application in textile humidity detection is still in the exploratory stage. The present invention proposes a fine-grained textile humidity sensing method based on commercial ultra-wideband UWB. The method utilizes the influence of the hygroscopicity of textile materials on UWB signals to achieve high-precision detection of humidity. Compared with the prior art, the present invention has the advantages of non-invasiveness, high precision, real-time detection and rapid response. It provides new ideas for the development of intelligent textile humidity sensing technology and has broad application prospects. Summary of the invention

[0004] The technical purpose of the present invention is to provide a UWB-based non-contact textile humidity detection method.

[0005] The technical solution of the present invention:

[0006] A non-contact textile humidity detection method based on UWB is as follows:

[0007] Step 1: Construct a theoretical model to bridge textile moisture content and wireless signal strength:

[0008] 1.1. The relationship between the dielectric constant of a textile and its water content is expressed by the following equation:

[0009] ε′ r (W) = β 1 W 3 +β 2 W 2 +β 3 W+β 4 (1)

[0010] where ε′ ris the real part of the relative dielectric constant of the textile, W is the moisture content; β 1 , β 2 , β 3 , β 4 is the coefficient. For different textile materials, the formula format is the same, but the coefficient is different;

[0011] 1.2. Based on the electromagnetic wave propagation theory and the Friis transmission formula, the reflection model of the electromagnetic wave signal is simplified as follows:

[0012]

[0013] Among them, |A t | and |A r | respectively represent the amplitude of the transmitted and received signals, G t and G r Respectively represent the gains of the transmitting and receiving antennas; λ represents the wavelength of the UWB signal, d represents the distance from the antenna to the target, and R represents the reflection coefficient of the target;

[0014] 1.3. Analysis of the reflection and refraction of electromagnetic waves at the boundaries of various materials relies on applying reasonable boundary conditions to the plane equations to derive the reflection coefficient:

[0015]

[0016] Where Z 1 and Z 2 Respectively represent the impedance of the two media of air and textile; the inherent impedance and dielectric constant satisfy the following relationship: Where Z 0 is the free space impedance, Z is Z 1 or Z 2 ; Then we can get:

[0017]

[0018] where ∈ 1 Represents the dielectric constant of air, which is 1, ∈ 2 represents the dielectric constant of the textile;

[0019] 1.4. Substitute formulas (2)-(4) layer by layer and combine them with formula (1) to establish the relationship equation between the humidity level of textile materials and signal characteristics:

[0020]

[0021] Among them, S r Indicates the UWB signal receiving signal strength. Related to water content; β iand γ are constants related to the type of textiles; based on the observed data collected from different textile materials, the corresponding parameters in the formula were estimated using the nonlinear least squares fitting method, and the statistical goodness of fit was evaluated using the r-squared and root mean square error;

[0022] Step 2: Robust signal processing for irrelevant noise and dependency elimination:

[0023] 2.1. Wireless signal preprocessing and noise elimination:

[0024] 2.1.1. Before the detection, the reflection signal of the empty environment without the textile target is collected in advance. This operation only needs to be collected once before the actual textile humidity detection. In the subsequent detection process of textile humidity, the reflection signal collected when the textile target exists and the reflection signal of the empty environment collected in advance are firstly used for background difference algorithm. Since the only variable in the collected signal before and after the presence of the textile is the textile, the reflection signals of other objects and layout settings in the environment remain unchanged except for the change of the reflection signal caused by the newly introduced textile.

[0025] 2.1.2. After collecting signals for a period of time, the signal transmission quality of different channels is determined by calculating the coefficient of variation of the received signals of different transceiver channels in the antenna array; selecting the most stable N transceiver channels can avoid the influence of some irrelevant environment and hardware interference;

[0026] According to the spatial position of the antenna corresponding to the selected transceiver channel, beamforming technology is used to further concentrate the reflected signals from the target direction and further suppress irrelevant interference from other directions; the specific method is as follows:

[0027] The reflected signal of the target is incident on the antenna array in parallel and received by the receiving antenna. Due to the different positions of the antennas, the signal reaches each antenna with different time delays. The first antenna in the array is selected as the reference, and according to the spatial position of the antenna and the phase difference of the received signal, the following weight expression is constructed to compensate for the phase difference caused by the different time delays of different antennas.

