Down feather moisture content test analysis method based on artificial intelligence

Abstract

The invention discloses a down feather moisture content test analysis method based on artificial intelligence, and relates to the technical field of down feather moisture detection. Comprising the following steps: collecting a plurality of groups of down feather samples with known water content as test samples, and carrying out cleaning pretreatment on the collected test samples; then, a laser interferometer is used for collecting optical characteristics of the test sample, meanwhile, a camera is used for recording appearance characteristics of the test sample, a down feather water content correction model is established, the obtained optical characteristics and appearance characteristics of the test sample serve as input of the model, the corresponding water content serves as a label, and the model is trained; the method comprises the following steps: collecting a down feather raw material to be detected, performing cleaning pretreatment, inputting collected characteristic data into a trained down feather water content correction model to obtain optically detected down feather water content, generating a correction coefficient on the basis of the water content in combination with the temperature and humidity of a drying environment, and correcting the down feather water content detected by a drying method to obtain a down feather water content correction result. Obtaining the accurate water content of the down.

Description

Technical Field

[0001] The invention relates to the technical field of down moisture detection, and in particular to a down moisture content testing and analysis method based on artificial intelligence. Background Art

[0002] Down feathers are one of the main by-products of poultry farming. They are soft, warm and fluffy, and are mainly used in clothing and bedding. Down in a broad sense refers to all feathers of poultry except wing feathers. Down in a narrow sense refers to feathers less than 5 cm in length that can be used as fillings for textiles. Down includes flake feathers and down feathers, of which down feathers have a higher utilization value. Due to the production characteristics of poultry feathers, chickens have very little down available, while waterfowl down has a larger amount of available down.

[0003] With the prosperous development of the down feather market, the down feather and product industry has become an industry with great development potential in my country's national economy. The evaluation standards for down feather quality are becoming more and more important. Among them, the detection of down feather moisture content is one of the important indicators for evaluating down feather quality. At present, the main method for down feather moisture detection is the oven method. The measurement process of this method is completed manually, which is time-consuming and inefficient. In addition, while drying the sample moisture, some effective ingredients contained in the sample, such as oil, will also be dissipated during the baking process, thus affecting the accuracy of the measurement results.

[0004] In order to solve the existence of such problems and improve the accuracy of moisture measurement results, it is necessary to calibrate the down moisture measurement results and use more reasonable detection methods to improve the accuracy of down moisture measurement results.

[0005] In the prior art, publication number CN102878785A discloses a temperature-controlled heat-conducting oil down dryer, in which the interlayer of the cylinder of the down dryer is filled with heat-conducting oil, and a plurality of temperature sensors are evenly distributed in the heat-conducting oil; a plurality of groups of electric heating tubes are evenly arranged in the interlayer of the cylinder; sieve plates are respectively arranged on the end covers at the left and right ends of the cylinder; air boxes are respectively arranged on the outside of the sieve plates; an air duct is connected between the left and right air boxes; an air inlet and an air exhaust port are connected to the middle of the air duct by a four-way reversing valve; an air preheating device and a wind hood are sequentially connected to the air inlet, and a radiator connected to the inner cavity of the cylinder is arranged on the outside of the air preheating device, and the air is heated by an external heat source to dry the down and remove moisture. During use, the air box works to pass air into the cylinder of the dryer through the air duct to heat the sample evenly, but the air box easily causes impurities in the down to fall off, and the moisture test result is biased large. Therefore, the moisture content of the down is measured only by the mass of the down after drying, which reduces the validity and accuracy of the moisture content data of the down.

[0006] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not constitute the prior art that is already known to one of ordinary skill in the art. Summary of the invention

[0007] The purpose of the present invention is to provide a down moisture content testing and analysis method based on artificial intelligence to solve the problems raised in the above background technology.

[0008] To achieve the above object, the present invention provides the following technical solutions:

[0009] A method for testing and analyzing moisture content of down based on artificial intelligence, the specific steps include:

[0010] Several groups of down materials with known moisture content are collected as test samples, and the test samples laid out on the test platform area are irradiated by a laser interferometer, and interference signals are recorded by a receiver under the test platform area. The collected interference signals are analyzed to obtain the total spacing between interference fringes and the total number of interference fringes;

[0011] Using a high-resolution camera to collect an image of the test sample laid out in the test platform area, and obtaining the base area of ​​the test sample in the test platform area and the height of the test sample according to the image of the test sample;

[0012] A down moisture content correction model is established, and the total spacing between interference fringes, the total number of interference fringes, the base area of ​​the test sample in the test platform area, and the height of the test sample are used as inputs of the model. The moisture content corresponding to each group of test samples is used as a label to train the down moisture content correction model to obtain a trained down moisture content correction model.

