Method for extracting characteristic signal of wet feeling when touching dry-state low-temperature fabric

Abstract

The invention provides a method for extracting a characteristic signal of a wet feeling when touching a dry-state low-temperature fabric. The method comprises the following steps: establishing a characteristic signal acquisition system when a human hand is in contact with the fabric; establishing a characteristic signal processing method; and screening key characteristic variables influencing the wet feeling intensity when the hand touches the waterless fabric. The method is suitable for collecting temperature, force and acceleration signals in the process that the human hand makes contact with the dry-state low-temperature fabric, analyzing the relation between the collected physical signals and the subjective wet feeling, and extracting key features influencing the wet feeling. Compared with the prior art, the method has the advantages that the main influence characteristics of the wet feeling generated when the hand touches the fabric can be obtained by reducing the surface temperature of the fabric under the condition of no water participation, and the method is favorable for guiding the design of related hardware parameters in the virtual reality technology.

Description

Technical Field

[0001] The invention relates to a method for extracting and analyzing characteristic signals of wet sensation generated by the combined action of cold stimulation and mechanical stimulation when human hands interact with dry fabrics, and belongs to the technical field of wet sensation. Background Art

[0002] The development of virtual technology has promoted the research on human sensation reproduction technology. In the process of contact between clothing and human skin, wetness is an important component of skin sensation, which plays an important role in skin sensing and human hand manipulation of objects. From the perspective of the formation mechanism of human sensation, there is no specific wet receptor in the skin. The skin is stimulated by cold signals and mechanical signals to form wet sensation, which is a sensation formed by the comprehensive neurophysiology and psychology. Studies have shown that low-voltage mechanical stimulation combined with cold stimulation can make the human body feel wet without the participation of water. The temperature signal that induces wetness mainly comes from the skin cooling caused by the skin contacting fabrics below the skin temperature. At the same time, the mechanical properties such as softness and roughness caused by the thickness and surface texture of the fabric will also affect the wetness of the human body. Collecting the temperature signal, force signal and acceleration signal when the human hand contacts the dry low-temperature fabric, and analyzing the key characteristic signals that affect the size of the wet sensation, is helpful to establish the subjective and objective connection of the wet sensation of flexible materials, and provide design parameters and theoretical support for the wet sensation reproduction of fabric-like flexible materials by virtual reality technology. Summary of the invention

[0003] The purpose of the present invention is to provide a method for extracting and analyzing wet sensation characteristic signals during the contact process between human hands and waterless fabrics.

[0004] In order to achieve the above object, the technical solution of the present invention is to disclose a method for extracting characteristic signals of wet sensation when touching dry low-temperature fabric, characterized in that it comprises the following steps:

[0005] Step 1, build a characteristic signal acquisition experimental platform for human hand contact with fabric, the experimental platform includes a horizontal base, a three-axis force sensor is embedded in the base, an adjustable temperature cold plate is arranged on the three-axis force sensor, the adjustable temperature cold plate is provided with a thermal conductive silicone pad, the sample is placed on the silicone pad, the three-axis force on the sample during the touch process is measured by the three-axis force sensor, and the obtained measurement value is sent to the host computer; the thermocouple temperature sensing line probe 1 is fixed on the touch part of the subject touching the sample, and is used to measure the skin temperature of the touch part, the thermocouple temperature sensing line probe 2 is fixed on the sample to measure the sample temperature in real time, the thermocouple temperature sensing line probe 1 and the thermocouple temperature sensing line probe 2 send the obtained measurement values ​​to the host computer; the acceleration sensor is fixed on the subject's skin, and is located directly above the thermocouple temperature sensing line probe 1 on the skin, and is used to measure the three-axis acceleration of the touch part, and the acceleration sensor sends the obtained measurement value to the host computer, thereby realizing the synchronous transmission and output storage of multi-dimensional signals of force, temperature and acceleration;

[0006] Step 2: extract and calculate the temperature, force and acceleration signals collected in step 1 to obtain signal characteristics, including:

[0007] The decrease in skin temperature during the whole touch process, ΔT, refers to the overall decrease in skin temperature during the touch process, ΔT = T1 – T0, T0 is the temperature at the start time t0 of the experiment, and T1 is the temperature at the end time t1 of the experiment;

[0008] The skin cooling rate during the whole touch process vT refers to the average rate at which the skin temperature decreases during the touch process, vT = (T1–T0) / (t1–t0);

[0009] The skin temperature change time Δt0 refers to the time from the start of touch to the start of the skin temperature drop, Δt0 = td – t0, td is the time point when the skin temperature data of the next sampling point is first observed to drop;

[0010] The amount of temperature drop within 1 second after the skin starts to change temperature, ΔT0, refers to the amount of skin temperature drop within 1 second from the moment when the skin temperature starts to drop, ΔT0 = Td1 – T0, Td1 is the skin temperature corresponding to the time point td1s later;

[0011] Tangential force mean Describes the average level of tangential force, in, is the value of the i-th sampling point in the tangential force signal, and N is the total number of sampling points;

[0012] Mean normal force represents the average level of normal force, in, is the value of the i-th sampling point in the normal force signal;

[0013] Peak tangential force represents the maximum tangential force,

[0014] Peak normal force represents the maximum normal force,

[0015] Tangential force variance Describes the degree of change of the tangential force signal,

[0016] Variance of normal force Describes the degree of change of the normal force signal,

[0017] Friction coefficient at each sampling point Among them, F xi is the horizontal force, F zi is the normal pressure;

[0018] The mean value of the friction coefficient

[0019] Friction coefficient peak F max , f max =MAX(F1,F2,…,F N );

[0020] Friction coefficient variance f v ,

[0021] The total vibration amount of tangential acceleration V x , In which, it is assumed that the start and end time of the time period are t s and t e ;

[0022] The total vibration of normal acceleration V z , In which, it is assumed that the start and end time of the time period are t s and t e ;

[0023] For the tangential acceleration signal a with time series x (t) and normal acceleration signal a z (t) Fourier transform is performed to obtain the tangential acceleration spectrum A x (f) and normal acceleration spectrum A z (f), and then calculate the integral of the square modulus of the spectrum to obtain the tangential spectrum energy and normal spectrum energy