[0028]

[0029] Where i = 1, 2, ..., n, where n is the number of antennas in the array, Δφ i (θ) represents the phase difference of the received signal corresponding to the spatial position of the antenna, θ represents the azimuth of the target, λ is the wavelength of the signal, and l is the length between adjacent antennas in the array;

[0030] Then the beamforming weights of different antennas in the array are obtained, denoted as j represents an imaginary unit; based on the above weights, the array composed of the selected channels can be beamformed.

[0031]

[0032] Among them, N is the number of stable channels, ω i is the weight vector, X i is the signal vector of the selected channel; the beamforming result reduces the influence of reflected signals in directions not of interest, making the moisture measurement more accurate;

[0033] Finally, the wavelet denoising method is used to filter out the environmental noise to avoid affecting the weak reflection signal of the textile. Wavelet denoising is mainly divided into the following three steps: Feature extraction: specify the wavelet decomposition level K, and perform wavelet decomposition on the signal to obtain frequency components of different levels and their corresponding wavelet coefficients; Low-pass filter: select a suitable threshold for each level of coefficients obtained by wavelet decomposition to suppress noise in the high-frequency area; Signal reconstruction: apply wavelet inverse transform to the frequency components that have been thresholded to reconstruct the signal and obtain the effective signal after denoising;

[0034] 2.2. Elimination of irrelevant dependencies:

[0035] 2.2.1. The received signal strength of wireless signals is closely related to the propagation distance. In order to make the proposed system practical, the limitation of distance dependence should be eliminated. The propagation loss of RF signals is expressed by the Friis free space loss equation:

[0036]

[0037] Among them, S t and S r is the received signal strength of the transmitted signal, and is the gain of the transmitting and receiving antennas in the dB domain;

[0038] Distance estimation: First, the sampling points of the device are increased, so that the device has twice as many sampling points in the same time, and a finer range-bin division is obtained; then, the signal of the bin where the target is located is interpolated using cubic spline to further improve the accuracy;

[0039] 2.2.2. Benchmark experiments were conducted to analyze the impact of different textile characteristics on the reflection signal, including material, size, shape and thickness. Corresponding experiments were conducted on these four characteristics with strict control of variables. The results showed that the impact of size and shape on the reflection signal was minimal under non-extreme conditions. As for the material factor, due to the different chemical materials that make up the finished product, although different materials exhibit different dielectric properties at the same humidity, the overall trend of change in dielectric capacity as the humidity level increases is the same. Therefore, for different materials, it is only necessary to estimate the corresponding parameters to complete the calibration. As for the thickness factor, the greater the thickness, the stronger the dielectric property. If the thickness change is abstracted as the superposition of textiles with unit thickness without layers, the problem can be simplified to a multi-layer reflection problem, which is reflected in the increase of the reflection coefficient R. Therefore, a correlation coefficient η is added to R in the model to obtain the final model.

[0040] Beneficial effects of the present invention:

[0041] 1) Through the UWB-based non-contact textile humidity detection method provided by the present invention, non-contact, non-invasive, high-precision, and high-response textile humidity detection can be achieved;

[0042] 2) A new theoretical model that bridges the gap between textile moisture content and wireless signals is provided, which can be used to explore the underlying physical principles of how changes in textile moisture content affect wireless signals;

[0043] 3) Fine signal processing method to solve the problem that textiles have a weak impact on signals due to their weak polarity, eliminate the interference of reflections from irrelevant objects in the environment, use antenna arrays to focus on enhancing signal reflections in specific directions, and eliminate the impact of high-frequency noise brought by the environment;

[0044] 4) Improve the ranging accuracy of the equipment at the software level, while eliminating the limitation of fixed distance to meet flexible detection needs. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 This is a schematic diagram of a UWB-based non-contact textile humidity detection system of the present invention.