[0013] The total spacing between interference fringes, the total number of interference fringes, the base area and height in the detection platform area corresponding to the down material to be detected are collected, and the acquired data are input into the trained down moisture content correction model to obtain the optical detection down moisture content of the down material to be detected;

[0014] Collecting the initial mass of the down material to be tested, drying the down material to be tested, obtaining the mass of the down material to be tested after drying, and calculating the moisture content of the down material to be tested according to the mass of the down material to be tested after drying and the initial mass;

[0015] The moisture content of the down material to be tested obtained by the drying method is corrected according to the optically detected down moisture content output by the down moisture content correction model to obtain the accurate moisture content of the down material.

[0016] Furthermore, it is characterized in that: samples of determined quality are obtained from each group of down raw material samples that have been cleaned and pretreated as test samples, and the test samples are spread flat on the test platform area, and a receiver is arranged under the test platform area; the test platform area is made of a light-transmitting material, wherein the test platform area is rectangular and the test platform area is placed horizontally.

[0017] Furthermore, a laser interferometer is used to illuminate the test sample laid out in the test platform area to ensure that the laser beam passes through the test sample, wherein the laser incident angle is 45°. The total spacing of the interference fringes and the total number of the interference fringes are generated based on the analysis of the collected interference signal. The formula is:

[0018]

[0019] In the formula, Δx is the interference fringe spacing, λ is the laser wavelength, and ΔL is the optical path difference. The formula for calculating the optical path difference ΔL is:

[0020] ΔL=2d*cosθ

[0021] Where d is the physical path length of the laser beam, and θ is the incident angle of the laser beam;

[0022] If a total of m interference fringes are collected, a total of m-1 interference fringes spacings are generated, so the total spacing of the interference fringes is calculated based on the formula:

[0023]

[0024] In the formula, Δx z is the total spacing of the interference fringes, Δx i is the i-th interference fringe spacing, where i = 1, 2, …, m-1.

[0025] Further, the base area of ​​the detection sample in the detection platform area and the height of the detection sample are obtained according to the detection sample image, wherein the method for obtaining the base area of ​​the detection sample in the detection platform area is to use the Canny edge detection algorithm to detect and extract the edge of the detection sample in the detection platform area, and calculate the area of ​​the extracted edge by using the pixel counting method, and the specific steps include: converting the identified contour area into a binary image, calculating the number of contour pixels in the contour area, and calculating the area of ​​the contour according to the ratio of known pixels to the actual area;

[0026] The height of the detection sample is measured by a laser sensor to measure the distance from the sample surface to the sensor, and the height of the detection sample is calculated based on the distance from the sample surface to the sensor and the distance from the sensor to the surface of the detection platform.

[0027] Furthermore, a down moisture content correction model is established, wherein the down moisture content correction model is constructed based on a convolutional neural network;

[0028] The initial neural network is a convolutional neural network, which consists of an input layer, a convolutional layer, a pooling layer, a fully connected layer and an output layer, wherein the activation function in the convolutional layer is a ReLU function, and the specific expression of the ReLU function is:

[0029] ReLU(s l(p,q) )=max(0,s l(p,q) )

[0030] Among them, l represents the lth convolutional layer, s l(p,q) It represents the qth eigenvalue of the pth feature map in the lth convolutional layer;

[0031] For the fully connected layer, the number of neurons in the fully connected layer is set to 64, the initial neural network learning rate is set to 0.001, and the number of training rounds is 50.

[0032] Furthermore, the mass of the down raw material to be tested is collected, and the down raw material to be tested is dried, wherein the method for drying the down raw material to be tested is: using an eight-basket oven to dry the down raw material to be tested, evenly distributing the down raw material to be tested in the eight baskets to ensure air circulation, then opening the oven and starting the blower to ensure even air circulation, setting a constant drying temperature to dry the down raw material, and again, weighing the sample mass every 30 minutes until the difference between two adjacent masses is no more than 0.1% of the total sample mass, and the drying is completed; finally, recording the final mass.