[0024] The peak value of the tangential acceleration spectrum P z , P x =MAX[E x (f)];

[0025] Normal acceleration spectrum peak P z , P z =MAX[E z (f)];

[0026] The main frequency of tangential acceleration fp x , is the frequency corresponding to the maximum spectrum energy, fp x =argP x = argMAX[E x (f)];

[0027] The main frequency of normal acceleration fp z , represents the frequency corresponding to the maximum energy of the normal spectrum, fp z = argMAX[E z (f)];

[0028] Frequency bandwidth BW of tangential acceleration x , in, and are the maximum and minimum values ​​of the main energy frequency range of the tangential acceleration, respectively;

[0029] Normal acceleration frequency bandwidth BW z , in, and are the maximum and minimum values ​​of the main energy frequency range of normal acceleration, respectively;

[0030] Step 3: The subjects touched samples with different surface temperatures, surface features and thicknesses to evaluate the predictive effect of the wetness scoring model and screen out the key characteristic variables that affect the wetness sensation.

[0031] Preferably, a thermoelectric semiconductor sheet is provided under the adjustable temperature cold plate, and the thermoelectric semiconductor sheet is connected to a water circulation device, thereby realizing the cooling or heating function of the adjustable temperature cold plate; both ends of the thermoelectric semiconductor sheet are connected to a digital temperature controller, and the voltage at both ends of the thermoelectric semiconductor sheet is adjusted by the digital temperature controller to realize the temperature adjustment of the adjustable temperature cold plate.

[0032] Preferably, plastic clips are used to fix the sample on the left and right edges of the temperature-adjustable cold plate to prevent it from sliding during the touch process.

[0033] Preferably, the first thermocouple temperature sensing wire probe and the second thermocouple temperature sensing wire probe are of the same model.

[0034] Preferably, a visual baffle is placed between the characteristic signal acquisition experimental platform and the subject, and a black elastic cloth is set on the visual baffle. The subject touches the sample after passing through the visual baffle through the black elastic cloth, so that the sample is invisible to the subject; a temperature-controllable heating pad is placed in front of the visual baffle to adjust the skin temperature of the subject, and a display unit is set to play a touch animation during the experiment to indicate the movement of the subject's fingers during the acquisition experiment.

[0035] Preferably, step 3 comprises the following steps:

[0036] Step 301: Experimental design

[0037] The experiment was conducted in a constant temperature and humidity climate room. The temperature conditions of the samples were set at 10°C and 15°C. Using the experimental platform built in step 1, the subjects actively touched the samples, collected objective signals, and recorded subjective evaluation results.

[0038] Step 302: Fabric performance test

[0039] The surface performance tester in the Kawabata fabric style instrument is used to measure the performance parameters of the fabric including the average friction coefficient MIU and the surface roughness SMD;

[0040] Step 303: Subject

[0041] The subjects had regular work and rest schedules and no skin or sensory diseases. The subjects did not drink alcohol, coffee or tea within 24 hours before the experiment. In order to keep the subjects in a thermoneutral state in a constant temperature and humidity climate room with set conditions, the subjects wore long-sleeved sports sweatshirts and long pants, short socks and thin sports shoes during the experiment.

[0042] Step 304: Evaluation indicators

[0043] (1) Temperature, force signal, and acceleration signal indicators

[0044] The test subjects' skin temperature, triaxial acceleration of the fingertips and triaxial force on the materials during the whole process of the index finger touching the materials were measured;

[0045] (2) Subjective evaluation

[0046] A visual analog scale was used to collect the wetness evaluation of the subjects. The two ends of the scale were 0 and 10 points, 0 for completely dry and 10 for completely wet. The subjects could mark and score anywhere on the scale.

[0047] Step 305: Experimental process

[0048] Before the experiment began, the subjects entered the climate chamber and sat quietly for a while. During this time, the operator fixed the thermocouple temperature probe on the subject's right index finger 10 mm away from the palm, and fixed the acceleration sensor on the right index finger nail. The operator also distributed a visual evaluation scale to guide the subjects to familiarize themselves with the experimental process and evaluation content. After the rest, the following steps were performed:

[0049] 1) The subject adjusts his sitting posture and becomes familiar with the touch speed and pressure according to the animation and the operator's verbal instructions. If the average pressure of three consecutive touches following the animation is 0.2-1N, the subject is considered to be familiar with the touch action and can continue with the subsequent process;

[0050] 2) control samples at both ends of the subject's perception visual scale of 0 and 10;

[0051] 3) The operator prepares the sample to be tested and places it on an adjustable temperature cold plate to cool it down to the temperature condition ±0.2°C. During this period, the subject uses a heating pad to heat up or a cold plate to cool down, and the initial fingertip temperature is adjusted to 30±0.2°C;

[0052] 4) The operator turns on data collection, and the subject, guided by the animation and metronome, touches the material to be tested from left to right with the index finger of his right hand for 3 seconds; the operator turns off data collection, and the subject marks the feeling evaluation on the visual scale with his left hand. The operator checks the average pressure during the touch period, and if it exceeds the range of 0.2 to 1N, retest the touch from step 1);

[0053] 5) After the subject rests for a while, the next touch process begins, starting from step 1;

[0054] Each material level was touched repeatedly multiple times, and each material sample was used only once to eliminate the influence of changes in fabric moisture during the touch process; after completing the touch test under one experimental condition, the subject rested in place for a period of time before testing the next material;

[0055] Step 306: Data analysis

[0056] Temperature signal, force signal, acceleration signal acquisition and corresponding signal transmission modules are selected to realize signal acquisition and feature analysis of the process of dynamic touch generating wet sensation. By statistically analyzing the experimental results, the changing pattern of wet sensation is studied, the relationship between temperature signal, force signal, acceleration signal and wet sensation is analyzed, and the important features of the wet sensation score prediction model are screened. The independent variable is the feature signal, and the dependent variable is the subjective wet sensation score.