[0046] Figure 2 Schematic diagram of the underlying principle of the textile of the present invention absorbing and retaining moisture and its influence on signals.

[0047] Figure 3 The present invention is a schematic diagram of the system structure of a UWB-based non-contact textile humidity detection method. Figure 4 The figure is a schematic diagram of the background difference experimental scene and the difference effect in the data preprocessing of the present invention, wherein (1) is the textile reflection signal, (2) is the empty environment reflection signal, and (3) is the background difference result.

[0048] Figure 5 The figure is a schematic diagram of the working principle and beam forming principle of the multi-antenna array of the present invention, wherein (1) is the antenna array receiving the environmental reflected echo, and (2) is the antenna array beam forming principle.

[0049] Figure 6 This is a schematic diagram of the beamforming effect applied in the present invention.

[0050] Figure 7 This is a schematic diagram of the results of improving the ranging accuracy of the present invention.

[0051] Figure 8 Schematic diagram of a benchmark experiment for studying the effects of different textile properties on wireless signals according to the present invention, wherein (1) is the effects of different materials on detection, (2) is the effects of different sizes on detection, and (3) is the effects of different thicknesses on detection.

[0052] Fig. 9 Schematic diagram of the comparison between the theoretical model of bridging textile humidity and wireless signals of the present invention and the observed results. DETAILED DESCRIPTION

[0053] The specific implementation of the present invention is further described below in conjunction with the technical solution and the accompanying drawings.

[0054] Reference Figure 1 , showing the layout and hardware information of the non-contact textile humidity detection system based on UWB of the present invention. The detection adopts X4MIMO UWB device with 8 transmit and 8 receive array antennas, with a transmission signal center frequency of 7.29GHz, a signal bandwidth of 1.5GHz, a transmission power of 0.084mw (ensuring safety in use and no harm to equipment and personnel), and a sampling rate of 2042Hz. The device can be connected to a mobile device to realize continuous humidity data collection and real-time display of results.

[0055] During the acquisition process, the device sends a carefully modulated UWB signal. The specific process is as follows:

[0056] First, the UWB signal can be simulated as a set of extremely narrow Gaussian pulse signals, as shown in formula (9):

[0057] Where M is the number of pulses, V i is the pulse amplitude, τ i is the transmission delay of the ith pulse, and σ is the standard deviation that determines the pulse width. Subsequently, the pulse signal is modulated to a carrier of a certain frequency and transmitted by the transmitting antenna. The modulated signal s(t) is shown in formula (10):

[0058] s(t)=x(t)cos(2πfc t), (10)

[0059] Where f c is the center frequency of the modulated carrier signal. Finally, the receiving antenna receives the reflected signal r(t) from different propagation paths, as shown in formula (11):

[0060] r(t)=s(t)×h(t)+n(t), (11)

[0061] where h(t) represents the channel impulse response (CIRs), which can be modeled as Where P represents the total number of reflection paths, α p To represent the propagation attenuation of the pth reflection path, δ(t-τ p ) represents the reflected pulse delay of the pth reflected path, φ p is the phase shift of the received signal corresponding to the pth reflection path, and n(t) is Gaussian noise. CIR describes the channel information between the transmitter and the receiver. It is usually represented as a two-dimensional complex matrix, which has a fast time dimension and a slow time dimension due to different sampling methods. By detecting the arrival of the signal, the fast time dimension sampling provides a preliminary estimate of the propagation distance to the target.

