Method and system for detecting water content of extract based on two-waveband hyperspectrum

Abstract

The invention discloses an extract water content detection method and system based on a two-waveband hyperspectrum, and belongs to the field of extract production. In order to realize efficient, lossless, real-time and accurate detection of the water content of the extract, the method for detecting the water content of the extract based on the two-waveband hyperspectrum comprises the following steps: acquiring a hyperspectral image of the extract in drying treatment; according to the spectral images of at least two different wavelengths, at least obtaining reflectivity corresponding to the two different wavelengths; obtaining the moisture index of the extract at least according to the reflectivity corresponding to the two different wavelengths; and calculating the dehydration rate of the extract according to the water index of the extract. The method can be applied to an extract production system, and has the advantages of high efficiency and real-time detection of the water content of the extract.

Description

Technical Field

[0001] The present invention relates to the field of extract production, and in particular to a method and system for detecting the water content of an extract based on a two-band hyperspectral method. Background Art

[0002] In the production process of traditional Chinese medicine, the moisture content of the extract (also known as extract, extract, extract ointment) is a key quality control indicator that affects the uniformity and stability of product quality. The traditional detection method is represented by the drying method (weight loss method). Its principle is to evaporate the water by heating, and measure the difference in mass before and after drying to calculate the moisture content. The specific operation process includes:

[0003] (1) Quickly sample about 1-5 g and seal it in a constant weight weighing dish;

[0004] (2) Accurately weigh the initial weight W1;

[0005] (3) Dry in a drying oven at a fixed or variable temperature for 2 to 4 hours (use vacuum drying at 60 to 80°C when containing volatile components);

[0006] (4) After cooling to room temperature, weigh W2 again and calculate the moisture content using the formula (W1-W2) / W1×100%. This method requires the use of an analytical balance (with an accuracy of 0.001g), drying equipment, a desiccator, and other instruments. However, this solution has the following significant drawbacks:

[0007] 1. Destructive testing and vacuum environment limitations: Offline sampling is required from the production line, resulting in production interruption and inability to continuously monitor. Real-time sampling is difficult during the vacuum drying process and requires shutdown.

[0008] 2. Low detection efficiency: A single test takes 2 to 4 hours, which cannot meet the timeliness requirements of dynamic adjustment of process parameters.

[0009] 3. Significant human errors: Operational steps such as sampling uniformity and cooling time control are prone to introduce deviations, affecting the accuracy of test results.

[0010] In addition, existing technologies offer a variety of alternative methods, such as (a) near-infrared spectroscopy (NIRS), which uses fiber optic probes to achieve non-contact detection based on the absorption characteristics of water to specific wavelengths of infrared light, allowing for quick analysis. (b) microwave detection technology, which leverages microwave penetration to rapidly evaporate water from the extract. Combined with a vacuum environment, this can significantly shorten drying time. (c) artificial neural network (ANN) prediction models, for example, use drying conditions (temperature, vacuum level, loading volume, etc.) as input parameters to construct ANN models to predict changes in the extract's moisture content.

[0011] However, these existing technologies still have many limitations:

[0012] 1. Technical bottlenecks in vacuum environments. Traditional real-time detection technologies (such as NIRS and microwave detection) experience reduced signal transmission stability in vacuum conditions and lack integrated sensors suitable for sealed cavities. For example, NIRS detection technology requires a bypass external circulation system, making it difficult to directly apply to vacuum drying equipment.

[0013] 2. Lack of multidimensional quality parameter detection. Existing methods can only detect moisture content alone and cannot simultaneously quantify changes in the extract's internal structure (such as bubble dynamics and porosity). The formation and collapse of bubbles directly affect drying efficiency and product uniformity. The lack of multidimensional quality parameter detection is not conducive to intelligent process control.

[0014] 3. The conflict between nondestructive testing and real-time control. Although technologies like NIRS and microwaves offer nondestructive properties, they are limited by data processing speed and model response time, making it difficult to meet the demand for real-time adjustment of process parameters. For example, static microwave vacuum dryers, while capable of rapid drying, often lack real-time feedback mechanisms.

[0015] Prior art 1: CN 117030628A, a hyperspectral intelligent analysis method for the quality of traditional Chinese medicine extracts, Zhejiang University.

[0016] Prior art 2: CN 116840110A, Guanxinning quality detection method and application based on hyperspectral imaging technology, Zhengda Qingchunbao Pharmaceutical Co., Ltd. and Zhejiang University of Technology.