[0057] Preferably, in step 306, the subjective evaluation index calculation method used is:

[0058] According to the experimental process, multiple wetness scores were obtained under each experimental condition, and box plots were drawn to screen and remove outliers. The samples after removing outliers were subjected to Kendall consistency test. If the consistency was good, subsequent analysis was performed; otherwise, normalization was performed before the next step of analysis.

[0059] Preferably, in step 306, the prediction model effect of the signal characteristics and the subjective evaluation index is:

[0060] With signal characteristic indicators as independent variables and wetness score as dependent variable, a prediction model of wetness score was established using random forest regression model to analyze the characteristic signals affecting wetness sensation.

[0061] Preferably, the random forest functions templateTree and fitrensemble of MATLAB are used to implement model building, and the subjects are introduced into the model as categorical variables in combination with one-hot encoding to analyze the impact of individual subjective score differences on the wetness score prediction model.

[0062] Preferably, the process of analyzing the impact of individual subjective score differences on the wetness score prediction model is as follows: perform Z-score standardization on the input feature data to make the impact of variables of different magnitudes on the model relatively consistent; perform one-hot encoding on the subject number as a categorical variable to consider the impact of individual differences; calculate the R2 and root mean square error RMSE of the training set and test set to observe the model effect.

[0063] The present invention is suitable for collecting temperature, force, and acceleration signals when a human hand touches a dry low-temperature fabric, analyzing the relationship between the collected physical signals and the subjective wet feeling, and extracting key features that affect the wet feeling. Compared with the prior art, the present invention can analyze the main influencing features of the wet feeling produced when a human hand touches a material by lowering the surface temperature of different materials without the involvement of water, which is helpful to guide the design of relevant hardware parameters in virtual reality technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 The experimental operation platform is shown in the figure:

[0065] 1-base; 2-force sensor; 3-cold plate; 4-semiconductor chip; 5-water circulation device; 6-digital temperature controller; 7-power supply; 8-plastic clamp; 9-sample; 10-force sensor transmitter; 11-development board; 12-thermocouple (skin); 13-thermocouple (sample); 14-thermocouple converter; 15-acceleration sensor; 16-computer; 17-baffle; 18-opening; 19-heating pad; 20-display screen)

[0066] Figure 2 This is a software system architecture diagram. DETAILED DESCRIPTION

[0067] The present invention will be further described below in conjunction with specific embodiments. It should be understood that these embodiments are only used to illustrate the present invention and are not intended to limit the scope of the present invention. In addition, it should be understood that after reading the content taught by the present invention, those skilled in the art can make various changes or modifications to the present invention, and these equivalent forms fall within the scope limited by the appended claims of the application equally.

[0068] The embodiment of the present invention discloses a method for analyzing wet sensation characteristics when a human hand contacts a dry low-temperature fabric, comprising the following steps:

[0069] Step 1: Establish a characteristic signal acquisition system for the contact between human hands and fabrics, including an experimental platform and a software system.

[0070] like Figure 1 As shown in the figure, the experimental platform includes a touch platform, a signal acquisition and transmission module, and an experimental auxiliary device. A base 1 is placed on a horizontal table, and a three-axis force sensor 2 (LZ-SWF40, Hefei Lizhi Technology, China) is embedded inside the base 1 to prevent the sensor from shaking. An adjustable temperature cold plate 3 (aluminum, 10cm*10cm, Yuliang Electronics, China) is pasted and fixed above the three-axis force sensor 2. A thermoelectric semiconductor chip 4 (Peltier effect) is pasted below the adjustable temperature cold plate 3, and the semiconductor chip is connected to a water circulation device 5 to realize the cooling or heating function of the cold plate. The two ends of the semiconductor chip 4 are connected to a digital display temperature controller 6 (RC-312M, Ruide Electronics, China) and a 12V power supply 7. The voltage at both ends of the cooling chip is adjusted by the digital display temperature controller 6 to adjust the temperature of the cold plate. The temperature adjustment range is -40 to 110°C, the resolution is 0.1°C, and the accuracy is ±1°C. A thermally conductive silicone pad (10cm*10cm*0.3mm, 13W / m·K) is pasted on the adjustable temperature cold plate 3 to accelerate the cooling of the material and prevent condensation water from being generated on the surface of the adjustable temperature cold plate 3. A sample 9 (10cm*10cm) is placed on the silicone pad, and a plastic clip 8 (10cm*1cm) is used to fix the sample on the left and right edges of the adjustable temperature cold plate 3 to prevent it from sliding during touch.

[0071] The force sensor 2 under the adjustable temperature cold plate 3 is used to measure the triaxial force on the material during the touch process. It is first connected to the digital signal transmitter 10 (LZ-JX3, Hefei Lizhi, China), and then connected to the development board 11 (STM32F407VET6, Dashi Zhiqian, China). A thermocouple temperature sensing wire probe 12 (TT-T-36-SLE, diameter 0.5 mm, OMEGA, USA) is fixed to the center of the right index finger with medical tape 10 mm from the palm to measure the skin temperature of the touched part. Another thermocouple temperature sensing wire probe 13 of the same model is fixed to the sample with medical tape to measure the material temperature in real time. Combined with the digital display temperature controller 6, the material temperature is adjusted. Both temperature sensing wires 12 and 13 are connected to the thermocouple temperature converter 14 (MCP9600, MICROCHIP, USA), and then connected to the development board 11. An acceleration sensor 15 (JY62, Vit Intelligent, China) is fixed to the skin just above the temperature sensing wire probe 12 using medical tape to measure the three-axis acceleration of the touch site, and then connected to the development board 11. Finally, the development board 11 is connected to the host 16 via USB to achieve multi-dimensional signal synchronous transmission and output storage of force, temperature and acceleration.