[0062] Reference Figure 2 , the main component of textiles is cellulose, which consists of many hydrophilic groups. The porous molecules of the fabric, and the ample spaces between them, provide the textile with strong moisture absorption and moisture retention capabilities. When the fabric is exposed to the outside, moisture molecules initially attach to the fibers, which is called surface wetting. The water molecules then diffuse into the amorphous regions inside the fibers, driven by permeability. The diffusion process requires the water molecules to break the binding forces to enter the spaces between the polymer chains. The basic principle behind UWB textile humidity sensing is that the dielectric constant of the fabric changes as the humidity level changes. The polarity of its molecules determines the dielectric constant of the material. As the molecules become more polarized, the dielectric constant increases. The dielectric constant of a material is usually expressed as a complex number ∈=∈′+j∈″. The real part ∈′ is the dielectric constant, which reflects the dielectric properties of the material. The imaginary part ∈″ is called the loss factor, which mainly affects the energy dissipation. In most cases, the dielectric constant is usually expressed as the ratio of the dielectric constant relative to that in a vacuum, defined as where ∈ 0 represents the dielectric constant in vacuum. The real part of the relative dielectric constant∈ r ' is closely related to the water content of the material. Textiles are composed of various cellulose molecules with weak polarity. When water molecules penetrate the fabric and reach a uniform and stable state, the water molecules with stronger polarity significantly improve the dielectric properties of the fabric, and the dielectric constant of the fabric increases.

[0063] Step 1: Construct a theoretical model to bridge textile moisture content and wireless signal strength:

[0064] The reason why textiles with weak dielectric properties can affect wireless signals is that the presence of water greatly increases the polarization characteristics of the molecules inside the textile, thus having higher dielectric properties. According to previous research results in the field of materials, the relationship between the dielectric constant of a textile and its water content can be expressed by the following equation (the cotton textile equation is used as an example):

[0065] ∈′ r (W)=-W 3 +7.28W 2 -0.87W+2.13 (12)

[0066] where ∈′ r is the real part of the relative dielectric constant of cotton, and W is the moisture content. For different textile materials, the formula format is the same, but the coefficients are different (i.e. -1, 7.28, -0.7, 2.13), which are obtained by fitting the actual observed data.

[0067] Based on the electromagnetic wave propagation theory and the Friis transmission formula, we can simplify the reflection model of electromagnetic wave signals as follows:

[0068]

[0069] Among them, |A t | and |A r | respectively represent the amplitude of the transmitted and received signals, G t and G r Represent the gains of the transmitting and receiving antennas respectively. λ represents the wavelength of the UWB signal, d represents the distance from the antenna to the target, and R represents the reflection coefficient of the target, which is used to represent the basic signal propagation model.

[0070] When electromagnetic waves propagate to the surface of two different media (such as water and air), reflection will occur. At this interface, part of the electromagnetic wave will be refracted into the new medium, and the other part will be reflected back. Analyzing the reflection and refraction of electromagnetic waves at the boundaries of various materials depends on applying reasonable boundary conditions to the plane equations. In this invention, the problem is regarded as a transmission line model to simplify the analysis. Therefore, we can derive the reflection coefficient:

[0071]

[0072] Where Z 1 and Z 2 They represent the impedance of the two media, air and textile, respectively. Impedance determines the amount of signal penetration and return when the signal is incident from a relatively uniform medium to another uniform medium. The inherent impedance and dielectric constant satisfy the following relationship: Where Z 0is the free space impedance. Substituting the impedance representation method into the transmission line model, the relationship between the reflection coefficient and the dielectric constants of the two media can be obtained:

[0073]

[0074] where ∈ 1 Represents the dielectric constant of air, and in most cases its value defaults to 1.

[0075] Substituting the above derived equations (13)-(15) one by one and combining them with equation (12), we can establish an equation that establishes a relationship between the moisture level of the textile material and the signal characteristics,

[0076]

[0077] Among them, S r Indicates the UWB signal receiving signal strength. Related to water content. i and γ are constants related to the type of textile. Based on the observation data collected from different materials, we can use the nonlinear least squares fitting method to estimate the corresponding parameters in the formula. The statistical goodness of fit is evaluated using r square (i.e., the square of the correlation between the observed value and the predicted response value) and the root mean square error (RSME). Taking cotton textiles as an example, after actual observation data measurement and parameter fitting method, the functional relationship between the reflection signal intensity of cotton textiles and the humidity time of textiles is obtained as follows:

[0078] in,

[0079] Step 2: Robust signal processing for irrelevant noise and dependency elimination, see Figure 3 The specific operation steps are as follows:

[0080] (1) Wireless signal preprocessing and noise elimination: To address the problem of weak reflectivity of textiles, a series of methods are designed to achieve target detection and noise filtering;

[0081] (1.1) Figure 4 As shown in Figure 1, textiles have a low dielectric constant and thus have a weaker ability to reflect signals. Figure 4As shown in (1), in the measured reflection signals, the reflection signal originally located at the location of the textile target is much weaker than the reflection signals of the wall and other strong reflectors in the environment. The textile reflection signal is submerged in the reflection signals of other objects, resulting in the inability to accurately locate the location of the target, and therefore the textile-related signal information cannot be obtained. Therefore, in the data preprocessing part, the background difference method is first used to suppress other irrelevant reflections from the environment to highlight the weak reflection signal of the target textile. Specifically, we collect the reflection signal of the empty environment when there is no textile target before the detection. This operation only needs to be collected once before the actual textile humidity detection. In the subsequent textile humidity detection process, the first step is to perform a background difference algorithm on the reflection signal collected when the textile target exists and the reflection signal of the empty environment collected in advance. Since the only variable in the collected signal before and after the existence of the textile is the textile, except for the change in the reflection signal caused by the newly introduced textile, the reflection signals of other objects and layout settings in the environment are almost unchanged. Therefore, no matter how strong the reflection ability of an object in the original environment is, it can be weakened by differentiating itself. The reflected signal in the textile area is a process of emergence from nothing, so the signal after differentiation is more prominent. Figure 4 (1) and Figure 4 In (3), the reflected signal of the target textile is significantly enhanced before and after background subtraction. Therefore, a more accurate target position can be obtained and preliminary target reflected signal information can be extracted.

[0082] (1.2)Reference Figure 5 We use multiple antennas to further enhance the quality of textile reflected signals. First, after collecting signals for a period of time, the signal transmission quality of different channels is determined by calculating the coefficient of variation of the received signals of different transceiver channels in the antenna array. Selecting the five most stable transceiver channels can avoid the influence of some irrelevant environmental and hardware interference. The selection of the number of stable transceiver channels here is the result of weighing the detection stability and processing time cost.

[0083] According to the spatial position of the antenna corresponding to the selected transceiver channel, beamforming technology can be used to further concentrate the reflected signal from the target direction, thereby further suppressing irrelevant interference from other directions. The specific method is as follows:

[0084] The reflected signal of the target is incident in parallel to the antenna array and received by the receiving antenna. Due to the different positions of the antennas, the signal reaches each antenna with different time delays. Therefore, we select the first antenna in the array as the reference, and construct the following weight expression based on the spatial position of the antenna in the array and the phase difference of the received signal to compensate for the phase difference caused by the different delays of different antennas.

[0085]

[0086] Where i = 1, 2, ..., n, where n is the number of antennas in the array, Δφ i (θ) represents the phase difference of the received signal corresponding to the spatial position of the antenna, λ is the wavelength of the signal, and l is the length between adjacent antennas in the array. For example, when the target azimuth When , the phase offset of the received signal of the second antenna in the array relative to the first antenna in the array is calculated as Then the beamforming weights of different antennas in the array are obtained, denoted as Based on the above weights, beamforming can be performed on the array of selected channels.

[0087]

[0088] Among them, N is the number of stable channels, ω i is the weight vector, X i is the signal vector of the selected channel. The beamforming result reduces the influence of reflected signals in directions of no interest, making the moisture measurement more accurate. Figure 6 As shown, beamforming is used to compensate for the phase difference between antenna arrays, so that the signal after shaping is more concentrated, the beam is narrower, and the directivity is more obvious, which is convenient for accurately locating the target reflection angle and resisting interference from reflected signals from other irrelevant directions of the environment.

