阅读说明：本技术 一种白酒蒸馏的在线快速量质摘酒方法 (Online rapid quality-measuring liquor taking method for liquor distillation ) 是由 张贵宇 庹先国 翟双 朱雪梅 彭英杰 曾祥林 于 2021-08-30 设计创作，主要内容包括：本发明公开了一种白酒蒸馏的在线快速量质摘酒方法,S1：通过酒醅蒸馏和冷凝工艺获得白酒基酒；S2：通过光谱仪实时检测基酒的光谱数据；S3：对光谱数据进行滤波预处理；S4：对滤波预处理后的光谱数据进行特征提取,保留利于量质分段摘酒的特征信息；S5：依据基酒光谱数据的属性信息,建立多维分析空间,每一个维度代表一个属性信息；S6：在多维空间下对基酒光谱矩阵进行分析,获得不同摘酒段数基酒的特征波长；S7：利用特征波长,采用模式识别算法,建立基于摘酒段数的分类模型,实现在线快速无损量质摘酒方法；对白酒基酒在线实时测量,脱离人工操作,能够解决摘酒操作完全依靠经验的缺陷,并且降低工人的劳动强度,同时提高了工艺的智能化程度。(The invention discloses an online rapid quality-measuring liquor taking method for liquor distillation, which comprises the following steps of S1: obtaining white spirit base wine through the process of distilling and condensing fermented grains; s2: detecting the spectrum data of the base wine in real time through a spectrometer; s3: carrying out filtering pretreatment on the spectral data; s4: performing feature extraction on the spectrum data after filtering pretreatment, and reserving feature information which is beneficial to quality subsection liquor picking; s5: establishing a multidimensional analysis space according to the attribute information of the base wine spectral data, wherein each dimension represents attribute information; s6: analyzing the spectrum matrix of the base liquor in a multidimensional space to obtain characteristic wavelengths of the base liquor of different liquor-picking sections; s7: establishing a classification model based on the number of liquor picking sections by using the characteristic wavelength and adopting a mode identification algorithm to realize an online rapid lossless quality-measuring liquor picking method; the method has the advantages that the white spirit base wine is measured on line in real time, manual operation is separated, the defect that the wine picking operation completely depends on experience can be overcome, the labor intensity of workers is reduced, and meanwhile, the intelligent degree of the process is improved.)

1. An online rapid quality-measuring liquor taking method for liquor distillation is characterized by comprising the following steps:

s1: obtaining white spirit base wine through the process of distilling and condensing fermented grains;

s2: detecting the spectrum data of the base wine in real time through a spectrometer;

s3: carrying out filtering pretreatment on the spectral data;

s4: performing feature extraction on the spectrum data after filtering pretreatment, and reserving feature information which is beneficial to quality subsection liquor picking;

s5: establishing a multidimensional analysis space according to the attribute information of the base wine spectral data, wherein each dimension represents attribute information;

s6: analyzing the spectrum matrix of the base liquor in a multidimensional space to obtain characteristic wavelengths of the base liquor of different liquor-picking sections;

s7: and establishing a classification model based on the number of liquor picking sections by using the characteristic wavelength and adopting a mode identification algorithm, thereby realizing an online fast lossless quality-measuring liquor picking method.

2. The online rapid quality-measuring liquor-taking method for liquor distillation according to claim 1, characterized in that the liquor-taking section number comprises a head, a middle section and a tail, and the middle section can be secondarily segmented according to the requirements of each liquor enterprise.

3. The method as claimed in claim 1, wherein the spectrometer comprises a near infrared spectrometer, and the spectral wave number range of the near infrared spectrometer is 4000-12000cm^-1。

4. The on-line fast quality-measuring liquor extracting method for liquor distillation according to claim 1, characterized in that in step S3, the raw spectral data contains noise signals such as random noise and baseline drift, and the filtering pretreatment of the spectral data is performed through algorithms such as smoothing and derivative.

5. The online rapid quality-measuring liquor taking method for white spirit distillation according to claim 4, characterized in that in step S3, the size of the smoothing window is adjusted according to the resolution of the original spectrum and the noise signal, high-frequency noise is filtered out after smoothing processing, the characteristic absorption peak of the spectrum is retained, and linear or nonlinear baseline drift is removed through a first derivative or a second derivative.

6. The online rapid quality-measuring liquor-taking method for liquor distillation according to claim 3, characterized in that in step S2, liquor-based liquor is detected in real time as it flows through a flow cell detection probe of a near-infrared spectrometer.

