A method for detecting the moisture content of vinasse

Abstract

The present application provides a kind of detection method of distiller's grains moisture content based on near infrared spectroscopy, through the measurement of near infrared reflectance spectroscopy, data preprocessing and analysis are carried out by combining physicochemical metrology method, establish the prediction model of distiller's grains moisture content, and then realize the fast and accurate measurement of distiller's grains moisture content.The advantage of the present application is that it can realize rapid, nondestructive testing and does not consume chemical reagent detection, near infrared spectroscopy testing means has good test stability, the detection result has higher repeatability, and in near infrared wave band, test signal can be transmitted in optical fiber at a distance, so that through optical fiber, remote data transmission can be carried out, realize the automation and integration of entire detection process, and multi-channel signal can be realized parallel processing, greatly save the test cost.

Description

Technical Field

[0001] This invention relates to the field of baijiu (Chinese liquor) brewing, and more particularly to a method for detecting the moisture content of lees. Background Technology

[0002] Moisture content is a crucial analytical and testing item in the entire process of baijiu (Chinese liquor) brewing, as water is involved in almost every stage of the process. In the critical fermentation stage, water provides a suitable environment for the microorganisms needed for fermentation and carries various soluble nutrients, providing them with the necessary food. Furthermore, the various soluble nutrients carried in the water are also important factors affecting microbial activity and baijiu formation. The various salts in the water can dissociate into ions, which significantly influence the working process and efficiency of microorganisms. Therefore, the detection and control of moisture content has always been a vital aspect of baijiu production, significantly impacting the quality of the finished product.

[0003] In current baijiu (Chinese liquor) production processes, some steps already incorporate moisture detection technologies. For example, in the raw material stage, moisture detection typically employs methods such as drying, infrared drying, and dielectric methods. These methods generally suffer from long testing times, damage to the sample, and complex procedures. More importantly, none of these methods can achieve real-time, online, and continuous monitoring of the moisture content of raw materials and intermediate products at each stage of the brewing process in automated production lines. Therefore, they are difficult to apply in actual production and provide reference information on moisture content during the brewing process. Furthermore, in the more crucial fermentation stage, due to the inherently closed nature of fermentation pits, these sampling and analysis methods are almost entirely ineffective for detecting moisture during fermentation. Summary of the Invention

[0004] In view of this, it is indeed necessary to provide a rapid, accurate, and cost-effective method for detecting the moisture content of the lees in the brewing process of baijiu, so as to monitor the moisture content of the entire brewing process and thereby improve the efficiency of the brewing process and the quality of the finished product.

[0005] A method for detecting the moisture content of distiller's grains, comprising the following steps:

[0006] S1. Basic data collection: Collect the near-infrared reflectance spectrum of the distiller's grains sample to obtain the near-infrared spectral sample, and at the same time measure the moisture content of each distiller's grains sample using traditional methods.

[0007] S2. Select the modeling sample set and perform spectral preprocessing: Remove abnormal samples from the near-infrared spectral samples, divide the remaining near-infrared spectral samples after removing abnormal samples into a modeling sample set and a prediction sample set, and preprocess the near-infrared spectra in the modeling sample set.

[0008] S3. Establishing a Data Processing Model: Different modeling methods are used to establish prediction models for the moisture content of distiller's grains. The optimal prediction model is evaluated and selected using the coefficient of determination and root mean square error of the calibration and validation sets.

[0009] S4. System Integration Development: System integration of the entire acquisition, preprocessing, modeling and analysis process.

[0010] Furthermore, in step S1, the moisture content of the distiller's grains sample is determined using a 105°C drying method.

[0011] Further, in step S2, the near-infrared spectra in the modeling sample set are preprocessed using first derivative (FD), second derivative (SD), standard normal transform (SNV), and multivariate scattering correction (MSC), and compared with the original spectra to determine the coefficient of determination R. 2 The closer the value is to 1, the smaller the root mean square error (RMSE), which serves as the basis for selecting the optimal preprocessing method.

