Inulin quality detection method based on data analysis

Abstract

An inulin quality detection method based on data analysis comprises the following steps that inulin sample data of all production batches are stored in a database, a data processing module extracts m inulin samples from the database to serve as training samples, the data processing module sends an inulin sample group to a feature processing module, and the feature processing module carries out feature analysis on the inulin sample group; the feature processing module extracts the feature vector of each inulin sample to obtain an inulin feature matrix. The data processing module establishes mapping between the sample feature xi and the quality rating vector Q through data annotation, obtains prior probabilities of all quality categories q1, q2,..., qn, determines a likelihood function of the inulin feature vector xnew under each quality category qi through a Gaussian model, and sends the likelihood function to the data processing module; the data processing module calculates the posterior probability of the sample xnew in each quality category qi, the data processing module obtains the corresponding posterior probabilities in all quality categories q1, q2,..., qn, and the data processing module selects the maximum posterior probability value from P (q1 | xnew), P (q2 | xnew)..., and P (qn | xnew) as the prediction quality. Quality detection can be rapidly achieved, inulin quality detection is achieved through multi-dimensional features, quality fluctuation is found in time, and follow-up process adjustment is facilitated.

Description

Technical Field

[0001] The present invention relates to the field of inulin detection, and in particular to an inulin quality detection method based on data analysis. Background Art

[0002] Inulin, as an important functional polysaccharide, is widely used in food, medicine and health products. The quality of inulin needs to be tested and classified during the production process, and the market prices of different inulin qualities vary greatly. The traditional inulin quality detection method mainly uses chromatography or mass spectrometry analysis. Specifically, the effective components of inulin are quantitatively analyzed by chromatography, and the molecular weight and structural information of the compounds in the sample are analyzed by mass spectrometry. The above method requires a special measurement laboratory, and the method is complicated to operate, has high requirements on equipment, and has a long detection time. The sample pretreatment process is cumbersome. Although the accuracy is high, it is difficult to adjust the process of the batch of inulin after the detection is completed due to the lag of the detection. In order to achieve rapid real-time detection feedback adjustment, it is necessary to introduce a detection method based on data analysis to reduce the dependence on expensive instruments and reagents, thereby reducing the overall cost of detection. Therefore, in order to solve the above problems, a method for detecting inulin quality based on data analysis is needed. Summary of the invention

[0003] In view of the shortcomings of the prior art, the present invention proposes a method for detecting inulin quality based on data analysis. The specific technical solution is as follows:

[0004] A method for detecting inulin quality based on data analysis, characterized in that:

[0005] The steps include:

[0006] S1: The inulin sample data of all production batches are stored in the database. The data processing module extracts m inulin samples from the database as training samples. The data processing module sends the inulin sample group to the feature processing module. The feature processing module extracts the feature vector of each inulin sample to obtain the inulin feature matrix

[0007]

[0008] Among them, the inulin feature matrix X represents the feature data of m samples, and each row represents a sample feature vector x i , each column represents a feature;

[0009] x i =[x i1 x i2 …x in ];

[0010] x i represents the feature vector of the i-th sample, xi1 ,x i2 ,…,x in are the n eigenvalues ​​of the inulin sample;

[0011] S2: The database stores the quality rating vector Q. The data processing module annotates the data with the sample feature x. i Establish a mapping between it and the quality rating vector Q:

[0012] in,

[0013] S3: The data processing module calculates q j The proportion of the number of samples to the total number of samples determines q j The prior probability of

[0014]

[0015] Among them, q j ∈{q 1 ,q 2 ,…,q n};

[0016] N is the total number of samples m, for q j The number of samples;

[0017] S4: The data processing module obtains all quality categories q 1 ,q 2 ,…,q n The prior probability of:

[0018] P(q 1 ),P(q 2 ),…P(q n );

[0019] S5: The data acquisition module collects new inulin samples and sends them to the feature extraction module. The feature extraction module extracts the inulin feature vector x new The data processing module determines the inulin feature vector x through the Gaussian model new In each quality category q i The likelihood function is:

[0020]

[0021] n is the characteristic dimension of inulin sample;

[0022] q j ∈{q 1 ,q 2 ,…,q n};

[0023] μj For quality category q j The mean vector of ;

[0024] |Σ j | is the quality category q j The covariance matrix of

[0025] S6: The data processing module obtains all quality categories q 1 ,q 2 ,…,q n The corresponding likelihood value in is:

[0026] P(x new |q 1 ),P(x new |q 1 )…P(x new |q n );

[0027] S7: Data processing module calculates sample x new In each quality category q i The posterior probability in :

[0028]

[0029] S8: The data processing module obtains all quality categories q 1 ,q 2 ,…,q n The corresponding posterior probability is:

[0030] P(q 1 ∣x new ),P(q 2 ∣x new )…,P(q n ∣x new );

[0031] S9: Data processing module in P(q 1 ∣x new ),P(q 2 ∣x new )…,P(q n ∣x new ) selects the largest posterior probability value as the prediction quality.

[0032] In order to better implement the present invention, it is possible to further: the data processing module processes the posterior probability P(q 1 ∣x new ),P(q 2 ∣x new )...,P(q n ∣x new) After being sorted in descending order, the data processing module selects the largest posterior probability as the prediction data.

[0033] Furthermore: the feature processing module extracts particle size, water content, sugar content, color parameters, pH value and ash content in the inulin sample as dimensions of the feature vector.

[0034] The beneficial effects of the present invention are as follows: the present invention can quickly realize quality detection, by setting a feature extraction module, quickly and in real time detect features from inulin samples, use inulin samples in the historical database to establish the prior probability of various grades, and distinguish different grades through the features detected in real time. Through the calculation of the posterior probability of samples of each category, the category with the highest posterior probability is selected as the prediction result of the grade. The quality detection of inulin is realized through multi-dimensional features, quality fluctuations are discovered in time, and subsequent process adjustments are convenient. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 It is the work flow chart of the present invention. DETAILED DESCRIPTION

[0036] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0037] like Figure 1 As shown:

[0038] A method for detecting inulin quality based on data analysis comprises the following steps:

[0039] S1: The inulin sample data of all production batches are stored in the database. The data processing module extracts m inulin samples from the database as training samples. The data processing module sends the inulin sample group to the feature processing module. The feature processing module extracts the feature vector of each inulin sample. The feature processing module extracts the particle size, water content, sugar content, color parameters, pH value and ash content from the inulin sample as the dimensions of the feature vector, and obtains the inulin feature matrix

[0040]

[0041] Among them, the inulin feature matrix X represents the feature data of m samples, and each row represents a sample feature vector x i , each column represents a feature;

[0042] xi =[x i1 x i2 ...x in ];

[0043] x i represents the feature vector of the i-th sample, x i1 ,x i2 ,...,x in are the n eigenvalues ​​of the inulin sample;

[0044] S2: The database stores the quality rating vector Q. The data processing module annotates the data with the sample feature x. i Establish a mapping between it and the quality rating vector Q:

[0045] in,

[0046] S3: The data processing module calculates q j The proportion of the number of samples to the total number of samples determines q j The prior probability of

[0047]

[0048] Among them, q j ∈{q 1 ,q 2 ,...,q n};

[0049] N is the total number of samples m, for q j The number of samples;

[0050] S4: The data processing module obtains all quality categories q 1 ,q 2 ,…,q n The prior probability of:

[0051] P(q 1 ),P(q 2 ),…P(q n );

[0052] S5: The data acquisition module collects new inulin samples and sends them to the feature extraction module. The feature extraction module extracts the inulin feature vector x new The data processing module determines the inulin feature vector x through the Gaussian model new In each quality category q i The likelihood function is:

[0053]

[0054] n is the characteristic dimension of inulin sample;

[0055] q j ∈{q 1 ,q 2 ,…,q n};