[0017] Prior Art 1 performs hyperspectral detection, calibration, masking, and average spectrum calculation on an extract sample. The preprocessing average spectrum or average spectrum feature extraction is then performed to obtain a processed spectrum. A quantitative calibration model for extract quality indicators is constructed based on the processed spectrum using a machine learning algorithm, thereby analyzing the extract sample. However, Prior Art 1 is intended to detect the relative density of the extract.

[0018] Prior Art 2 uses a hyperspectral method to detect the particle size distribution, water content, and active ingredient content of Guanxinning Granules, using a quantitative calibration model based on the ResNet neural network model. While this approach uses hyperspectral analysis to detect water content, its data processing is based on ANNs.

[0019] Based on this, how to achieve a non-destructive solution that can meet the timeliness requirements of process control, can simultaneously extract multi-dimensional quality parameters, and / or can avoid human operation errors is a difficult problem that needs to be solved urgently in this field. Summary of the Invention

[0020] In order to alleviate or partially alleviate the above technical problems, the solutions of the present invention are as follows:

[0021] The present invention provides a method for detecting the water content of an extract based on a two-band hyperspectral method, comprising the following steps:

[0022] Step S1: collecting a hyperspectral image of the extract during the drying process;

[0023] Step S2: obtaining reflectances corresponding to at least two different wavelengths based on spectral images of at least two different wavelengths;

[0024] Step S3: obtaining the extract moisture index based on at least the reflectance corresponding to the two different wavelengths;

[0025] Step S4: Calculate the dehydration rate of the extract according to the moisture index of the extract.

[0026] Furthermore, the extract moisture content detection method based on the two-band hyperspectral method further includes:

[0027] Step S5: Calculating the bubble characteristic value according to the extract moisture index.

[0028] Furthermore, the two different wavelengths include a first wavelength and a second wavelength; and the first wavelength is 1660 nm and the second wavelength is 1420 nm.

[0029] Furthermore, the two different wavelengths include a first wavelength and a second wavelength; the first wavelength is taken from a first wavelength band, and the second wavelength is taken from a second wavelength band; and the first wavelength band is 1655nm to 1665nm, and the second wavelength band is 1415nm to 1425nm.

[0030] Furthermore, the dehydration rate of the extract was calculated based on the extract moisture index and the linear equation.

[0031] Furthermore, the bubble characteristic value is obtained according to the number of bubbles and / or bubble diameter in the extract.

[0032] Furthermore, the calculated dehydration rate of the extract is fed back to the drying system to regulate the drying process.

[0033] Furthermore, the calculated bubble characteristic value of the extract is fed back to the drying system and combined with the dehydration rate to control the drying process.

[0034] On the other hand, the present invention provides a water content detection system for extracts based on a two-band hyperspectral method, comprising:

[0035] A drying system configured to dry the extract;

[0036] The detection system is equipped with a hyperspectral imager and is used to collect the spectrum of the extract to obtain a hyperspectral image; and

[0037] The detection system is also used to obtain the dehydration rate of the extract according to the extract moisture content detection method based on the two-band hyperspectral spectrum as described in any of the above items, and send it to the drying system to achieve control of the drying process.

[0038] Furthermore, the detection system is also used to obtain the bubble characteristic value of the extract and send it to the drying system, and in combination with the dehydration rate, jointly realize the regulation of the drying process.

[0039] The technical solution of the present invention has one or more of the following beneficial technical effects:

[0040] (1) Hyperspectral imaging is used to break through the limitations of traditional destructive sampling and is suitable for integrated sensors in vacuum drying equipment to achieve real-time, non-destructive in-situ detection of the moisture content of the extract.

[0041] (2) The detection purpose can be achieved by performing simple processing on the data in the sensitive band with high accuracy, which greatly reduces the detection time and makes it possible for the drying system to dynamically adjust the drying parameters.

[0042] (3) The present invention can also avoid human operation errors caused by uneven sampling, weighing errors, etc.

[0043] (4) While detecting moisture, it can also quantify the changes in the internal structure of the extract (such as bubble dynamics), which can provide a feedback mechanism for the drying process and establish a closed-loop control system based on real-time detection data.