[0012] Furthermore, in step S3, the modeling method includes principal component regression (PCR), partial least squares (PLS), or support vector machine regression (SVR).

[0013] Furthermore, in step S4, based on the application requirements of the detection system in the production site, the entire acquisition, preprocessing, modeling and analysis are systematically integrated.

[0014] Furthermore, step S5 includes an on-site verification and optimization step following step S4.

[0015] Furthermore, in step S5, the optimization includes optimization of the spectral acquisition mode, optimization of the data analysis method, optimization of the detection method, optimization of the modeling method, expansion and optimization of the sample domain, and optimization of the spectral characteristic indicators.

[0016] Compared with existing technologies, this invention provides a method for detecting the moisture content of distiller's grains during the brewing process. This method utilizes near-infrared spectroscopy to detect the moisture content of the distiller's grains during brewing. Through research on spectral acquisition methods, data preprocessing methods, data model construction, and detection system integration methods, a rapid, accurate, and cost-effective moisture content detection system for key stages of the entire baijiu brewing process is established. This method based on distiller's grains moisture content detection has advantages such as speed, accuracy, non-contact operation (no direct contact with the distiller's grains), and independence from the working environment. It can monitor the moisture content of the entire brewing process, thereby improving brewing efficiency and finished product quality. Attached Figure Description

[0017] Figure 1 This is a flowchart of the near-infrared spectroscopy detection process for sample moisture content provided by the present invention.

[0018] Figure 2 This is the near-infrared spectrum of the distiller's grains sample.

[0019] Figure 3 It is the near-infrared spectrum of the modeling sample set.

[0020] Figure 4 The results are based on the detection of the moisture content of the predicted distiller's grains sample using an optimized model.

[0021] Explanation of main component symbols

[0022] none

[0023] The following detailed description, in conjunction with the accompanying drawings, will further illustrate the present invention. Detailed Implementation

[0024] The method for detecting the moisture content of distiller's grains provided by the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0025] Please see Figure 1 This invention provides a method for detecting the moisture content of distiller's grains, which includes the following steps:

[0026] S1. Basic data collection: Near-infrared reflectance spectra of distiller's grains samples were collected to obtain near-infrared spectral samples. At the same time, the moisture content of each distiller's grains sample was measured using traditional methods. The moisture content of the distiller's grains samples was determined by the 105℃ drying method. The collected near-infrared reflectance spectra corresponded one-to-one with the moisture content data measured by the 105℃ drying method.

[0027] S2. Selecting the modeling sample set and performing spectral preprocessing: Abnormal samples are removed from the collected near-infrared spectral samples. The remaining near-infrared spectral samples are divided into a modeling sample set and a prediction sample set. The near-infrared spectra in the modeling sample set are preprocessed using multiple methods including first derivative (FD), second derivative (SD), standard normal transform (SNV), and multivariate scattering correction (MSC) and compared with the original spectra. The coefficient of determination R0 is used to determine the modeling sample set. 2 The closer the value is to 1, the smaller the root mean square error (RMSE), which serves as the basis for selecting the optimal preprocessing method.

[0028] S3. Establish data processing models: Different modeling methods are used to establish prediction models for the moisture content of distiller's grains. The optimal prediction model is evaluated and selected by the coefficient of determination and root mean square error of the calibration set and the validation set. The modeling methods include principal component regression (PCR), partial least squares (PLS) or support vector machine regression (SVR).

[0029] S4. System Integration Development: Based on the application requirements of the detection system in the production field, the entire data acquisition, preprocessing, modeling, and analysis processes are systematically integrated; and

[0030] S5. On-site verification and optimization: Based on the optimization of spectral acquisition mode, data analysis method, detection method, modeling method, sample domain expansion and optimization, and spectral characteristic index, the accuracy and practicality of the quantitative discrimination model are repeatedly improved.