7. The online rapid quality-measuring liquor-taking method for white spirit distillation according to claim 1, wherein in step S5, the attribute information at least includes base liquor near infrared spectrum data wavelength, base liquor sample number, and base liquor-taking section number.

8. The online rapid quality-measuring liquor-taking method for liquor distillation according to claim 1, characterized in that in step S6, the correlation algorithm is adopted to analyze the spectrum matrix of the base liquor in a multidimensional space to obtain an importance score coefficient matrix of each wavelength of the spectrum, and the characteristic wavelengths of the base liquor of different liquor-taking sections are obtained according to the ranking of the importance score coefficients.

9. The method for on-line fast quality liquor taking in distillation of white spirit according to claim 1, wherein in step S7, neural networks and support vector machine algorithms are adopted as pattern recognition algorithms.

Technical Field

The invention relates to the technical field of liquor taking, in particular to an online rapid quality-measuring liquor taking method for liquor distillation.

Background

Chinese spirits have experience summary of 'producing fragrance by fermentation, extracting fragrance by distillation and key in liquor picking', and it can be seen that liquor picking is a key process link for brewing white spirit. At present, most wineries pick wine by using traditional manual wine picking modes such as flower-watching wine picking, quality (tasting) wine picking, flower-watching measurement and the like, and the mode depends on manual experience, is greatly influenced by human factors, has the problems of different vinosity from person to person, instable base wine segmentation and the like.

In view of this, some scholars have developed automated liquor-picking techniques, such as:

1) the indirect measurement of the alcoholic strength is realized by detecting the density of the base liquor, and the liquor is picked in sections according to the alcoholic strength; however, the relation between the alcoholic strength and the density of the base liquor is influenced by the temperature, the accuracy of sectional liquor picking is influenced by the condensation temperature of a condenser, and the alcoholic strength cannot accurately reflect the comprehensive effect of the flavor components;

2) taking the time and flow of flowing wine as a sectional judgment basis according to the manual wine picking experience; however, the quality of the base wine is directly determined by the fermented grains with different fermentation qualities, the fermented grains in different cellars, different layers of the same cellars and different fermentation rounds of the same cellars have different fermentation conditions, and the accuracy of segmented wine picking is very poor only by taking the time and flow of flowing wine as the judgment basis of segmentation;

3) judging the size and the dissipation speed of hops by an image analysis method according to the manual liquor picking experience to realize segmented liquor picking; however, "picking up the wine by watching the flowers" is the experience summary of the brewer, and the image analysis method only replaces the naked eyes, so that the judgment error caused by factors such as eye fatigue is avoided, but the scientific judgment basis is not realized;

4) and monitoring the steam temperature of the wine, and comparing the monitored temperature with a preset sectional wine picking temperature to realize sectional wine picking.

5) And monitoring the steam pressure of the wine, and comparing the monitored pressure with the preset sectional pressure to realize sectional wine picking.

For the 4 th and 5 th points, the liquor collection is characterized by 'slow fire distillation and low temperature flow liquor', proper pressure and temperature are favorable for extracting trace aroma substance components, but currently, only the monitored pressure and temperature are respectively compared with fixed preset values. The quality of liquor picking is the comprehensive influence of a plurality of monitoring parameters, and the relation among the parameters is not considered in the current automatic liquor picking method. The methods still rely on manual experience to determine the judgment basis for automatically and sectionally picking the wine, and scientific quality-measuring wine picking is not realized. The quality of the base wine is the comprehensive action of alcohol and flavor-developing compounds, and the single sectional judgment basis is inaccurate. Therefore, an online rapid quality-measuring liquor taking method for liquor distillation, which can solve the problems, is urgently needed.

Disclosure of Invention

Aiming at the defects in the prior art, the invention provides the on-line quick quality-measuring liquor-picking method for liquor distillation, which can solve the defect that liquor-picking operation completely depends on experience by measuring liquor base liquor on line in real time through a spectrometer and separating from manual experience operation, reduces the labor intensity of workers and improves the intelligence degree of the process.