[0089] Finally, the wavelet denoising method is used to filter out the environmental noise to avoid affecting the weak reflection signal of the textile. Wavelet denoising is mainly divided into the following three steps: Feature extraction: Specify the wavelet decomposition level K, and perform wavelet decomposition on the signal to obtain frequency components of different levels and their corresponding wavelet coefficients. Low-pass filter: Select a suitable threshold for each level of coefficients obtained by wavelet decomposition to suppress noise in the high-frequency area. Signal reconstruction: Apply the inverse wavelet transform to the frequency components that have been thresholded to reconstruct the signal back to obtain the denoised effective signal. After wavelet denoising, the high-frequency signal fluctuations caused by environmental noise and hardware thermal noise are significantly weakened, reducing the interference of high-frequency noise on the weak reflection signal of the textile.

[0090] (2) Elimination of irrelevant dependencies: Eliminate the influence of the distance between the target and the device and the properties of the textile itself. Due to the energy consumption in the transmission process, distance has the most serious impact on signal quality. In order to achieve a system that is not restricted by placement, the influence of distance should be eliminated first. Secondly, a novel solution is proposed for the influence of properties between different textiles, so that our system is not affected by these factors.

[0091] (2.1) The received signal strength of wireless signals is closely related to the propagation distance. In order to make the proposed system practical, we should eliminate the limitation of distance dependence. The propagation loss of RF signals can be expressed by the Friis free space loss equation:

[0092]

[0093] Among them, S t and S r is the received signal strength of the transmitted signal, and is the gain of the transmitting and receiving antennas in the dB domain, R is the reflection coefficient, and d is the propagation distance. From this equation, we can see that the distance has a great influence on the received signal S r The effect is reflected at -20log 10 (2d) This term. By obtaining the exact distance d between the target and the device, the reflected signal received at any distance can be compensated by 20log 10 (2d) eliminates the distance term in the equation, thus obtaining the characteristic value S of the reflected signal r ′=S r +20log 10 (2d) will no longer be affected by the distance. Combining the derived signal and the theoretical model of textile moisture, through theoretical calculation, when the distance estimation error is less than 1%, it will only cause a 0.5% fluctuation in the moisture content estimation error. For example, assuming that the actual target distance is 4m, the calculated distance compensation term value is 20log 10 (2·4)≈18.06, the calculated moisture content is 10%; while the device estimates the target distance to be 3.96m, the calculated distance compensation term value is 20log 10 (2·3.96)≈17.97, the calculated water content is 9.6%, and this error can be ignored.

[0094] In order to obtain high-precision distance estimation, we first increased the sampling points of the device, so that the device has doubled the sampling points in the same time, so that a finer range-bin division can be obtained. Then, cubic spline interpolation is used to further improve the accuracy of the signal in the bin where the target is located. The effect is that the distance resolution of the original binA and binB is about 5cm. Assuming that the corresponding distance of binA is 100cm and the corresponding distance of binB is 105cm, if the range-bin (103cm) where the target is located falls between A and B, its final distance bin estimate can only be A or B, that is, the final ranging error is 2 to 3cm. If 10 times interpolation is used for the above results, 10 sampling points will be added between binA and binB, and the final distance estimate we may get is 103.5cm, so that a more accurate ranging result can be obtained. Considering the trade-off between ranging accuracy and time cost, we chose to perform 20 times interpolation, which supports us to achieve a ranging error of less than 1% within a range of 4m. The accuracy improvement effect is as follows Figure 7 As shown in the figure, the target position is 0.37m. When no operation is performed, due to the limitation of resolution, the estimated result falls at 0.35m, and the distance error is 5.41%. After 20 times interpolation of the collected signal, the estimated distance result falls at 0.365m, and the calculated estimation error is 1.35%. The significant improvement in distance estimation accuracy will support the elimination of subsequent distance terms.