[0044] In addition, other beneficial effects of the present invention will be mentioned in the specific embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 This is a comparison chart of the extract’s initial RGB image and hyperspectral image at different times;

[0046] Figure 2 It is a reflectance curve diagram of different spectral bands at different drying times;

[0047] Figure 3 It is the first-order differential curve of the core spectral band;

[0048] Figure 4 It is a graph showing the change of extract moisture index with drying time by screening out sensitive bands;

[0049] Figure 5 It is the relationship between the dehydration rate of the extract and the extract index;

[0050] Figure 6 This is a specific flow chart of the extract water content detection method based on two-band hyperspectral;

[0051] Figure 7 It is an extract moisture content detection system based on two-band hyperspectral. DETAILED DESCRIPTION

[0052] To make the objectives, technical solutions, and advantages of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the embodiments described are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0053] To facilitate a clear description of the technical solutions of the embodiments of the present invention, in the embodiments of the present invention, terms such as "first" and "second" are used to distinguish between identical or similar items having substantially the same functions and effects. Those skilled in the art will appreciate that terms such as "first" and "second" do not limit the quantity or order of execution.

[0054] Explanation of terms:

[0055] Waveband: Light is essentially an electromagnetic wave, and its propagation is wave-like. Electromagnetic waves cover a wide range, from gamma rays (extremely short wavelengths) to radio waves (extremely long wavelengths). A waveband is an artificial division of wavelengths. In this context, a waveband can refer to either a wavelength range (e.g., 1415nm to 1425nm) or a specific wavelength (e.g., 1420nm).

[0056] Reflectivity is the ratio of the radiant flux reflected by an object's surface to the incident radiant flux. It reflects an object's ability to reflect light of different wavelengths. It is dimensionless and ranges from 0 to 1. Reflectivity is an inherent property of an object, dependent on its chemical composition, surface structure, roughness, and other factors. It is independent of lighting conditions (although actual measurements are affected by lighting and generally require correction). The pixel values ​​of a hyperspectral image are essentially the quantified results of an object's reflectivity in a specific wavelength band. After correction, they can directly reflect the characteristics of the material.

[0057] The method for detecting the water content of extract based on two-band hyperspectral analysis of the present invention comprises the following steps:

[0058] Step S1: Acquire a hyperspectral image of the extract during the drying process.

[0059] Specifically, a hyperspectral imager can be used to collect the spectrum of the extract in the drying box to obtain a hyperspectral image.

[0060] Alternatively, the spectrum of the extract can be collected directly in the drying workshop using a hyperspectral imager to obtain a hyperspectral image.

[0061] For example, the temperature in the drying oven can be set to 85° C. The extract can be, for example, a traditional Chinese medicine extract.

[0062] Hyperspectral imaging technology captures and analyzes point-by-point spectral information within a spatial region, detecting the unique spectral signatures of a single object at different spatial locations. A hyperspectral imager can be compared to an array of hundreds or thousands of closely spaced spectrometers, all observing the same area simultaneously. Each unit operates independently and acquires spectral data at its corresponding spatial location.

[0063] Hyperspectral images simultaneously record the spatial information (pixel location distribution) and spectral information (radiant brightness values ​​of each pixel in dozens to hundreds of continuous bands) of the target area. They can be regarded as a "data cube" composed of multiple single-band images superimposed on each other. Each band corresponds to a specific wavelength (such as 400nm, 500nm, 600nm, etc.), so each pixel contains a complete spectral curve.

[0064] Hyperspectral images can reflect the detailed spectral characteristics of an object. However, the spectral response of an object can vary due to factors such as light intensity, shadow distribution, climatic conditions, and the shooting environment. This phenomenon can cause objects with the same physical properties to exhibit different spectral characteristics, or objects with different physical properties to exhibit similar spectral characteristics.

[0065] Figure 1 This is a comparison of the RGB images and hyperspectral images of the extract at different initial times. The figure shows the RGB images and hyperspectral images corresponding to different drying or drying times, with the upper part showing the RGB image and the lower part showing the hyperspectral image.

[0066] It can be seen that with the increase of drying time, RGB images cannot significantly reflect the changes of the extract, while hyperspectral images can better reflect the changes of the extract.

[0067] The inventors discovered that hyperspectral images can reflect that the bubbles inside the extract first increase and then decrease. Based on this discovery, in addition to water content detection, the present invention can also realize the extraction of multidimensional quality parameters. This information can provide a basis for optimizing the drying process.

[0068] Figure 2 The following is a graph showing the reflectivity of different spectral bands at different drying times. The measured data shows that for almost all bands, the reflectivity of the extract decreases with drying time.