[0033] Furthermore, the formula for calculating the moisture content of the down material to be tested based on the mass of the down material to be tested after drying and the initial mass is:

[0034]

[0035] Where W 2 is the moisture content of the down material to be tested, m 0 is the initial mass of the down material to be tested, m t It is the quality of the down raw material to be tested after drying.

[0036] Furthermore, the logic for correcting the calculated moisture content of the down material to be detected according to the optically detected down moisture content output by the down moisture content correction model is as follows: the optically detected down moisture content W of the down material to be detected output by the down moisture content correction model is: 1 and the calculated moisture content W of the down material to be tested 2Calculate the correction coefficient, and adjust the calculated moisture content W of the down material to be tested according to the correction coefficient. 2 Correction is made to obtain accurate moisture content calculation of down raw materials;

[0037] The formula for calculating the precise moisture content of down raw materials is:

[0038] W z =(1+K)*W 2

[0039] Where W z is the precise moisture content of the down material, K is the correction factor, and the correction factor is based on the formula:

[0040]

[0041] Where K is the correction coefficient, ω 1 ,ω 2 and ω 3 W 1 and W 2 Average value and W 2 The ratio of the dry environment temperature T and the dry environment humidity H, where ω 1 ,ω 2 and ω 3 are greater than 0 and ω 1 >ω 2 >ω 3 .

[0042] Compared with the prior art, the present invention has the following beneficial effects:

[0043] Several groups of down samples with known moisture content are collected as test samples, and the collected test samples are cleaned and pre-treated to reduce test interference. The optical characteristics of the test samples, including the total spacing between interference fringes and the total number of interference fringes, are collected using a laser interferometer. At the same time, the appearance characteristics of the test samples, including the base area of ​​the test samples in the test platform area and the height of the test samples, are recorded with a camera. A down moisture content correction model is established, and the optical characteristics and appearance characteristics of the test samples are used as inputs to the model. The corresponding test sample down moisture content is used as a label to train the model. The down raw materials to be tested are collected. After cleaning and pre-treatment, the total spacing between the interference fringes, the total number of interference fringes, the base area and the height in the test platform area of ​​the down raw materials to be tested are collected, and the collected feature data are input into the complete In the trained down moisture content correction model, the optically detected down moisture content of the down raw material to be tested is obtained. Based on the optically detected down moisture content combined with the drying temperature and humidity, a correction coefficient is generated, and the down moisture content obtained by the drying method is corrected to obtain the accurate moisture content of the down raw material; this method uses laser interferometry technology to analyze the moisture content by measuring the changes in the optical properties of the down samples under different humidity conditions. This method does not require direct contact with the sample and avoids interference from impurities. At the same time, images of the down samples are taken, and the moisture content is judged by analyzing the changes in the appearance of the down under different humidity conditions. At the same time, the model is trained according to the optical characteristics and appearance characteristics to improve the robustness of the correction. The down moisture content obtained by the drying method is corrected by the correction coefficient, which effectively improves the accuracy of the down moisture content detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 It is a schematic diagram of the overall method flow of the present invention. DETAILED DESCRIPTION

[0045] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments.

[0046] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0047] Example:

[0048] See also Figure 1 , the present invention provides a technical solution:

[0049] A method for testing and analyzing moisture content of down based on artificial intelligence, the specific steps include:

[0050] Step 1: Collect several groups of down materials with known moisture content as test samples, use a laser interferometer to irradiate the test samples laid out in the test platform area, record the interference signal through the receiver under the test platform area, and analyze the collected interference signal to obtain the total spacing between the interference fringes and the total number of interference fringes;

[0051] Before testing, visible impurities on the surface of the down raw material sample can be removed. The method for removing visible impurities on the surface of the down raw material sample can be through visual inspection and manual selection to remove impurities. The steps include spreading the down sample flat on a clean surface, carefully inspecting it with hands or tweezers, and removing visible impurities; or using an air spray gun to remove visible impurities on the surface of the sample to avoid affecting subsequent moisture measurements.