[0072] A visual baffle 17 (430mm*290mm) made of thick cardboard is placed between the above-mentioned touch platform and signal acquisition device and the subject. A rectangular opening 18 of 200mm*50mm is opened on the side facing the subject, and the opening is covered with black elastic cloth. The subject passes his arm through the horizontal cut on the elastic cloth. During this process, the subject cannot see the experimental materials to avoid the visual influence on the subjective evaluation of the fabric. A temperature-controllable heating pad 19 (520mm*260mm) is placed on the table in front of the baffle to adjust the skin temperature of the subject. An iPad display 20 (250mm*174mm, Apple, USA) is placed on the left side of the baffle to play touch animations during the experiment to indicate the movement of fingers during the collection experiment.

[0073] Software system architecture Figure 2 As shown in the figure, the main task is to realize the acquisition, transmission, synchronous processing, application and storage of temperature signals, force signals and acceleration signals, which is divided into two parts: front-end data acquisition and back-end data processing. The front-end software completes the configuration, acquisition, data sorting and upload of temperature, pressure and acceleration sensors. The data acquisition frequency is 20Hz. The back-end completes the initialization setting of the front-end acquisition system, reads and processes the collected data, and completes the storage and display of temperature, pressure and acceleration data. It can realize the import and export of collected data and Excel data tables.

[0074] Step 2: Establish a method for processing characteristic signals. Collect and calculate skin temperature, force and acceleration signals.

[0075] The processing method of characteristic signals established in step 2 is to extract and calculate the collected temperature, force and acceleration signals to obtain signal characteristics. The specific method is as follows:

[0076] 1. Signal feature calculation method

[0077] a. Temperature signal

[0078] In the study of wetness, the skin temperature drop and skin cooling rate are often used to analyze the relationship between coldness and wetness. In order to reduce the individual differences in skin temperature caused by different index finger skin thickness, the following four characteristics are calculated:

[0079] The decrease in skin temperature during the entire touch process (ΔT) refers to the overall decrease in skin temperature during the touch process;

[0080] The skin cooling rate during the whole touch process (vT) refers to the average rate at which the skin temperature decreases during the touch process;

[0081] Skin temperature change start time (Δt0), which refers to the time from the start of touch to the start of skin temperature drop;

[0082] The amount of temperature drop within 1 second after the skin starts to change temperature (ΔT0) refers to the amount of skin temperature drop within 1 second from the moment the skin temperature starts to drop.

[0083] Assuming the starting time is t0 and the temperature is T0, and the ending time is t1 and the temperature is T1, then ΔT = T1–T0, in °C; vT = (T1–T0) / (t1–t0), in °C / s. Assuming the time point when the skin temperature data of the next sampling point is first observed to drop is td (the corresponding skin temperature is still T0), and the corresponding skin temperature 1 second after this time point is Td1, then Δt0 = td–t0, in seconds; ΔT0 = Td1–T0, in °C.

[0084] b. Force signal

[0085] The forces during the interaction between skin and fabric are related to the mechanical properties and touch of the fabric. The following nine characteristics are calculated:

[0086] Tangential force mean Describes the average level of tangential force, which is related to the overall softness of the material and is calculated as, in is the value of the i-th sampling point in the tangential force signal, N is the total number of sampling points, and N = 40 in the experiment.

[0087] Mean normal force It represents the average level of normal force and is calculated as in is the value of the i-th sampling point in the normal force signal, N=40.

[0088] Peak tangential force Indicates the maximum tangential force, calculated by the formula

[0089] Peak normal force Indicates the maximum value of normal force, calculated by the formula

[0090] Tangential force variance Describes the degree of change in the tangential force signal, which is related to the surface roughness or unevenness of the material. The calculation formula is:

[0091] Variance of normal force Describes the degree of change of the normal force signal, and the calculation formula is:

[0092] Friction coefficient at each sampling point In the experiment, N=40, where F xi is the horizontal force, F zi is the normal pressure. According to the above formula, the average value of the friction coefficient is calculated The formula is Peak friction coefficient f max , the formula is f max =MAX(f1,f2,…,f N ) ; and the friction coefficient variance f v , the formula is These three parameters are related to the surface roughness of the material.

[0093] c. Acceleration signal

[0094] Acceleration is one of the important signals for evaluating the surface texture of the material touched by the hand. The following eight features are calculated:

[0095] The total vibration amount of tangential acceleration V x , It is assumed that the start and end time of the time period are ts and te, and the unit is m. 2 / s 3 .

[0096] The total vibration of normal acceleration V z , It is assumed that the start and end time of the time period are ts and te, and the unit is m. 2 / s 3 .

[0097] For the tangential acceleration signal a with time series x (t) and normal acceleration signal a z(t) Fourier transform is performed to obtain the tangential acceleration spectrum A x (f) and normal acceleration spectrum A z (f), and then calculate the integral of the square modulus of the spectrum to obtain the tangential spectrum energy and normal spectrum energy

[0098] The peak value of the tangential acceleration spectrum P x , P x =MAX[E x (f)], unit is m / s 2 .

[0099] Normal acceleration spectrum peak P z , P z =MAX[E z (f)], unit is m / s 2 .

[0100] The main frequency of tangential acceleration fp x , is the maximum spectrum energy (spectral peak value P x ) corresponds to the frequency, fp x =argP x = argMAX[E x (f)], unit is Hz.

[0101] The main frequency of normal acceleration fp z , represents the maximum value of the normal spectrum energy (spectral peak P z ) corresponds to the frequency, fp z = argMAX[E z (f)], unit is Hz.

[0102] Frequency bandwidth BW of tangential acceleration x , in and They are the maximum and minimum values ​​of the main energy frequency range of the tangential acceleration, in Hz.

[0103] Normal acceleration frequency bandwidth BW z , in and They are the maximum and minimum values ​​of the main energy frequency range of the normal acceleration, in Hz.

[0104] 2. Feature signal extraction method

[0105] A low-pass filter is used to remove high-frequency vibration signals, and all signals are denoised. Then, the signals are windowed and calibrated to obtain signal data during the time period of contact between the skin and the fabric. The signal features are then calculated using the signal feature calculation method.

[0106] 3. Screening the key characteristic variables that affect the intensity of wet sensation when human hands touch water-free fabrics. Subjects touched materials with different surface temperatures, surface features, and thicknesses to evaluate the predictive effect of the wet sensation scoring model and screen out the key characteristic variables that affect wet sensation.