[0056] μ j For quality category q j The mean vector of ;

[0057] |Σ j | is the quality category q j The covariance matrix of

[0058] S6: The data processing module obtains all quality categories q 1 ,q 2 ,...,q n The corresponding likelihood value in is:

[0059] P(x new |q 1 ),P(x new |q 1 )...P(x new |q n );

[0060] S7: Data processing module calculates sample x new In each quality category q i The posterior probability in :

[0061]

[0062] S8: The data processing module obtains all quality categories q 1 ,q 2 ,…,q n The corresponding posterior probability is:

[0063] P(q 1 ∣x new ),P(q 2 ∣x new )...,P(q n ∣x new );

[0064] S9: Data processing module in P(q 1 ∣x new ),P(q 2 ∣x new )...,P(q n ∣x new ) selects the largest posterior probability value as the predicted quality. Specifically, the data processing module calculates the posterior probability P(q 1 ∣x new),P(q 2 ∣x new )...,P(q n ∣x new ) After being sorted in descending order, the data processing module selects the largest posterior probability as the prediction data.

[0065] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

[0066] In addition, it should be understood that although the present specification is described according to implementation modes, not every implementation mode contains only one independent technical solution. This description of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment may also be appropriately combined to form other implementation modes that can be understood by those skilled in the art.

Claims

1. A method for detecting inulin quality based on data analysis, characterized in that: The steps include: S1: The inulin sample data of all production batches are stored in the database. The data processing module extracts m inulin samples from the database as training samples. The data processing module sends the inulin sample group to the feature processing module. The feature processing module extracts the feature vector of each inulin sample to obtain the inulin feature matrix Among them, the inulin feature matrix X represents the feature data of m samples, and each row represents a sample feature vector x i , each column represents a feature; x i =[x i1 x i2 ...x in ]; x i represents the feature vector of the i-th sample, x i1 ,x i2 ,...,x in are the n eigenvalues ​​of the inulin sample; S2: The database stores the quality rating vector Q. The data processing module annotates the data with the sample feature x. i Establish a mapping between it and the quality rating vector Q: Where, Q=[q1 q2...q n ]; i=1,2,…,m; q j ∈{q1,q2,…,q n }; S3: The data processing module calculates q j The proportion of the number of samples to the total number of samples determines q j The prior probability of Among them, q j ∈{q1,q2,…,q n }; N is the total number of samples m, for q j The number of samples; S4: The data processing module obtains all quality categories q1, q2, …, q n The prior probability of: P(q1),P(q2),…P(q n ); S5: The data acquisition module collects new inulin samples and sends them to the feature extraction module. The feature extraction module extracts the inulin feature vector x new The data processing module determines the inulin feature vector x through the Gaussian model new In each quality category q i The likelihood function is: n is the characteristic dimension of inulin sample; q j ∈{q1,q2,…,q n }; μ j For quality category q j The mean vector of ; |Σ j | is the quality category q j The covariance matrix of S6: The data processing module obtains all quality categories q1, q2, …, q n The corresponding likelihood value in is: P(x new |q1),P(x new |q1)…P(x new |q n ); S7: Data processing module calculates sample x new In each quality category q i The posterior probability in : S8: The data processing module obtains all quality categories q1, q2, …, q n The corresponding posterior probability is: P(q1∣x new ),P(q2∣x new )...,P(q n ∣x new ); S9: Data processing module in P(q1|x new ),P(q2|x new )...,P(q n ∣x new ) selects the largest posterior probability value as the prediction quality.

2. The inulin quality detection method based on data analysis according to claim 1, characterized in that: The data processing module calculates the posterior probability P(q1|x new ),P(q2|x new )…,P(q n ∣x new ) After being sorted in descending order, the data processing module selects the largest posterior probability as the prediction data.

3. The inulin quality detection method based on data analysis according to claim 2, characterized in that: The feature processing module extracts particle size, water content, sugar content, color parameters, pH value and ash content in the inulin sample as dimensions of the feature vector.

Images