[0052] A sample of a determined mass is obtained from each group of down raw material samples that have been cleaned and pretreated as a test sample, and the test sample is spread flat on a test platform area, the test platform area is made of a light-transmitting material, and a receiver is arranged below the test platform area; the test platform area is made of a light-transmitting material, wherein the test platform area is rectangular and is placed horizontally, wherein a laser interferometer selects a helium-neon laser with a wavelength of 632.8 nm to ensure high precision;

[0053] Use a laser interferometer to illuminate the test sample laid out in the test platform area, ensuring that the laser beam passes through the test sample. The laser incident angle is 45°. The total spacing of the interference fringes and the total number of interference fringes are generated based on the collected interference signal analysis formula:

[0054]

[0055] In the formula, Δx is the interference fringe spacing, λ is the laser wavelength, and ΔL is the optical path difference. The formula for calculating the optical path difference ΔL is:

[0056] ΔL=2d*cosθ

[0057] Where d is the physical path length of the laser beam, and θ is the incident angle of the laser beam;

[0058] If a total of m interference fringes are collected, a total of m-1 interference fringes spacings are generated, so the total spacing of the interference fringes is calculated based on the formula:

[0059]

[0060] In the formula, Δx z is the total spacing of the interference fringes, Δx i is the i-th interference fringe spacing, where i = 1, 2, …, m-1.

[0061] Interference fringes are light and dark stripe patterns produced by the coherent interference of light waves. When light waves pass through different media, such as water-containing down, the change in refractive index will lead to the formation of interference fringes. The characteristics of interference fringes, such as the total spacing and number, can reflect the physical properties of the down sample. When the moisture content is high, the refractive index of the down may increase, resulting in a decrease in the spacing of the interference fringes. Conversely, when the moisture content is low, the refractive index decreases and the spacing of the interference fringes may increase. As the moisture content changes, the spacing of the interference fringes changes, which in turn affects the number of fringes. When the spacing of the interference fringes decreases, the number of fringes increases; when the spacing of the interference fringes increases, the number of fringes decreases.

[0062] Step 2: Use a high-resolution camera to capture an image of the test sample laid out in the test platform area, and obtain the base area of ​​the test sample in the test platform area and the height of the test sample based on the image of the test sample;

[0063] The base area of ​​the detection sample in the detection platform area and the height of the detection sample are obtained according to the detection sample image, wherein the method for obtaining the base area of ​​the detection sample in the detection platform area is to use the Canny edge detection algorithm to detect and extract the edge of the detection sample in the detection platform area, and calculate the area of ​​the extracted edge by using the pixel counting method, and the specific steps include: calculating the gradient of the image in the horizontal direction and the vertical direction, calculating the number of contour pixels in the contour area, and calculating the area of ​​the contour according to the ratio of the known pixels to the actual area;

[0064] The height of the detection sample is measured by a laser sensor to measure the distance from the sample surface to the sensor, and the height of the detection sample is calculated based on the distance from the sample surface to the sensor and the distance from the sensor to the surface of the detection platform.

[0065] The Canny edge detection algorithm is used to extract edge features. The specific steps include calculating the image gradient and calculating the gradient of the image in the horizontal and vertical directions. The formula for calculating the gradient of the image in the horizontal and vertical directions is:

[0066]

[0067] In the formula, G x is the horizontal gradient of the image, G y is the vertical gradient of the image, ■ represents the convolution operation, and I is the input image; the formula for calculating the gradient magnitude and direction is:

[0068]

[0069] In the formula, G is the gradient amplitude and θ is the direction;

[0070] Non-maximum suppression is performed along the gradient direction, local maxima are retained, non-edge points are suppressed, high thresholds and low thresholds are used to distinguish strong edges from weak edges, and edge detection is completed by connecting weak edges and strong edges.

[0071] Calculate the number of contour pixels within the contour area, and calculate the area of ​​the contour based on the known ratio of pixels to the actual area. The contour is white, the background is black, and the contour area is the number of white pixels multiplied by the actual area of ​​each pixel.