[0095] (2.2)Reference Figure 8 We conducted some benchmark experiments to analyze the impact of different textile properties on the reflected signal, including different materials (cotton, polyester, silk, nylon, viscose), different sizes (80cm×80cm, 100cm×100cm, 120cm×120cm) and different thicknesses (0.25mm, 0.90mm, 1.9mm). We strictly controlled the variables and conducted corresponding experiments on these three properties. The results are shown in the figure below. Figure 8 It shows that in non-extreme cases (such as when the target is too small to be detected), the effect of size and shape on the reflected signal is minimal, and the range of detection we can provide is sufficient to cover most sizes of textile products on the market. As for material factors, due to the different chemical materials that make up the finished product, different materials exhibit different dielectric properties at the same humidity. But overall, as the humidity level increases, the trend of change in the dielectric capacity exhibited by different materials is the same, so for different materials, it is only necessary to estimate the corresponding parameters of the model to complete the calibration. For the thickness factor, the greater the thickness, the stronger the dielectric property. By abstracting the thickness change as the superposition of textiles with unit thickness without layers, the problem can be simplified to a multi-layer reflection problem. It is manifested in the increase of the reflection coefficient R, so a correlation coefficient η is added to R in the model to obtain the final model. After formula arrangement, the newly introduced thickness correlation coefficient η can be arranged in W * Therefore, the thickness effect can also be calibrated by fitting parameters to solve the thickness effect. Fig. 9 As shown, the numerical comparison between the detection results of the proposed theoretical model bridging textile humidity and wireless signals and the actual observation values ​​has an extremely high degree of fit, which can achieve a detection error of 1.4%.

Claims

1. A non-contact textile humidity detection method based on UWB, characterized by the following: Step 1: Construct a theoretical model to bridge the moisture content of textiles and wireless signal strength: 1.

1. The relationship between the dielectric constant of a textile and its water content is expressed by the following equation: e ′ r (W)=β1W 3 +β2W 2 +β3W+β4(1) where ε ′ r is the real part of the relative dielectric constant of the textile, W is the moisture content; β1, β2, β3, β4 are coefficients. For different textile materials, the formula format is the same, but the coefficients are different; 1.

2. Based on the electromagnetic wave propagation theory and the Friis transmission formula, the reflection model of the electromagnetic wave signal is simplified as follows: in, |A t | and |A r | respectively represent the amplitude of the transmitted and received signals, G t and G r Respectively represent the gains of the transmitting and receiving antennas; λ represents the wavelength of the UWB signal, d represents the distance from the antenna to the target, and R represents the reflection coefficient of the target; 1.

3. Analysis of the reflection and refraction of electromagnetic waves at the boundaries of various materials relies on applying reasonable boundary conditions to the plane equations to derive the reflection coefficient: Where Z1 and Z2 represent the impedance of the two media, air and textile, respectively; the inherent impedance and dielectric constant satisfy the following relationship: Where Z0 is the free space impedance, and Z is Z1 or Z2; then we can get: Where ∈1 represents the dielectric constant of air, which is 1, and ∈2 represents the dielectric constant of textiles; 1.

4. Substitute formulas (2)-(4) layer by layer and combine them with formula (1) to establish the relationship equation between the humidity level of textile materials and signal characteristics: Among them, S r Indicates the UWB signal receiving signal strength. Related to water content; β i and γ are constants related to the type of textiles; based on the observed data collected from different textile materials, the corresponding parameters in the formula were estimated using the nonlinear least squares fitting method, and the statistical goodness of fit was evaluated using the r-squared and root mean square error; Step 2: Robust signal processing for irrelevant noise and dependency elimination: 2.

1. Wireless signal preprocessing and noise elimination: 2.1.

1. Before the detection, the reflection signal of the empty environment without the textile target is collected in advance. This operation only needs to be collected once before the actual textile humidity detection. In the subsequent detection process of textile humidity, the reflection signal collected when the textile target exists and the reflection signal of the empty environment collected in advance are firstly used for background difference algorithm. Since the only variable in the collected signal before and after the presence of the textile is the textile, the reflection signals of other objects and layout settings in the environment remain unchanged except for the change of the reflection signal caused by the newly introduced textile. 2.1.