[0107] In the step three, key characteristic variables that affect the intensity of wet sensation when a person touches the water-free fabric are screened.

[0108] The specific method is as follows:

[0109] 1. Experimental Design

[0110] The experiment was conducted in a constant temperature and humidity climate chamber (T: 20±1°C, RH: 65±5%, Air speed: 0.4m / s). To explore the effect of different material temperatures on wet sensation, the temperature conditions for touching the material were set at 10°C and 15°C. Using the experimental platform described in step 1, the subjects actively touched the material, collected objective signals, and recorded subjective evaluation results.

[0111] 2. Fabric performance test

[0112] The surface performance tester (KES-FB4) in the Kawabata fabric style tester (KES) was used to measure the average friction coefficient MIU, surface roughness SMD and other performance parameters of the fabric.

[0113] 3. Subjects

[0114] The subjects had regular work and rest schedules and no skin or sensory diseases. The subjects did not drink alcohol, coffee, or tea within 24 hours before participating in the experiment. In order to keep the subjects in a thermoneutral state in a constant temperature and humidity climate room with set conditions, the subjects were required to wear long-sleeved sports sweatshirts and long pants, short socks and thin sports shoes during the experiment, and the total thermal resistance of the clothing was about 0.8clo.

[0115] 4. Evaluation indicators

[0116] (1) Temperature, force signal, and acceleration signal indicators

[0117] The skin temperature (℃) and triaxial acceleration (m / s) of the test subject's index finger during the whole process of touching the material were measured. 2 ) and the triaxial force (N) applied to the material. The signal acquisition frequency is 20 Hz, and the single acquisition time is about 3 to 5 seconds, including a touch action and the subject's reaction time, that is, 60 to 100 acquisition points.

[0118] (2) Subjective evaluation

[0119] The visual analog scale was used to collect the wetness evaluation of the subjects. The scale was 10 cm long, with 0 and 10 points at the two ends, 0 for completely dry and 10 for completely wet. The subjects could mark and score anywhere on the scale (0 to 10 points, accurate to 0.1). To ensure that the experimental evaluation values ​​were on a unified basis, reference samples were set at 0 and 10 points at both ends of the scale. During the experiment, the evaluation of other materials should be compared with the reference samples.

[0120] 5. Experimental Procedure

[0121] Before the experiment began, the subjects entered the climate chamber and sat quietly for 10 minutes. During this time, the subjects signed the informed consent form. The operator fixed the temperature probe on the subject's right index finger 10 mm from the palm side, and fixed the acceleration sensor on the right index finger nail. The operator also issued a visual evaluation scale to guide the subjects to familiarize themselves with the experimental process and evaluation content. After the rest, the following steps were performed:

[0122] 1) The subject adjusts his sitting posture and becomes familiar with the touch speed and pressure according to the animation (2.5s, 40mm / s) and the operator's verbal instructions. If the average pressure of three consecutive touches following the animation is 0.2-1N, the subject is considered to be familiar with the touch action and can continue with the subsequent process, which takes about 1 minute;

[0123] 2) control samples at both ends of the subject's perception visual scale of 0 and 10;

[0124] 3) The operator prepares the sample to be tested and places it on a cold plate to cool it down to the temperature condition ±0.2°C. During this period, the subject uses a heating pad to heat up or a cold plate to cool down, and the initial fingertip temperature is adjusted to 30±0.2°C;

[0125] 4) The operator turns on data collection, and the subject, guided by the animation and metronome, touches the material to be tested from left to right with the tip of his right index finger for 3 seconds; the operator turns off data collection, and the subject marks the feeling evaluation on the visual scale with his left hand. The operator checks the average pressure during the touch period, and if it exceeds the range of 0.2 to 1N, retest the touch from step 1).

[0126] 5) The subject rests for 5 minutes and then enters the next touch process, starting the cycle from step 1).

[0127] Each material level was touched 5 times repeatedly, and each material sample was used only once to eliminate the influence of fabric moisture changes during the touch process. After completing the touch test under one experimental condition, the subject rested for 10 minutes before testing the next material to avoid frequent touches that weakened the judgment of sensory differences. The 12 experimental conditions (6 materials × 2 temperature levels) were randomly ordered, and each subject collected 5*12=60 signals. After all tests were completed, the subject rested for 10 minutes and left after no discomfort.

[0128] 6. Data Analysis

[0129] The temperature signal, force signal, acceleration signal acquisition and corresponding signal transmission modules are selected to realize the signal acquisition and feature analysis of the process of dynamic touch generating wetness. Through statistical analysis of the experimental results, the changing law of wetness is studied, the relationship between temperature signal, force signal, acceleration signal and wetness is analyzed, and the important features of the wetness score prediction model are screened. The independent variable is the feature signal, and the dependent variable is the subjective wetness score.

[0130] 1) Calculation method of subjective evaluation index

[0131] According to the experimental process, 12 wetness scores were obtained under each experimental condition, and a box plot was drawn to screen and remove outliers. The samples after removing outliers were subjected to Kendall consistency test. If the consistency was good (P<0.05, Kendall W>0.6), subsequent analysis was performed; otherwise, normalization was performed before the next step of analysis.

[0132] 2) Prediction model effect of signal characteristics and subjective evaluation indicators

[0133] With the signal characteristic index as the independent variable and the wetness score as the dependent variable, a random forest regression model was used to establish a prediction model for the wetness score, and analyze the characteristic signals that affect the wetness. The random forest functions templateTree and fitrensemble of MATLAB were used to implement the model, and the subjects were introduced into the model as categorical variables in combination with one-hot encoding to analyze the impact of individual subjective score differences on the wetness score prediction model. The analysis process is as follows: the input feature data is Z-score standardized to make the impact of variables of different magnitudes on the model relatively consistent; the subject number is used as a categorical variable for one-hot encoding to consider the impact of individual differences; the R of the training set and the test set are calculated. 2 And root mean square error RMSE (Root Mean Square Error) are used to observe the model effect.

[0134] The above method is further explained below in combination with specific data.