[0031] In step S1, in this embodiment, the traditional method for measuring the moisture content of each lees sample is the 105℃ drying method. The 105℃ drying method refers to drying the lees sample at a temperature of 105℃. Samples were collected from lees samples in fermentation tanks at different fermentation stages during the laboratory brewing of light-aroma baijiu. One portion of lees was divided into six equal portions, dried at 105℃ for different times, and then the mass of each lees sample was measured, and near-infrared reflectance spectra were obtained. The actual moisture content of the lees corresponding to each spectrum can be calculated based on the moisture content of the lees at the time of sampling and the mass measured after drying. Using this method, 105 lees samples were obtained with a one-to-one correspondence between lees moisture content and near-infrared reflectance spectra, and their spectra are shown below. Figure 2 As shown. Figure 2 In the figure, the vertical axis represents reflectivity.

[0032] In addition, near-infrared light refers to electromagnetic waves with a wavelength range of 780nm-2526nm (12820-3959cm-1). Its spectral information mainly reflects the overtone and sum-frequency absorption of amino groups such as CH, OH, NH, and SH in organic molecules. Quantitative determination based on the absorption characteristics of a chemical component in the near-infrared region can determine the content and composition of an unknown sample from its spectrum.

[0033] In step S2, in this embodiment, after removing the abnormal samples in step S1, 62 near-infrared spectral samples are selected as the modeling sample set. Figure 3 This is the near-infrared spectrum of the modeling sample set, where the vertical axis represents the reflectance of the distiller's grains sample.

[0034] In step S3, the multivariate data processing software The Unscrambler 10.4 was used for spectral preprocessing, and different modeling methods were employed. Some results are shown in Table 1 (evaluation statistics of each modeling method on the calibration and validation sets after unprocessed, SNV, MSC, and second-order derivative spectral preprocessing). The coefficient of determination R0 was selected from these methods. 2 The model closest to 1 with the smallest root mean square error (RMSE) is used as the final prediction model PLS (calibration set R). 2 =0.906, RMSE=0.0654; Validation set R 2 =0.863, RMSE=0.0797). In this embodiment, the original spectrum of the distiller's grains sample was used to model the model using partial least squares (PLS). The determination coefficient R of the moisture content prediction model of the distiller's grains established in this way is 0.863, RMSE=0.0797. 2 The value closest to 1 has the smallest root mean square error (RMSE).

[0035] Table 1. Evaluation statistics of each modeling method on the calibration and validation sets, both before and after preprocessing with SNV, MSC, and second-order guided spectra.

[0036]

[0037] Table 1 shows that the most suitable modeling method for this batch of modeling sample sets is to use the original spectra and employ partial least squares (PLS) for modeling. The coefficient of determination R of the prediction model for the moisture content of distiller's grains established by this method is [missing information]. 2 The value closest to 1 has the smallest root mean square error (RMSE).

[0038] In step S4, in this embodiment, given the large volume and number of fermentation tanks in the actual production site, the spectral acquisition system can be integrated into a compact anti-interference device and pre-embedded in the fermentation tank. This facilitates the spectral acquisition system to collect near-infrared reflectance spectra of the lees sample during the fermentation process. Combined with remote data transmission technology, the data is transmitted to the central control computer, and the moisture content of the lees is obtained using the data processing model established in step S3.

[0039] In step S5, the optimization includes optimizing the spectral acquisition mode, data analysis method, detection method, modeling method, sample domain expansion and optimization, and spectral characteristic indicators. Step S5 is optional and can be omitted.

[0040] Figure 4 To apply the selected prediction model to the detection results of the predicted moisture content in the collected distiller's grains samples. Figure 4 It can be seen that the predicted value of the moisture content of the distiller's grains sample is almost consistent with the measured value. Therefore, the selected prediction model has high accuracy in predicting the moisture content of the distiller's grains sample.