2. After collecting signals for a period of time, the signal transmission quality of different channels is determined by calculating the coefficient of variation of the received signals of different transceiver channels in the antenna array; selecting the most stable N transceiver channels can avoid the influence of some irrelevant environment and hardware interference; According to the spatial position of the antenna corresponding to the selected transceiver channel, beamforming technology is used to further concentrate the reflected signals from the target direction and further suppress irrelevant interference from other directions; the specific method is as follows: The reflected signal of the target is incident on the antenna array in parallel and received by the receiving antenna. Due to the different positions of the antennas, the signal reaches each antenna with different time delays. The first antenna in the array is selected as the reference, and according to the spatial position of the antenna and the phase difference of the received signal, the following weight expression is constructed to compensate for the phase difference caused by the different time delays of different antennas. in, i=1,2,…,n, where n is the number of antennas in the array, Δφ i (θ) represents the phase difference of the received signal corresponding to the spatial position of the antenna, θ represents the azimuth of the target, λ is the wavelength of the signal, and l is the length between adjacent antennas in the array; Then the beamforming weights of different antennas in the array are obtained, denoted as j represents an imaginary unit; based on the above weights, the array composed of the selected channels can be beamformed. Among them, N is the number of stable channels, ω i is the weight vector, X i is the signal vector of the selected channel; the beamforming result reduces the influence of reflected signals in directions not of interest, making the moisture measurement more accurate; Finally, the wavelet denoising method is used to filter out the environmental noise to avoid affecting the weak reflection signal of the textile. Wavelet denoising is mainly divided into the following three steps: Feature extraction: specify the wavelet decomposition level K, and perform wavelet decomposition on the signal to obtain frequency components of different levels and their corresponding wavelet coefficients; Low-pass filter: select a suitable threshold for each level of coefficients obtained by wavelet decomposition to suppress noise in the high-frequency area; Signal reconstruction: apply wavelet inverse transform to the frequency components that have been thresholded to reconstruct the signal and obtain the effective signal after denoising; 2.

2. Elimination of irrelevant dependencies: 2.2.

1. The received signal strength of wireless signals is closely related to the propagation distance. In order to make the proposed system practical, the limitation of distance dependence should be eliminated. The propagation loss of RF signals is expressed by the Friis free space loss equation: in, S t and S r is the received signal strength of the transmitted signal, and is the gain of the transmitting and receiving antennas in the dB domain; Distance estimation: First, the sampling points of the device are increased, so that the device has twice as many sampling points in the same time, and a finer range-bin division is obtained; then, the signal of the bin where the target is located is interpolated using cubic spline to further improve the accuracy; 2.2.

2. Benchmark experiments were conducted to analyze the impact of different textile characteristics on the reflection signal, including material, size, shape and thickness. Corresponding experiments were conducted on these four characteristics with strict control of variables. The results showed that the impact of size and shape on the reflection signal was minimal under non-extreme conditions. As for the material factor, due to the different chemical materials that make up the finished product, although different materials exhibit different dielectric properties at the same humidity, the overall trend of change in dielectric capacity as the humidity level increases is the same. Therefore, for different materials, it is only necessary to estimate the corresponding parameters to complete the calibration. As for the thickness factor, the greater the thickness, the stronger the dielectric property. If the thickness change is abstracted as the superposition of textiles with unit thickness without layers, the problem can be simplified to a multi-layer reflection problem, which is reflected in the increase of the reflection coefficient R. Therefore, a correlation coefficient η is added to R in the model to obtain the final model.

Citation Information

Patent Citations

  • UWB-based positioning method and system

    CN113038597A

  • Gesture interaction control system based on UWB

    CN113453147A

Cited By

  • Data processing system and method for textile auxiliary agent production

    CN121051367A