[0072] Step 3: Establish a down moisture content correction model, take the total spacing between interference fringes, the total number of interference fringes, the base area of ​​the test sample in the test platform area and the height of the test sample as the input of the model, take the moisture content corresponding to each group of test samples as the label, train the down moisture content correction model, and obtain the trained down moisture content correction model;

[0073] Establishing a down moisture content correction model, wherein the down moisture content correction model is constructed based on a convolutional neural network;

[0074] The initial neural network is a convolutional neural network, which consists of an input layer, a convolutional layer, a pooling layer, a fully connected layer and an output layer, wherein the activation function in the convolutional layer is a ReLU function, and the specific expression of the ReLU function is:

[0075] ReLU(s l(p,q) )=max(0,s l(p,q) )

[0076] Among them, l represents the corresponding convolutional layer, s l(p,q) It represents the qth eigenvalue of the pth feature map in the lth corresponding convolutional layer;

[0077] For the fully connected layer, the number of neurons in the fully connected layer is set to 64, the initial neural network learning rate is set to 0.001, and the number of training rounds is 50.

[0078] Step 4: Collect the total spacing between the interference fringes, the total number of interference fringes, the base area and height in the detection platform area corresponding to the down material to be detected, and input the acquired data into the trained down moisture content correction model to obtain the optical detection down moisture content of the down material to be detected;

[0079] Establishing a down moisture content correction model, wherein the down moisture content correction model is constructed based on a convolutional neural network;

[0080] The initial neural network is a convolutional neural network, which consists of an input layer, a convolutional layer, a pooling layer, a fully connected layer and an output layer, wherein the activation function in the convolutional layer is a ReLU function, and the specific expression of the ReLU function is:

[0081] ReLU(s l(p,q) )=max(0,s l(p,q) )

[0082] Among them, l represents the lth convolutional layer, s l(p,q) It represents the qth eigenvalue of the pth feature map in the lth convolutional layer;

[0083] For the fully connected layer, the number of neurons in the fully connected layer is set to 64, the initial neural network learning rate is set to 0.001, and the number of training rounds is 50.

[0084] Step 5: collecting the initial mass of the down material to be tested, drying the down material to be tested, obtaining the mass of the down material to be tested after drying, and calculating the moisture content of the down material to be tested according to the mass of the down material to be tested after drying and the initial mass;

[0085] The mass of the down raw materials to be tested is collected, and the down raw materials to be tested are dried. The method for drying the down raw materials to be tested is as follows: using an eight-basket oven to dry the down raw materials to be tested, evenly distributing the down raw materials to be tested in the eight baskets to ensure air circulation, then opening the oven and starting the blower to ensure even air circulation, setting a constant drying temperature to dry the down raw materials, and then weighing the sample mass every 30 minutes until the difference between two adjacent masses is no more than 0.1% of the total sample mass, and the drying is completed; finally, recording the final mass.

[0086] The eight-basket oven is an experimental device used to determine the moisture content of substances such as down, food, etc. Its working principle is to evaporate the moisture in the sample by controlling the temperature and air flow, thereby measuring the moisture content. The eight-basket oven usually has eight independent baskets, which can hold multiple samples at the same time. There is a heating element inside the oven to heat the air to reach the set temperature. The temperature is set to 100±5℃ and a constant temperature is used for drying. The blower is used to enhance air flow, make the temperature in the oven uniform, and accelerate the evaporation of moisture.

[0087] The formula for calculating the moisture content of the down material to be tested based on the mass of the down material to be tested after drying and the initial mass is:

[0088]

[0089] Where W 2 is the moisture content of the down material to be tested, m 0 is the initial mass of the down material to be tested, m t It is the quality of the down raw material to be tested after drying.

[0090] Step 6: Correct the moisture content of the down material to be tested obtained by the drying method according to the optically detected down moisture content output by the down moisture content correction model to obtain the accurate moisture content of the down material;

[0091] The logic for correcting the calculated moisture content of the down material to be detected according to the optically detected down moisture content of the down material to be detected output by the down moisture content correction model is as follows: 1 and the calculated moisture content W of the down material to be tested 2 Calculate the correction coefficient, and adjust the calculated moisture content W of the down material to be tested according to the correction coefficient. 2 Correction is made to obtain accurate moisture content calculation of down raw materials;

[0092] The formula for calculating the precise moisture content of down raw materials is:

[0093] W z =(1+K)*W 2

[0094] Where W z is the precise moisture content of the down material, K is the correction factor, and the correction factor is based on the formula:

[0095]