In order to achieve the aim, the invention provides an online rapid quality-measuring liquor taking method for liquor distillation, which comprises the following steps: s1: obtaining white spirit base wine through the process of distilling and condensing fermented grains; s2: detecting the spectrum data of the base wine in real time through a spectrometer; s3: carrying out filtering pretreatment on the spectral data; s4: performing feature extraction on the spectrum data after filtering pretreatment, and reserving feature information which is beneficial to quality subsection liquor picking; s5: establishing a multidimensional analysis space according to the attribute information of the base wine spectral data, wherein each dimension represents attribute information; s6: analyzing the spectrum matrix of the base liquor in a multidimensional space to obtain characteristic wavelengths of the base liquor of different liquor-picking sections; s7: and establishing a classification model based on the number of liquor picking sections by using the characteristic wavelength and adopting a mode identification algorithm, thereby realizing an online fast lossless quality-measuring liquor picking method. The online real-time measurement of the white spirit base liquor through the spectrometer is separated from manual experience operation, the defect that liquor picking operation completely depends on experience can be overcome, the labor intensity of workers is reduced, and meanwhile, the intelligent degree of the process is improved. Realizing quality-based liquor taking according to the coordination of aroma-producing and flavor-producing compounds in the base liquor body; meanwhile, the accuracy of sectional liquor picking and the consistency of the quality of the base liquor are improved, and the timeliness of online liquor picking in the liquor picking process is realized.

Preferably, the number of liquor-picking sections comprises a foreshot, a middle-section liquor and a feints, and the middle-section liquor can be secondarily segmented according to the requirements of each liquor enterprise. The principle that the base liquor falls into a liquor container to form different foam sizes and retention time is that the base liquor is based on the mixed liquor of alcohol and water with different concentrations and has different surface tensions under certain pressure and temperature. The foam generated by the alcohol is easy to dissipate due to small tension; the alcohol concentration is gradually reduced in the distillation process, the dissipation speed of foam generated by the alcohol is continuously reduced, meanwhile, the content of water in miscible and alcohol is gradually increased, the relative density of the water is higher than that of the alcohol, the tension is high, and the dissipation speed of the foam is low. In the traditional process, wine is picked in sections by looking at flowers, but the technical requirement on workers is high, particularly under the conditions that the number of sections is large, and the difference of hops of each section is very slight, the wine cannot be distinguished only by the traditional technology, the spectral analysis of the method accurately measures the spectral data of the whole process through an intelligent instrument, and the number of the sections for wine picking can be distinguished and detected through qualitative analysis.

Preferably, in step S2, the spectrometer includes a near-infrared spectrometer, and the spectrum wave number range of the near-infrared spectrometer is 4000-12000cm^-1. The content of aroma and flavor compounds in the white spirit is very low, the content of a plurality of compounds is lower than the lowest detection limit of a conventional analysis and detection instrument, and the analysis of the conventional detection instrument is long in time consumption and requires complex treatment on a detected object. The near infrared spectrum can solve the problems, the analysis speed is high, samples do not need to be processed, and the comprehensive reaction of the near infrared spectrum on the hydrogen-containing groups of organic molecules in the white spirit to be detected is utilized to explain the comprehensive coordination effect of the aroma-generating and flavor-generating compounds. Detection by other spectrometers is also possible.

Preferably, in step S3, the raw spectral data includes noise signals such as random noise and baseline drift, and the spectral data is subjected to filtering preprocessing by algorithms such as smoothing and derivative. The analysis method of the present invention is not limited to the above-described feature extraction and dimension reduction method.

Preferably, in step S3, the size of the smoothing window is adjusted according to the resolution of the original spectrum and the noise signal, the high-frequency noise is filtered after the smoothing process, the spectral characteristic absorption peak is retained, and then the linear or linear baseline drift is removed by the first derivative or the second derivative.

Preferably, in step S2, the white spirit base is detected in real time as the white spirit base flows through the flow cell detection probe of the near-infrared spectrometer.

Preferably, the near infrared spectrum data belong to high-dimensional data, the full spectrum data are utilized for analysis modeling, the calculated amount is large, the model robustness is poor, the spectral data dimensionality is reduced through the characteristic extraction of the spectral data, and the characteristic information which is beneficial to quality subsection liquor picking is reserved.

Preferably, in step S5, the attribute information at least includes a base liquor near infrared spectrum data wavelength, a base liquor sample number, and a base liquor alcohol-extraction stage number. And establishing a multidimensional analysis space according to the attribute information of the base wine near infrared spectrum data, including the base wine near infrared spectrum data wavelength, the base wine sample number, the base wine picking section number and the like, wherein each dimension represents one attribute information. The traditional spectral analysis is established in a two-dimensional space, the structural characteristic information of a detected object is ignored, all attributes are converted into the two-dimensional space, and the inaccuracy of a complex detected object classification model is caused.

