Association analysis method for process parameters and quality indexes of tobacco primary processing

Abstract

The invention discloses a correlation analysis method for process parameters and quality indexes of tobacco primary processing, which comprises the following steps: acquiring process parameters and quality indexes corresponding to each process of tobacco primary processing, and forming a corresponding process parameter data matrix and a corresponding quality index data matrix; all the process parameters and the quality indexes are subjected to normalization processing; calculating the relative entropy of each process parameter and the quality index; calculating an absolute difference value between each process parameter and each quality index element and two pole differences in an absolute difference value sequence; according to each relative entropy, the absolute difference value and the two-pole difference, calculating a grey correlation degree; and according to the grey correlation degree, calculating a comprehensive correlation degree score for each process parameter and the corresponding quality index. According to the method, more process parameters influencing the quality indexes of the cut tobacco making process can be more accurately obtained, and the stability of homogenizing production of cigarette products is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of tobacco shred quality control, and more specifically, to a method for analyzing the correlation between process parameters and quality indicators of tobacco shred. Background Art

[0002] The tobacco silk making process includes loosening and rehydration, adding materials, cutting, and proportional blending. Its main function is to process tobacco sheets and convert tobacco sheets into tobacco shreds with uniform proportions that meet the quality requirements of manufacturers, which is of great significance to ensuring the sensory quality of cigarette products. The existing silk making process has complex coupling between the various processes, and different batches of raw materials and different process parameters will lead to differences in sensory quality. There are many data collection methods for existing silk making equipment, but the data utilization rate is low. There is a lack of reasonable use of collected data, which leads to the silk making process parameters being mainly controlled by manual experience. The correlation and influence relationship between each process is vaguely defined, and it can only be controlled within the approximate quality indicators. It is difficult to achieve precise control of the silk making process, and it cannot effectively support the intelligent control and high-quality development of the cigarette manufacturing process. The existing correlation relationship evaluation method for the silk making process is affected by the coupling between processes, the complexity of the correlation relationship, and the huge number of process parameter combinations. The evaluation method is mainly based on the hierarchical analysis method and the TOPSIS method, and subjective factors are difficult to avoid. Moreover, as the core link of cigarette production, the silk making process has a direct impact on the sensory quality, combustion performance and consumer experience of cigarette products. Therefore, how to more deeply explore the correlation between various parameters of the silk-making process and quality indicators, how to make full use of the sensor data between equipment, explore the multi-factor potential, cross-time and space, and multi-influence chain correlation between each process of the silk-making process and quality indicators, find out the main factors affecting quality indicators, and form a data-driven silk-making process analysis and precise control method, which can more accurately obtain more process parameters that affect the quality indicators of the silk-making process, and is of great significance to the homogenized production of cigarette products. Summary of the invention

[0003] The present invention provides a correlation analysis method for process parameters and quality indicators of tobacco silk making, which solves the problem that the correlation between process parameters and quality indicators in existing silk making production depends on manual experience and the process quality cannot be accurately regulated objectively. More process parameters that affect the quality indicators of the silk making process can be obtained more accurately, and the stability of the homogenized production of cigarette products is improved.

[0004] To achieve the above objectives, the present invention provides the following technical solutions:

[0005] A method for correlation analysis between process parameters and quality indicators of tobacco shreds, comprising:

[0006] Obtaining the process parameters and quality indicators corresponding to each process of tobacco silk making, and forming the corresponding process parameter data matrix and quality indicator data matrix;

[0007] Normalize all process parameters and quality indicators;

[0008] Calculate the relative entropy of each process parameter and quality index;

[0009] Calculate the absolute difference between each process parameter and each quality index element and the two extreme differences within the absolute difference sequence;

[0010] The grey relational degree is calculated according to the relative entropy, the absolute difference and the two extreme differences;

[0011] According to the grey correlation degree, a comprehensive correlation score is calculated for each process parameter and the corresponding quality index.