[0096] Where K is the correction coefficient, ω 1 ,ω 2 and ω 3 W 1 and W 2 Average value and W 2 The ratio of the dry environment temperature T and the dry environment humidity H, where ω 1 ,ω 2 and ω 3 are all greater than 0, since the correction coefficient is based on the optical detection of down moisture content W 1 and the moisture content W of the down material to be tested 2 The calculated correction coefficient is mainly based on the optical detection of down moisture content and the calculated moisture content of the down raw material to be detected. The weight coefficient ω 1 The maximum value is the drying environment temperature. In addition, the drying environment temperature also has a great influence on the drying. Except for certain circumstances, the drying environment humidity has a smaller influence on the drying result than the temperature. Therefore, ω is set 1 >ω 2 >ω 3 .

[0097] Drying environment temperature T, temperature is an important factor affecting the evaporation rate of water. In the drying method test, the increase in temperature usually accelerates the evaporation of water. Different test temperatures may lead to different moisture measurement results. Therefore, introducing temperature as an independent variable in the model can more accurately correct the measurement error caused by temperature changes; the evaporation rate of water increases at high temperatures, resulting in a lower measured moisture content; on the contrary, the evaporation rate of water slows down at low temperatures, which may result in a higher measured moisture content; therefore, the drying environment temperature T is proportional to the correction factor;

[0098] W 1 and W 2 Average value and W 2 The ratio of the optical down moisture content detected by the non-contact method and the down moisture content detected by the drying method is used to calculate the average value and compare the average value with the down moisture content detected by the drying method. The correction parameters obtained can make the down moisture content detected by the drying method have a smaller error and be closer to the accurate value.

[0099] Dry environment humidity H, environmental humidity also has a significant impact on the measurement results of moisture. In a high humidity environment, down may absorb moisture in the air, resulting in a higher measured moisture content; while in a low humidity environment, the drying speed of the down is accelerated, which may result in a lower measured moisture content. Therefore, when the measured moisture content is higher in a high humidity environment, the correction factor should be reduced to avoid the influence of environmental humidity on the moisture content. The dry environment humidity H is inversely proportional to the correction factor.

[0100] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0101] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product. Those skilled in the art may appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein may be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed by hardware or software methods depends on the specific application and design constraints of the technical solution.

[0102] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, and may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0103] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. A method for testing and analyzing moisture content of down based on artificial intelligence, It is characterized in that The specific steps include: Several groups of down materials with known moisture content are collected as test samples, and the test samples laid out on the test platform area are irradiated by a laser interferometer, and interference signals are recorded by a receiver under the test platform area. The collected interference signals are analyzed to obtain the total spacing between interference fringes and the total number of interference fringes; Using a high-resolution camera to collect an image of the test sample laid out in the test platform area, and obtaining the base area of ​​the test sample in the test platform area and the height of the test sample according to the image of the test sample; A down moisture content correction model is established, and the total spacing between interference fringes, the total number of interference fringes, the base area of ​​the test sample in the test platform area, and the height of the test sample are used as inputs of the model. The moisture content corresponding to each group of test samples is used as a label to train the down moisture content correction model to obtain a trained down moisture content correction model. The total spacing between interference fringes, the total number of interference fringes, the base area and height in the detection platform area corresponding to the down material to be detected are collected, and the acquired data are input into the trained down moisture content correction model to obtain the optical detection down moisture content of the down material to be detected; Collecting the initial mass of the down material to be tested, drying the down material to be tested, obtaining the mass of the down material to be tested after drying, and calculating the moisture content of the down material to be tested according to the mass of the down material to be tested after drying and the initial mass; The moisture content of the down material to be tested obtained by the drying method is corrected according to the optically detected down moisture content output by the down moisture content correction model to obtain the accurate moisture content of the down material.

2. According to claim 1, a method for testing and analyzing moisture content of down based on artificial intelligence, Features: Samples of determined mass are obtained from each group of down raw material samples that have been cleaned and pretreated as test samples, and the test samples are spread flat on a test platform area, under which a receiver is arranged; the test platform area is made of a light-transmitting material, wherein the test platform area is rectangular and is placed horizontally.