Preferably, in step S6, the spectrum matrix of the base wine is analyzed in a multidimensional space by using a correlation algorithm to obtain an importance score coefficient matrix of each wavelength of the spectrum, and the characteristic wavelengths of the base wine according to different wine-picking sections are obtained according to the ranking of the importance score coefficients.

Preferably, in step S7, algorithms such as neural network and support vector machine are adopted as the pattern recognition algorithm. The classification modeling method of the present invention is not limited to the pattern recognition method described.

The invention has the beneficial effects that: compared with the prior art, the invention provides an online rapid quality-measuring liquor taking method for liquor distillation, which comprises the following steps: s1: obtaining white spirit base wine through the process of distilling and condensing fermented grains; s2: detecting the spectrum data of the base wine in real time through a spectrometer; s3: carrying out filtering pretreatment on the spectral data; s4: performing feature extraction on the spectrum data after filtering pretreatment, and reserving feature information which is beneficial to quality subsection liquor picking; s5: establishing a multidimensional analysis space according to the attribute information of the base wine spectral data, wherein each dimension represents attribute information; s6: analyzing the spectrum matrix of the base wine in a multidimensional space to obtain characteristic wavelengths of the base wine according to different wine-picking sections; s7: and establishing a classification model based on the number of liquor picking sections by using the characteristic wavelength and an identification algorithm, and realizing an online rapid lossless quality-measuring liquor picking method. The online real-time measurement of the white spirit base liquor through the spectrometer is separated from manual experience operation, the defect that liquor picking operation completely depends on experience can be overcome, the labor intensity of workers is reduced, and meanwhile, the intelligent degree of the process is improved. The quality-based liquor taking can be realized according to the coordination effect of the aroma-producing and flavor-producing compounds in the base liquor body; meanwhile, the accuracy of sectional liquor picking and the consistency of the quality of the base liquor are improved, and the timeliness of online liquor picking in the liquor picking process is realized.

Drawings

FIG. 1 is a near infrared spectrum of the present invention;

FIG. 2 is a simplified step structure of the present invention;

FIG. 3 is a flow chart illustrating the detailed steps of the present invention.

Detailed Description

In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. The components of embodiments of the present invention generally described and illustrated in the figures herein may be arranged and designed in a wide variety of different configurations. The following detailed description of the embodiments of the present invention, presented in the figures, is not intended to limit the scope of the invention, as claimed, but is merely representative of selected embodiments of the invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

It should be noted that: like reference numbers and letters refer to like items in the following figures, and thus, once an item is defined in one figure, it need not be further defined and explained in subsequent figures. In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc. indicate orientations or positional relationships based on the orientations or positional relationships shown in the drawings or the orientations or positional relationships that the products of the present invention are conventionally placed in use, and are only used for convenience in describing the present invention and simplifying the description, but do not indicate or imply that the devices or elements referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus, should not be construed as limiting the present invention. Furthermore, the terms "first," "second," "third," and the like are used solely to distinguish one from another and are not to be construed as indicating or implying relative importance.

The said collection of liquor from flower is the traditional art of mastering the alcohol content in the distillation process of white spirit, and is used up to now. The 'flower watching' means that the alcohol content of the distillate is known by observing the size and the retention length of the hops; the liquor picking refers to that when liquor flows, liquor flowing time is gradually increased, alcohol concentration is gradually reduced from high, flavor components are continuously changed, and base liquor is picked and stored in sections according to liquor body quality. Therefore, the term "picking up the spent liquor" refers to the process of distilling white liquor by picking up spent liquor so as to separate the high-alcohol liquor from the low-alcohol liquor (tail or top of the liquor). After the wine base flows out, the wine head is firstly pinched off, and then the wine base is divided into four grades of super-excellent, superior first grade and superior second grade through first-smelling, second-watching and third-grade. But all the wine with strong aroma, large hop and sweet taste is high-quality wine. The segments of the extracted base liquor are mainly observed by sensory evaluation according to experience.

The near infrared spectrum region is consistent with the frequency combination of the vibration of the hydrogen-containing group (O-H, N-H, C-H) in the organic molecule and the absorption region of each level of frequency multiplication, the characteristic information of the hydrogen-containing group in the organic molecule in the sample can be obtained by scanning the near infrared spectrum of the sample, and the analysis of the sample by using the near infrared spectrum technology has the advantages of convenience, rapidness, high efficiency, accuracy, lower cost, no damage to the sample, no consumption of chemical reagents, no environmental pollution and the like, so the technology is favored by more and more people. The invention adopts the near-infrared spectrometer to carry out real-time online measurement on the white spirit base wine, and has the advantages of low cost, rapid detection and no influence on the environment.