[0135] 1. Experimental material selection

[0136] The test samples selected 6 materials with large differences in thickness and surface properties, namely S1~S6, as shown in Table 1. Before the experiment, the materials were cut into several samples, placed in a climate chamber for 48 hours, and then put into sealed bags and placed in a refrigerator (5~7℃) for pre-cooling. Before using the samples, adjust the temperature of the cold plate, take the samples out of the sealed bags and lay them flat on the cold plate, and it will take 5~15s to reach the material temperature conditions required for the experiment (±0.2℃). All samples are not allowed to be used twice to avoid contamination of the materials by secretions such as oil on the skin.

[0137] Table 1 Basic parameters of experimental materials

[0138]

[0139]

[0140] 2. Subject Screening

[0141] The subjects were screened using a health questionnaire, and finally 12 young female subjects (f = 0.5, α = 0.05 and β = 0.9, G*Power 3.1, Germany) were recruited to participate in the experiment. They were 21.1 ± 1.7 years old, 162.1 ± 7.1 cm tall, 56.4 ± 5.9 kg in weight, and 21.5 ± 1.5 kg / m 2 .

[0142] 3. Experimental Design

[0143] The experimental platform and software system described in step 1 are used to simulate the dynamic process of human hands interacting with water-free materials, and characteristic signals are calculated and extracted as variables that affect wet sensation.

[0144] Each material was tested at two temperature levels (10°C, 15°C), for a total of 12 experimental conditions (6 materials × 2 temperature levels). Following the method described in step three, each subject was tested 60 times (12 conditions × 5 repetitions), and the experimental and analysis results were recorded.

[0145] 4. Results and Discussion

[0146] 4.1 Subjective evaluation results

[0147] The outliers were removed from the subjective scores of the subjects, and the consistency was good after Kendall consistency test (W=0.58, P<0.005). The subjective evaluation results are shown in Table 2.

[0148] Table 2 Wetness score results (mean ± standard deviation)

[0149]

[0150] The wetness scores of the 12 materials were ranked from high to low as follows: E3>E9>E2>E5>E1>E8>E4>E7>E10>E11>E6>E12. In general, the materials with lower temperature (E1-E6) and thicker thickness (E2, E3, E8, E9) had stronger wetness. When the surface temperature of the materials was 10℃ (E1-E6), the subjective scores of different materials varied greatly, ranging from 1.84 to 6.26. When the surface temperature of the materials was 15℃ (E7-E12), the subjects gave weaker wetness scores (0.43-1.97), and the wetness differences between different materials were small, indicating that when touching materials with higher surface temperatures, it is difficult to directly predict their wetness differences by comparing the physical parameters of the materials alone, emphasizing the importance of introducing multi-dimensional signal characteristics of the touch process to explore the wetness characteristics.

[0151] 4.2 Signal feature screening results

[0152] According to the signal processing method described in step 2, 144 feature data samples (12 subjects × 12 experimental conditions) were obtained. Each feature data included 4 temperature features, 9 force features, 8 acceleration features, and the subject number as a categorical variable to explore the impact of individual differences on subjective evaluation, a total of 22 variables. The importance scores of the above 22 variables for wetness score (mean of 100 operations) were calculated using the out of bag (OOB) error method of the random forest model with one-hot encoding. Furthermore, the top 10 most important variables with importance scores greater than 0.05 were screened out by sorting the feature importance scores of the variables in the model in descending order.

[0153] The 10 most important variables for the wetness scoring model are as follows: skin cooling time Δt0, tangential acceleration frequency bandwidth BW x , normal acceleration spectrum peak value P z , friction coefficient variance f v , Normal vibration energy V z , mean tangential force Categorical variables, mean normal force Normal acceleration frequency bandwidth BW z , tangential acceleration main frequency fp x Among them, the response speed of the skin temperature drop (Δt0) in the initial stage of touching the low-temperature material has the greatest impact on the wetness sensation.

[0154] 4.3 Identification of important characteristic variables affecting wetness perception

[0155] The signal features of the 10 screened variables were subjected to principal component analysis (PCA) dimensionality reduction. The subject classification variables were not included in the dimensionality reduction process. The dimensionality reduction results showed that the nine key signal features of the wetness score could be appropriately reduced to five dimensions, and the variance explanation ratio of the five dimensions reached 82.59%. Then, a random forest regression model was used to model these reduced-dimensional signal features. The wetness score was used as the dependent variable, and 70% of the data was divided into a training set and 30% of the data as a test set to evaluate the accuracy of the model. The modeling effect was compared using all 22 parameters (21 signal features plus 1 subject classification variable), the 10 most important variables that were screened, and the variables after PCA dimensionality reduction. Table 3 shows the modeling effect.

[0156] Table 3 Effect of random forest in establishing wetness score model

[0157]

[0158] *Parameters for evaluating the effect of the random forest model: RMSE refers to the root mean square error of the model, which is the square root of the average of the squared differences between the predicted value and the actual observation. The closer it is to 0, the smaller the error of the model; R 2 Refers to the degree to which the model explains the data. The closer it is to 1, the better the model's explanatory power.

[0159] Compared with directly using all 21 signal features plus the subject classification variables, the 10 most important variables screened out were used for modeling. The test set R 2 The value of the RMSE of the test set increased by 0.22, and the RMSE of the test set decreased by 0.13, indicating that the number of input parameters using all signal features is large. Some interference variables have been successfully removed through signal feature screening, reducing the overfitting degree of the model. Secondly, compared with the modeling using 10 variables after signal screening, the model training set R of the input parameters after PCA signal feature dimensionality reduction is 2 and the test set R 2 The results show that although feature dimensionality reduction can reduce the complexity of the model, the data information lost in the dimensionality reduction process makes the model effect worse. Therefore, directly using the 10 parameters after screening has the best effect in wetness score modeling. 2 The maximum RMSE of the test set is the smallest, the model is more interpretable, and the error of the predicted value is reduced.