3. The method for testing and analyzing moisture content of down based on artificial intelligence according to claim 2, Features: Use a laser interferometer to illuminate the test sample laid out in the test platform area, ensuring that the laser beam passes through the test sample. The laser incident angle is 45°. The total spacing of the interference fringes and the total number of interference fringes are generated based on the collected interference signal analysis formula: In the formula, Δx is the interference fringe spacing, λ is the laser wavelength, and ΔL is the optical path difference. The formula for calculating the optical path difference ΔL is: △L=2d*cosθ Where d is the physical path length of the laser beam, and θ is the incident angle of the laser beam; If a total of m interference fringes are collected, a total of m-1 interference fringes spacings are generated, so the total spacing of the interference fringes is calculated based on the formula: In the formula, Δx z is the total spacing of the interference fringes, Δx i is the i-th interference fringe spacing, where i = 1, 2, …, m-1.

4. The method for testing and analyzing moisture content of down based on artificial intelligence according to claim 3, Features: The base area of ​​the detection sample in the detection platform area and the height of the detection sample are obtained according to the detection sample image, wherein the method for obtaining the base area of ​​the detection sample in the detection platform area is to detect and extract the edge of the detection sample in the detection platform area using the Canny edge detection algorithm, and calculate the area of ​​the extracted edge using the pixel counting method, and the specific steps include: converting the identified contour area into a binary image, calculating the number of contour pixels in the contour area, and calculating the area of ​​the contour according to the ratio of known pixels to the actual area; The height of the detection sample is measured by a laser sensor to measure the distance from the sample surface to the sensor, and the height of the detection sample is calculated based on the distance from the sample surface to the sensor and the distance from the sensor to the surface of the detection platform.

5. The method for testing and analyzing moisture content of down based on artificial intelligence according to claim 4, Features: Establishing a down moisture content correction model, wherein the down moisture content correction model is constructed based on a convolutional neural network; The initial neural network is a convolutional neural network, which consists of an input layer, a convolutional layer, a pooling layer, a fully connected layer and an output layer, wherein the activation function in the convolutional layer is a ReLU function, and the specific expression of the ReLU function is: ReLU(s l(p，q) )=max(0,s l(p，q) ) Among them, l represents the lth convolutional layer, s l(p,q) It represents the qth eigenvalue of the pth feature map in the lth convolutional layer; For the fully connected layer, the number of neurons in the fully connected layer is set to 64, the initial neural network learning rate is set to 0.001, and the number of training rounds is 50.

6. The method for testing and analyzing moisture content of down based on artificial intelligence according to claim 1, Features: The mass of the down raw materials to be tested is collected, and the down raw materials to be tested are dried. The method for drying the down raw materials to be tested is as follows: using an eight-basket oven to dry the down raw materials to be tested, evenly distributing the down raw materials to be tested in the eight baskets to ensure air circulation, then opening the oven and starting the blower to ensure even air circulation, setting a constant drying temperature to dry the down raw materials, and then weighing the sample mass every 30 minutes until the difference between two adjacent masses is no more than 0.1% of the total sample mass, and the drying is completed; finally, recording the final mass.

7. The method for testing and analyzing moisture content of down based on artificial intelligence according to claim 6, Features: The formula for calculating the moisture content of the down material to be tested based on the mass of the down material to be tested after drying and the initial mass is: Where W 2 is the moisture content of the down material to be tested, m 0 is the initial mass of the down material to be tested, m t It is the quality of the down raw material to be tested after drying.

8. The method for testing and analyzing moisture content of down based on artificial intelligence according to claim 1, Features: The logic for correcting the calculated moisture content of the down material to be tested according to the optically detected down moisture content output by the down moisture content correction model is as follows: the optically detected down moisture content W of the down material to be tested output by the down moisture content correction model is 1 and the calculated moisture content W of the down material to be tested 2 Calculate the correction coefficient, and adjust the calculated moisture content W of the down material to be tested according to the correction coefficient. 2 Make corrections to obtain the accurate moisture content of down raw materials; The formula for calculating the precise moisture content of down raw materials is: W Z =(1+K)*W 2 Where W z is the precise moisture content of the down material, K is the correction factor, and the correction factor is based on the formula: Where K is the correction coefficient, ω 1 ,ω 2 and ω 3 W 1 and W 2 Average value and W 2 The ratio of the dry environment temperature T and the dry environment humidity H, where ω 1 ,ω 2 and ω 3 are greater than 0 and ω 1 >ω 2 >ω 3 .

Images