[0012] Preferably, the process parameters include: flow rate of each process, water addition ratio, valve opening, hot air temperature, and water addition flow rate.

[0013] Preferably, the quality indicators include: discharge temperature and discharge moisture content of each process.

[0014] Preferably, the forming of the corresponding process parameter data matrix and quality indicator data matrix includes:

[0015] The various process parameters included in the process are marked as , , … , where i represents the number of species;

[0016] The process parameters corresponding to each process are marked as , j represents the number of data contained, is the jth process parameter included in the i-th process parameter;

[0017] The process parameter data matrix is:

[0018] , where n is the number of types corresponding to the process parameters, and m is the number of data corresponding to each group.

[0019] Preferably, the forming of the corresponding process parameter data matrix and quality indicator data matrix further includes:

[0020] The various quality indicators of the process output are marked as , , … , and mark the quality index corresponding to each process as , is the jth quality indicator contained in the i-th quality indicator;

[0021] The quality indicator data matrix is: , where h is the number of types of quality indicators.

[0022] Preferably, the normalization of all process parameters and quality indicators includes:

[0023] right All elements in are normalized using the formula: , represents the normalized process parameters;

[0024] right All elements in are normalized using the formula: , Represents the normalized quality index.

[0025] Preferably, the calculation of the relative entropy of each process parameter and the quality index includes:

[0026] The calculation formula is as follows: , where Represents a sequence of process parameters Quality indicator sequence The relative entropy of , l is the number of data corresponding to each group.

[0027] Preferably, the step of calculating the absolute difference between each process parameter and each quality indicator element and the two extreme differences in the absolute difference sequence includes:

[0028] For any process parameter sequence and quality indicator sequence , corresponding to the absolute difference The calculation formula is: ;

[0029] By calculating the absolute difference between all process parameters and the specified quality index, an array consisting of absolute difference sequences is obtained. , and get the The extreme difference of internal elements and ,in express The global minimum value in express The global maximum value in .

[0030] Preferably, the calculation of the correlation degree according to the relative entropy, the absolute difference and the two-pole difference comprises:

[0031] Relevance value The calculation formula is: .

[0032] Preferably, scoring each process parameter and the corresponding quality indicator according to the correlation degree includes:

[0033] The formula is used to calculate the average correlation of each sequence to obtain the comprehensive correlation score of each element. The calculation formula is: .

[0034] The present invention provides a method for analyzing the correlation between the process parameters and the quality indexes of tobacco silk making, firstly, the relative entropy method is used to judge the distribution similarity between the input process parameters and the output quality indexes, the resolution coefficient of the grey correlation analysis is determined according to the distribution similarity, and finally the correlation score between the input process parameters and the output quality indexes is calculated. The method solves the problem that the correlation between the process parameters and the quality indexes in the existing silk making production depends on manual experience and the process quality cannot be accurately regulated objectively, more process parameters affecting the quality indexes of the silk making process can be obtained more accurately, and the stability of the homogenized production of cigarette products is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the specific embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below.

[0036] Figure 1 It is a schematic diagram of a correlation analysis method between process parameters and quality indicators of tobacco shreds provided by the present invention.

[0037] Figure 2 It is a schematic diagram of relative entropy provided by an embodiment of the present invention.

[0038] Figure 3 It is a schematic diagram of a tobacco thread making process provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0039] In order to enable persons skilled in the art to better understand the solutions of the embodiments of the present invention, the embodiments of the present invention are further described in detail below in conjunction with the accompanying drawings and implementation modes.

[0040] In view of the problem that the correlation between process parameters and quality indicators of the current tobacco silk production line relies on manual experience and cannot objectively and accurately control the process quality, the present invention provides a correlation analysis method between process parameters and quality indicators of tobacco silk production, which aims to explore the dynamic correlation between silk production process parameters and quality indicators, find out the process parameters that affect the quality indicators, and improve the stability of homogenized production of cigarette products.