[0069] Based on this discovery, the present invention utilizes reflectivity, in particular, utilizes reflectivity of at least two wavelength bands, and more particularly utilizes reflectivity of two specific wavelengths, to non-destructively and in situ detect the water content of the extract.

[0070] Figure 3 This is the first-order differential curve of the core spectral band. By taking the first-order differential of the reflectivity, we screened out the sensitive bands and found that the reflectivity is most sensitive at wavelengths of 1660nm and 1420nm.

[0071] Alternatively, the sensitive bands may be screened by wavelet transform, Fourier transform or the like.

[0072] Step S2: According to the spectral images of at least two different wavelengths, at least the reflectivities corresponding to the two different wavelengths are obtained.

[0073] The pixel value of a hyperspectral image is essentially the quantitative result of the reflectivity of an object in a specific band. The corresponding reflectivity is obtained from the hyperspectral image, especially for a specific wavelength or band.

[0074] Preferably, in the present invention, the reflectivity corresponding to the first wavelength of 1660 nm and the second wavelength of 1420 nm can be obtained.

[0075] Alternatively, a first wavelength may be selected in the first wavelength band, and a second wavelength may be selected in the second wavelength band, and the reflectivity corresponding to the first wavelength and the reflectivity corresponding to the second wavelength may be obtained respectively.

[0076] The first wavelength band is 1655nm to 1665nm, and the second wavelength band is 1415nm to 1425nm.

[0077] Step S3: obtaining the moisture index of the extract based on at least the reflectivity corresponding to the two different wavelengths.

[0078] In the present invention, the NDWI can be constructed by any reasonable method, as long as the NDWI can reflect the trend of decreasing moisture content of the extract with increasing drying time.

[0079] The extract moisture index (NDWI) can be exemplarily constructed as follows:

[0080]

[0081] Among them, the first wavelength λ1 = 1660nm, the first wavelength λ2 = 1420nm, is the reflectivity measured at the first wavelength λ1, is the reflectivity measured at the first wavelength λ2.

[0082] Figure 4The following graph shows the variation of the extract moisture index calculated using sensitive bands as a function of drying time. As can be seen from the graph, in this embodiment, the extract moisture index shows a downward trend with increasing drying time. In other words, the extract moisture index obtained from hyperspectral imagery decreases with increasing drying time. This demonstrates that the present invention can determine the extract moisture index, which reflects moisture content, with minimal calculations. Furthermore, the curve trend in the graph demonstrates the effectiveness of the present invention in determining or indirectly determining the moisture content of an extract.

[0083] Step S4: Calculate the dehydration rate of the extract according to the moisture index of the extract.

[0084] Figure 5 It is a relationship diagram between the dehydration rate of the extracted ointment and the ointment index. Through the correlation analysis between the extract moisture index (NDWI) and the extract dehydration amount (i.e., the actual moisture reduction or reduction of the extract after heating for a certain period of time), it can be seen that as the drying time increases, the extract moisture index continues to decrease, while the extract dehydration amount or dehydration rate continues to increase. The dehydration rate and the extract moisture index show an overall negative correlation, which is consistent with the actual situation. Among them, the dehydration rate can be an indicator used by the present invention to measure the water content.

[0085] In one embodiment, the relationship between the dehydration rate y and the extract moisture index x can be described by a linear equation:

[0086] y=-209.04x+41.727;

[0087] This linear relationship allows the use of hyperspectral images to determine the extract moisture index, which can then be used to calculate the dehydration rate. Different extract components may require different calculation formulas, which can be calculated and fitted using conventional experimental sampling data, and will not be further elaborated in this invention.

[0088] In other words, the dehydration rate of the extract can be calculated based on the extract moisture index and the linear equation. Figure 6 This reflects the specific process of the above extract moisture content detection method based on two-band hyperspectral.

[0089] Furthermore, since hyperspectral images can detect bubble information, such as size and number, by constructing bubble feature values, for example, based on the number of bubbles and / or bubble diameter, an indicator that can reflect the bubble information is obtained.

[0090] During the experimental collection phase, based on the obtained extract moisture index and the sampled bubble characteristic values, a mapping relationship between the extract moisture index and the bubble characteristic values ​​can be obtained. Due to different methods of constructing the bubble characteristic values, the corresponding mapping relationship may also be different. Therefore, in a preferred embodiment of the present invention, the following steps may also be included:

[0091] Optionally, step S5: calculating the bubble characteristic value according to the extract moisture index.