[0160] 5 Conclusion

[0161] According to the method described in this specification, the important characteristic variables that affect the wet feeling when a person touches a low-temperature anhydrous material are evaluated, and the following conclusions are obtained:

[0162] Among the input parameters of the wetness score prediction model, the 10 most important influencing variables are the skin cooling start time Δt0, the tangential acceleration frequency bandwidth BWx , normal acceleration spectrum peak value P z , friction coefficient variance f v , Normal vibration energy V z , mean tangential force Categorical variables, mean normal force Normal acceleration frequency bandwidth BW z , tangential acceleration main frequency fp x . In addition to the skin cooling response speed (Δt0) and classification variables, which are ranked the most important, the others are force signal and acceleration signal characteristics, indicating that the influence of force signal and acceleration signal on wet sensation is more complex than that of temperature signal, covering the surface texture characteristics and normal vibration characteristics of the material, and also indicating that wet sensation is closely related to the roughness and softness of the material surface. This discovery provides theoretical support for the successful simulation of wet sensation under dry conditions by combining cold stimulation and vibration signals in virtual reality technology in recent years.

Claims

1. A method for extracting characteristic signals of wet sensation when touching dry low-temperature fabric, characterized in that: The following steps are involved: Step 1, build a characteristic signal acquisition experimental platform for human hand contact with fabric, the experimental platform includes a horizontal base, a three-axis force sensor is embedded in the base, an adjustable temperature cold plate is arranged on the three-axis force sensor, the adjustable temperature cold plate is provided with a thermal conductive silicone pad, the sample is placed on the silicone pad, the three-axis force on the sample during the touch process is measured by the three-axis force sensor, and the obtained measurement value is sent to the host computer; the thermocouple temperature sensing line probe 1 is fixed on the touch part of the subject touching the sample, and is used to measure the skin temperature of the touch part, the thermocouple temperature sensing line probe 2 is fixed on the sample to measure the sample temperature in real time, the thermocouple temperature sensing line probe 1 and the thermocouple temperature sensing line probe 2 send the obtained measurement values ​​to the host computer; the acceleration sensor is fixed on the subject's skin, and is located directly above the thermocouple temperature sensing line probe 1 on the skin, and is used to measure the three-axis acceleration of the touch part, and the acceleration sensor sends the obtained measurement value to the host computer, thereby realizing the synchronous transmission and output storage of multi-dimensional signals of force, temperature and acceleration; Step 2: extract and calculate the temperature, force and acceleration signals collected in step 1 to obtain signal characteristics, including: The decrease in skin temperature during the whole touch process, ΔT, refers to the overall decrease in skin temperature during the touch process, ΔT = T1 – T0, T0 is the temperature at the start time t0 of the experiment, and T1 is the temperature at the end time t1 of the experiment; The skin cooling rate during the whole touch process vT refers to the average rate at which the skin temperature decreases during the touch process, vT = (T1–T0) / (t1–t0); The skin temperature change time Δt0 refers to the time from the start of touch to the start of the skin temperature drop, Δt0 = td – t0, td is the time point when the skin temperature data of the next sampling point is first observed to drop; The amount of temperature drop within 1 second after the skin starts to change temperature, ΔT0, refers to the amount of skin temperature drop within 1 second from the moment when the skin temperature starts to drop, ΔT0 = Td1 – T0, Td1 is the skin temperature corresponding to the time point td1s later; Tangential force mean Describes the average level of tangential force, in, is the value of the i-th sampling point in the tangential force signal, and N is the total number of sampling points; Normal force mean represents the average level of normal force, in, is the value of the i-th sampling point in the normal force signal; Peak tangential force represents the maximum tangential force, Peak normal force represents the maximum normal force, Tangential force variance Describes the degree of change of the tangential force signal, Variance of normal force Describes the degree of change of the normal force signal, Friction coefficient at each sampling point Among them, F xi is the horizontal force, F zi is the normal pressure; The mean value of the friction coefficient Friction coefficient peak value f max , f max =MAX(f1,f2,…,f N ); Friction coefficient variance f v , The total vibration amount of tangential acceleration V x , In which, it is assumed that the start and end time of the time period are t s and t e ; The total vibration of normal acceleration V z , In which, it is assumed that the start and end time of the time period are t s and t e ; For the tangential acceleration signal a with time series x (t) and normal acceleration signal a z (t) Fourier transform is performed to obtain the tangential acceleration spectrum A x (f) and normal acceleration spectrum A z (f), and then calculate the integral of the square modulus of the spectrum to obtain the tangential spectrum energy and normal spectrum energy The peak value of the tangential acceleration spectrum P x , P x =MAX[E x (f)]; Normal acceleration spectrum peak P z , P z =MAX[E z (f)]; The main frequency of tangential acceleration fp x , is the frequency corresponding to the maximum spectrum energy, fp x =argP x =argMAX[E x (f)]; The main frequency of normal acceleration fp z , represents the frequency corresponding to the maximum energy of the normal spectrum, fp z =argMAX[E z (F)]; Frequency bandwidth BW of tangential acceleration x , in, and are the maximum and minimum values ​​of the main energy frequency range of the tangential acceleration, respectively; Normal acceleration frequency bandwidth BW z , in, and are the maximum and minimum values ​​of the main energy frequency range of normal acceleration, respectively; Step 3: The subjects touched samples with different surface temperatures, surface features and thicknesses to evaluate the predictive effect of the wetness scoring model and screen out the key characteristic variables that affect the wetness sensation.

2. The method for extracting characteristic signals of wet sensation when touching dry low-temperature fabric according to claim 1, characterized in that: A thermoelectric semiconductor sheet is provided under the adjustable temperature cold plate, and the thermoelectric semiconductor sheet is connected to a water circulation device, thereby realizing the cooling or heating function of the adjustable temperature cold plate; two ends of the thermoelectric semiconductor sheet are connected to a digital temperature controller, and the voltage at both ends of the thermoelectric semiconductor sheet is adjusted by the digital temperature controller to realize the temperature adjustment of the adjustable temperature cold plate.