Example 1: referring to fig. 1 to 3, the invention discloses an online rapid quality-measuring liquor-taking method for liquor distillation, which comprises the following steps: s1: obtaining white spirit base wine through the process of distilling and condensing fermented grains; s2: detecting the spectrum data of the base wine in real time through a spectrometer; s3: carrying out filtering pretreatment on the spectral data; s4: performing feature extraction on the spectrum data after filtering pretreatment, and reserving feature information which is beneficial to quality subsection liquor picking; s5: establishing a multidimensional analysis space according to the attribute information of the base wine spectral data, wherein each dimension represents attribute information; s6: analyzing the spectrum matrix of the base liquor in a multidimensional space to obtain characteristic wavelengths of the base liquor of different liquor-picking sections; s7: and establishing a classification model based on the number of liquor picking sections by using the characteristic wavelength and an identification algorithm, and realizing an online rapid lossless quality-measuring liquor picking method. The online real-time measurement of the white spirit base liquor through the spectrometer is separated from manual experience operation, the defect that liquor picking operation completely depends on experience can be overcome, the labor intensity of workers is reduced, and meanwhile, the intelligent degree of the process is improved. The quality-based liquor taking can be realized according to the coordination effect of the aroma-producing and flavor-producing compounds in the base liquor body; meanwhile, the accuracy of sectional liquor picking and the consistency of the quality of the base liquor are improved, and the timeliness of online liquor picking in the liquor picking process is realized.

Example 2: referring to fig. 1 to 3, the number of liquor-picking sections in the present embodiment includes a head, a middle section and a tail, and the middle section can be secondarily segmented according to the requirements of each liquor enterprise. The principle that the base liquor falls into a liquor container to form different foam sizes and retention time is that the base liquor is based on the mixed liquor of alcohol and water with different concentrations and has different surface tensions under certain pressure and temperature. The foam generated by the alcohol is easy to dissipate due to small tension; the alcohol concentration is gradually reduced in the distillation process, the dissipation speed of foam generated by the alcohol is continuously reduced, meanwhile, the content of water in miscible and alcohol is gradually increased, the relative density of the water is higher than that of the alcohol, the tension is high, and the dissipation speed of the foam is low. In the traditional process, wine is picked in sections by looking at flowers, but the technical requirement on workers is high, particularly under the conditions that the number of sections is large, and the difference of hops of each section is very slight, the wine cannot be distinguished only by the traditional technology, and the spectral analysis of the method accurately measures the spectral data of the whole process through an intelligent instrument, and can distinguish and detect the number of the sections of wine picked through quantitative analysis.

Example 3: referring to fig. 1 to 3, in step S2 of the present embodiment, the spectrometer includes a near-infrared spectrometer, and the spectral wave number range of the near-infrared spectrometer is 4000-12000cm^-1. The content of aroma and flavor compounds in the white spirit is very low, the content of a plurality of compounds is lower than the lowest detection limit of a conventional analysis and detection instrument, and the analysis of the conventional detection instrument is long in time consumption and requires complex treatment on a detected object. The near infrared spectrum can solve the problems, the analysis speed is high, samples do not need to be processed, and the comprehensive reaction of the near infrared spectrum on hydrogen-containing groups in organic molecules in the white spirit to be detected is utilized to explain the comprehensive coordination effect of the aroma-generating and flavor-generating compounds. Detection by other spectrometers is also possible. In step S2, the white spirit base wine is detected in real time while flowing through the flow cell detection probe of the near-infrared spectrometer.

Example 4: referring to fig. 1 to fig. 3, in step S3 of the present embodiment, the original spectral data includes noise signals such as random noise and baseline drift, and the filtering preprocessing of the spectral data is performed through algorithms such as smoothing and derivative. The analysis method of the present invention is not limited to the above-described feature extraction and dimension reduction method. In step S3, the size of the smoothing window is adjusted according to the resolution of the original spectrum and the noise signal, high-frequency noise is filtered after smoothing, the spectral characteristic absorption peak is retained, and linear and nonlinear baseline drift is removed by the first derivative or the second derivative.