[0041] The method for detecting the moisture content of distiller's grains provided by this invention has the following advantages: First, this invention proposes to use near-infrared reflectance spectroscopy to rapidly detect the moisture content of distiller's grains during the brewing process. Specifically, by measuring near-infrared reflectance spectra and combining this with physicochemical metrology methods for data preprocessing and analysis, the spectral characterization mechanism is studied, and a rapid method for obtaining the moisture content of various target samples throughout the brewing process is established, thereby achieving rapid and accurate measurement of moisture content throughout the entire brewing process. Second, the target sample (distiller's grains sample) targeted by the method for detecting the moisture content of distiller's grains provided by this invention generally does not require special preprocessing. It can achieve rapid, non-destructive detection without consuming chemical reagents, making it a truly green detection method, very suitable for the specific working environment of the brewing process. Third, the near-infrared spectroscopy testing method used in this invention has good testing stability, high repeatability of the detection results, and is less affected by external factors. In the near-infrared band, the test signal can be transmitted over long distances in optical fiber, enabling long-distance data transmission and automation and integration of the entire detection process. Furthermore, multiple signals can be processed in parallel, greatly saving testing costs.

[0042] Furthermore, those skilled in the art may make other changes within the spirit of this invention. Of course, all such changes made in accordance with the spirit of this invention should be included within the scope of protection claimed by this invention.

Claims

1. A method for detecting the moisture content of distiller's grains, applied in the fermentation process of baijiu (Chinese liquor), characterized in that, Includes the following steps: S1. Basic data acquisition: Collect the near-infrared reflectance spectrum of the distiller's grains sample to obtain the near-infrared spectral sample; S2. Select the modeling sample set and perform spectral preprocessing: Remove abnormal samples from the near-infrared spectral samples, divide the remaining near-infrared spectral samples after removing abnormal samples into a modeling sample set and a prediction sample set, and preprocess the near-infrared spectra in the modeling sample set. The preprocessing includes at least multivariate scattering correction (MSC). S3. Establishing a Data Processing Model: Different modeling methods are used to establish prediction models for the moisture content of distiller's grains. The optimal prediction model is evaluated and selected using the coefficient of determination and root mean square error of the calibration and validation sets. S4. System Integration Development: System integration of the entire acquisition, preprocessing, modeling and analysis process.

2. The method for detecting the moisture content of distiller's grains as described in claim 1, characterized in that, In step S1, the moisture content of the distiller's grains sample is determined using a 105°C drying method.

3. The method for detecting the moisture content of distiller's grains as described in claim 1, characterized in that, In step S2, the near-infrared spectra in the modeling sample set are preprocessed using first derivative (FD), second derivative (SD), standard normal transform (SNV), and multivariate scattering correction (MSC), and compared with the original spectra. The closer the coefficient of determination R2 is to 1, the smaller the root mean square error RMSE, which serves as the basis for selecting the optimal preprocessing method.

4. The method for detecting the moisture content of distiller's grains as described in claim 1, characterized in that, In step S3, the modeling method includes principal component regression (PCR), partial least squares (PLS), or support vector machine regression (SVR).

5. The method for detecting the moisture content of distiller's grains as described in claim 1, characterized in that, In step S4, based on the application requirements of the detection system in the production site, the entire acquisition, preprocessing, modeling and analysis are systematically integrated.

6. The method for detecting the moisture content of distiller's grains as described in claim 1, characterized in that, Further, after step S4, a field verification and optimization step is included.

7. The method for detecting the moisture content of distiller's grains as described in claim 6, characterized in that, The optimizations include optimization of spectral acquisition modes, data analysis methods, detection methods, modeling methods, sample domain expansion and optimization, and optimization of spectral characteristic indicators.

8. The method for detecting the moisture content of distiller's grains as described in claim 1, characterized in that, In step S3, the original spectrum of the distiller's grains sample is modeled using partial least squares (PLS).

9. The method for detecting the moisture content of distiller's grains as described in claim 1, characterized in that, In step S4, the spectral acquisition system is integrated into an anti-interference device, and the anti-interference device is pre-embedded in the fermentation tank.

10. The method for detecting the moisture content of distiller's grains as described in claim 9, characterized in that, During fermentation, the spectral acquisition system collects near-infrared reflectance spectra of the lees sample and transmits the data to the central control computer using remote data transmission technology. The moisture content of the lees is then obtained using the data processing model established in step S3.

Images