[0092] By acquiring the dehydration rate and bubble characteristic values ​​in real time, these data can be transmitted to the drying system for adjusting the extract drying process.

[0093] In addition, reference Figure 7 The present invention also discloses an exemplary extract moisture content detection system based on two-band hyperspectral imaging, which includes a detection system and a drying system. The detection system includes a hyperspectral imager for collecting the spectrum of the extract to obtain a hyperspectral image.

[0094] In addition to the drying box shown in the exemplary two-band hyperspectral extract moisture content detection system, the extract can also be placed directly in a drying workshop without a drying box. For example, the extract is placed in the drying workshop at a distance of 0.5 meters from the hyperspectral imager.

[0095] In the detection system, the aforementioned two-band hyperspectral-based extract moisture content detection method is used to obtain the dehydration rate or bubble characteristic value. This acquisition process is in situ, non-destructive, and more importantly, can be obtained in real time.

[0096] Based on the real-time acquired dehydration rate and bubble characteristic values, these data are transmitted to the drying system for adjusting the extract drying process.

[0097] In summary, compared with the existing technology, the solution proposed in the present invention does not require complex calculations and the collection of a large amount of training sample data. It can achieve the detection purpose by performing simple processing on the data in the sensitive band, and has high accuracy. It is conducive to real-time dynamic adjustment of drying parameters and the establishment of a closed-loop control system based on real-time detection data.

[0098] To better illustrate the present invention, numerous specific details are provided in the detailed description above. Those skilled in the art will appreciate that the present invention can be practiced without certain specific details. In some instances, methods, means, components, and circuits well known to those skilled in the art have not been described in detail in order to highlight the main purpose of the present invention.

[0099] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A method for detecting the water content of extract based on a two-band hyperspectral method, characterized in that: The steps include: Step S1: collecting a hyperspectral image of the extract during the drying process; Step S2: obtaining reflectances corresponding to at least two different wavelengths based on spectral images of at least two different wavelengths; Step S3: obtaining the extract moisture index based on at least the reflectance corresponding to the two different wavelengths; Step S4: Calculating the dehydration rate of the extract according to the moisture index of the extract.

2. The method for detecting water content of extract based on two-band hyperspectral analysis according to claim 1, characterized in that: The extract moisture content detection method based on the two-band hyperspectral method further includes: Step S5: Calculate the bubble characteristic value according to the extract moisture index.

3. The method for detecting water content of extract based on two-band hyperspectral analysis according to claim 1 or 2, characterized in that: The two different wavelengths include a first wavelength and a second wavelength; and, The first wavelength is 1660 nm and the second wavelength is 1420 nm.

4. The method for detecting water content of extract based on two-band hyperspectral analysis according to claim 1 or 2, characterized in that: The two different wavelengths include a first wavelength and a second wavelength; The first wavelength is taken from a first wavelength band, the second wavelength is taken from a second wavelength band; and, The first band is 1655nm~1665nm, and the second band is 1415nm~1425nm.

5. The method for detecting water content of extract based on two-band hyperspectral analysis according to claim 1 or 2, characterized in that: The dehydration rate of the extract was calculated based on the extract moisture index and the linear equation.

6. The method for detecting water content of extract based on two-band hyperspectral analysis according to claim 2, characterized in that: The bubble characteristic value is obtained according to the number of bubbles and / or the bubble diameter in the extract.

7. The method for detecting water content of extract based on two-band hyperspectral analysis according to claim 1 or 2, characterized in that: The calculated dehydration rate of the extract is fed back to the drying system to control the drying process.

8. The method for detecting water content of extract based on two-band hyperspectral analysis according to claim 7, characterized in that: The calculated bubble characteristic values ​​of the extract are also fed back to the drying system and combined with the dehydration rate to control the drying process.

9. A water content detection system for extracts based on a two-band hyperspectral method, characterized in that: include: A drying system configured to dry the extract; The detection system is equipped with a hyperspectral imager and is used to collect the spectrum of the extract to obtain a hyperspectral image; and The detection system is also used to obtain the dehydration rate of the extract according to the extract moisture content detection method based on two-band hyperspectral according to any one of claims 1 to 8, and send it to the drying system to realize the regulation of the drying process.

10. The method for detecting water content of extract based on two-band hyperspectral analysis according to claim 9, characterized in that: The detection system is also used to obtain the bubble characteristic value of the extract and send it to the drying system, and in combination with the dehydration rate, jointly realize the regulation of the drying process.

Images