Example 5: referring to fig. 1 to fig. 3, in step S5 of the present embodiment, the attribute information at least includes a base liquor near infrared spectrum data wavelength, a base liquor sample number, and a base liquor alcohol-picking section number. And establishing a multidimensional analysis space according to the attribute information of the base wine near infrared spectrum data, including the base wine near infrared spectrum data wavelength, the base wine sample number, the base wine picking section number and the like, wherein each dimension represents one attribute information. The traditional spectral analysis is established in a two-dimensional space, the structural characteristic information of a detected object is ignored, all attributes are converted into the two-dimensional space, and the inaccuracy of a complex detected object classification model is caused.

Example 6: referring to fig. 1 to 3, in step S6 of the present embodiment, a correlation algorithm is used to analyze the spectrum matrix of the base liquor in a multidimensional space to obtain an importance score coefficient matrix of each wavelength of the spectrum, and the characteristic wavelengths of the base liquors of different liquor-picking sections are obtained according to the ranking of the importance score coefficients. In step S7, the present embodiment includes using algorithms such as neural network and support vector machine as the pattern recognition algorithm. The classification modeling method of the present invention is not limited to the pattern recognition method described. In the embodiment, the near infrared spectrum data belongs to high-dimensional data, the full spectrum data is utilized for analysis and modeling, the calculated amount is large, the model robustness is poor, the spectral data dimension is reduced through the characteristic extraction of the spectral data, and the characteristic information which is beneficial to quality subsection liquor picking is reserved.

Example 7: referring to fig. 1 to 3, the operation steps of the present embodiment are as follows:

1) obtaining the base liquor of the white spirit through the fermented grains distillation and condensation process, wherein the number of liquor picking sections can be divided into a head section, a middle section liquor and a tail section, and the middle section liquor can be segmented according to the requirements of each liquor enterprise;

2) the liquor base liquor flows through a flow cell detection probe of the near-infrared spectrometer to obtain near-infrared spectrum data of the base liquor in real time, wherein the spectrum wave number range is 4000-^-1(ii) a In particular, as shown in FIG. 1, the spectral wavenumber range is 4500-12000cm^-1

3) The original near infrared spectrum data contains noise signals such as random noise, baseline drift and the like, and the filtering pretreatment of the spectrum data is carried out through algorithms such as smoothing, derivative and the like; the size of the smoothing window is adjusted according to the resolution of the original spectrum and the noise signal, high-frequency noise is filtered after smoothing processing, the spectral characteristic absorption peak is reserved, and linear or nonlinear baseline drift is removed through a first derivative or a second derivative.

4) The near infrared spectrum data belong to high-dimensional data, the full spectrum data are utilized for analysis modeling, the calculated amount is large, the model robustness is poor, the spectral data dimensionality is reduced through the characteristic extraction of the spectral data, and the characteristic information which is beneficial to quality subsection liquor picking is reserved;

5) and establishing a multidimensional analysis space according to the attribute information of the base wine near infrared spectrum data, including the base wine near infrared spectrum data wavelength, the base wine sample number, the base wine picking section number and the like, wherein each dimension represents one attribute information. The innovation of establishing the multidimensional analysis space is as follows: the traditional spectral analysis is established in a two-dimensional space, the structural characteristic information of a detected object is ignored, all attributes are converted into the two-dimensional space, and the inaccuracy of a complex detected object classification model is caused.

6) And analyzing the near-infrared spectrum matrix of the base wine in a multidimensional space by adopting a correlation algorithm to obtain an importance score coefficient matrix of each wavelength of the spectrum, and obtaining the characteristic wavelengths of the base wine according to different wine-picking sections according to the sequencing of the importance score coefficients.

7) And establishing a classification model based on the number of liquor picking sections by using the characteristic wavelength and adopting pattern recognition algorithms such as a neural network, a support vector machine and the like, thereby realizing an online rapid lossless quality-free liquor picking method.

The invention has the advantages that:

1) the defect that the liquor picking operation completely depends on experience is overcome, the labor intensity of workers is reduced, and the intelligent degree is improved;

2) realizing quality-based liquor taking according to the coordination of the aroma-producing and flavor-producing compounds in the base liquor body;

3) the accuracy of sectional liquor picking and the consistency of the quality of the base liquor are improved;

4) and the timeliness of online liquor picking in the liquor picking process is realized.

The above disclosure is only a few specific embodiments of the present invention, but the present invention is not limited thereto, and the technical methods of the patent application can also be applied to quality grading, authenticity identification and other detection of other products (not limited to white spirit), and all fall within the scope of the present invention.