3. The method for extracting characteristic signals of wet sensation when touching dry low-temperature fabric according to claim 1, characterized in that: Plastic clips are used to fix the sample on the left and right edges of the temperature-adjustable cold plate to prevent it from sliding during the touch process.

4. The method for extracting characteristic signals of wet sensation when touching dry low-temperature fabric according to claim 1, characterized in that: The thermocouple temperature sensing wire probe 1 is of the same model as the thermocouple temperature sensing wire probe 2.

5. The method for extracting characteristic signals of wet sensation when touching dry low-temperature fabric according to claim 1, characterized in that: A visual baffle is placed between the characteristic signal acquisition experimental platform and the subject, and a black elastic cloth is set on the visual baffle. The subject passes through the visual baffle through the black elastic cloth and touches the sample, so that the sample is invisible to the subject; a temperature-controllable heating pad is placed in front of the visual baffle to adjust the skin temperature of the subject, and a display unit is set to play a touch animation during the experiment to indicate the movement of the subject's fingers during the acquisition experiment.

6. The method for extracting characteristic signals of wet sensation when touching dry low-temperature fabric according to claim 1, characterized in that: The step 3 comprises the following steps: Step 301: Experimental design The experiment was conducted in a constant temperature and humidity climate room. The temperature conditions of the samples were set at 10°C and 15°C. Using the experimental platform built in step 1, the subjects actively touched the samples, collected objective signals, and recorded subjective evaluation results. Step 302: Fabric performance test The surface performance tester in the Kawabata fabric style instrument is used to measure the performance parameters of the fabric including the average friction coefficient MIU and the surface roughness SMD; Step 303: Subject The subjects had regular work and rest schedules and no skin or sensory diseases. The subjects did not drink alcohol, coffee or tea within 24 hours before the experiment. In order to keep the subjects in a thermoneutral state in a constant temperature and humidity climate room with set conditions, the subjects wore long-sleeved sports sweatshirts and long pants, short socks and thin sports shoes during the experiment. Step 304: Evaluation indicators (1) Temperature, force signal, and acceleration signal indicators The test subjects' skin temperature, triaxial acceleration of the fingertips and triaxial force on the materials during the whole process of the index finger touching the materials were measured; (2) Subjective evaluation The visual analog scale is used to collect the wetness evaluation of the subjects, with 0 points and 10 points at the two ends of the scale, 0 points representing completely dry, and 10 points representing completely wet. The subjects can mark and score anywhere on the scale; Step 305, Experimental process Before the experiment began, the subjects entered the climate chamber and sat quietly for a while. During this time, the operator fixed the thermocouple temperature probe on the subject's right index finger 10 mm away from the palm, and fixed the acceleration sensor on the right index finger nail. The operator also distributed a visual evaluation scale to guide the subjects to familiarize themselves with the experimental process and evaluation content. After the rest, the following steps were performed: 1) The subject adjusts his sitting posture and becomes familiar with the touch speed and pressure according to the animation and the operator's verbal instructions. If the average pressure of three consecutive touches following the animation is 0.2-1N, the subject is considered to be familiar with the touch action and can continue with the subsequent process; 2) control samples at both ends of the subject's perception visual scale of 0 and 10; 3) The operator prepares the sample to be tested and places it on an adjustable temperature cold plate to cool it down to the temperature condition ±0.2°C. During this period, the subject uses a heating pad to heat up or a cold plate to cool down, and the initial fingertip temperature is adjusted to 30±0.2°C; 4) The operator turns on data collection, and the subject, guided by the animation and metronome, touches the material to be tested from left to right with the index finger of his right hand for 3 seconds; the operator turns off data collection, and the subject marks the feeling evaluation on the visual scale with his left hand. The operator checks the average pressure during the touch period, and if it exceeds the range of 0.2 to 1N, retest the touch from step 1); 5) After the subject has rested for a while, the next touch process begins, starting from step 1; Each material level was touched repeatedly multiple times, and each material sample was used only once to eliminate the influence of changes in fabric moisture during the touch process; after completing the touch test under one experimental condition, the subject rested in place for a period of time before testing the next material; Step 306: Data analysis Temperature signal, force signal, acceleration signal acquisition and corresponding signal transmission modules are selected to realize signal acquisition and feature analysis of the process of dynamic touch generating wet sensation. By statistically analyzing the experimental results, the changing pattern of wet sensation is studied, the relationship between temperature signal, force signal, acceleration signal and wet sensation is analyzed, and the important features of the wet sensation score prediction model are screened. The independent variable is the feature signal, and the dependent variable is the subjective wet sensation score.

7. A method for extracting characteristic signals of wet sensation when touching dry low-temperature fabrics as claimed in claim 6, characterized in that: In step 306, the subjective evaluation index calculation method used is: According to the experimental process, multiple wetness scores were obtained under each experimental condition, and box plots were drawn to screen and remove outliers. The samples after removing outliers were subjected to Kendall consistency test. If the consistency was good, subsequent analysis was performed; otherwise, normalization was performed before the next step of analysis.

8. The method for extracting characteristic signals of wet sensation when touching dry low-temperature fabric according to claim 6, characterized in that: In step 306, the prediction model effect of the signal characteristics and the subjective evaluation index is: With signal characteristic indicators as independent variables and wetness score as dependent variable, a prediction model of wetness score was established using random forest regression model to analyze the characteristic signals affecting wetness sensation.

9. The method for extracting characteristic signals of wet sensation when touching dry low-temperature fabric according to claim 8, characterized in that: The model was established using MATLAB's random forest functions templateTree and fitrensemble, and the subjects were introduced into the model as categorical variables in combination with one-hot encoding to analyze the impact of individual subjective score differences on the wetness score prediction model.

10. The method for extracting characteristic signals of wet sensation when touching dry low-temperature fabric according to claim 9, characterized in that: The process of analyzing the impact of individual subjective score differences on the wetness score prediction model is as follows: perform Z-score standardization on the input feature data to make the impact of variables of different magnitudes on the model relatively consistent; use the subject number as a categorical variable for one-hot encoding to consider the impact of individual differences; calculate the R of the training set and the test set 2 And root mean square error RMSE to observe the model effect.

Images