[0041] like Figure 1 As shown, a correlation analysis method for process parameters and quality indicators of tobacco shreds includes:

[0042] S1: Obtaining the process parameters and quality indicators corresponding to each process of tobacco silk making, and forming the corresponding process parameter data matrix and quality indicator data matrix;

[0043] S2: Normalize all process parameters and quality indicators.

[0044] S3: Calculate the relative entropy of each process parameter and quality index.

[0045] S4: Calculate the absolute difference between each process parameter and each quality indicator element and the two extreme differences within the absolute difference sequence.

[0046] S5: Calculate the grey relational degree according to the relative entropy, the absolute difference and the two extreme differences.

[0047] S6: Calculate a comprehensive correlation score for each process parameter and the corresponding quality index according to the grey correlation.

[0048] Specifically, the following steps are included: a). Collecting the process parameters and quality indicators of silk thread. The silk thread process includes loosening and moisture recovery, primary feeding, secondary feeding, leaf silk drying and other processes. The process parameters include the flow rate, water addition ratio, valve opening, hot air temperature, water addition flow rate and the like of each process.

[0049] The process parameters included in the process are marked as , , ... . Where n represents the type of process parameters. For a specific process parameter sequence , the parameters of each process included are recorded as , , ... , where m represents the process parameter sequence The number of elements contained. represents the jth process parameter contained in the i-th process parameter. These process parameter sequences form the following matrix:

[0050] ;

[0051] The quality indicators include the discharge temperature and moisture content of each process. The quality indicators of each process output are marked as , , ... .in Represents the type of quality indicator. For the specified k-th group of quality indicator sequence , the parameters of each quality index are recorded as , , ... , where the process parameters represents the jth quality indicator output from the i-th quality indicator. These different types of quality indicator sequences form the following matrix:

[0052] ;

[0053] In this patent, n is uniformly recorded as the number of process types, h is the number of quality indicator types, and m is the number of data in each set of process parameter sequences or quality indicators. It should be noted that the collected process parameter and quality indicator data should be steady-state data, that is, data after the production line is running stably. Non-steady-state data is not within the scope of this patent.

[0054] b). Normalize all process parameters and quality indicators.

[0055] For any process parameter sequence ,right The method for normalizing all elements in is as follows: ;

[0056] In the formula, Represents the normalized process parameters, with a value range of . It means taking the maximum value of the elements in the brackets. Indicates taking the minimum value of the elements in the brackets.

[0057] For any quality index sequence Normalization is performed, and the formula is as follows: ;

[0058] Similarly, Represents the normalized quality index, with a value range of .

[0059] c). Calculate the relative entropy of each process parameter and quality index. Relative entropy is also called KL divergence (Kullback–Leibler divergence), and its calculation formula is as follows: ;

[0060] In the formula, Represents a sequence of process parameters Quality indicator sequence The relative entropy of The closer the distribution is to ,but The closer it is to 0. That is, the closer the fluctuation of the process parameters is to the fluctuation of the quality index, the The smaller the value of . represents the sum of the polynomials in it, represents a logarithmic function. In one embodiment, the calculated relative entropy is Figure 2 shown.

[0061] d). Calculate the absolute difference and two-level difference between process parameters and quality indicators. Calculate the absolute difference between each process parameter and each quality indicator element. For any process parameter sequence and quality indicator sequence , corresponding to the absolute difference The calculation is as follows: ;

[0062] In the formula, Indicates the absolute value of the internal element. According to the above formula, the absolute difference between all process parameters and the specified quality index is calculated to obtain an array consisting of absolute difference sequences. .

[0063] Further, please The extreme difference of internal elements and ,in express The global minimum value in express The global maximum value in .

[0064] e). Calculate the correlation degree based on the improved grey correlation analysis method. First, Calculate the correlation value of each sequence. The calculation method is as follows: ;

[0065] From the above formula, we can see that The value range of each element in is Among them, the resolution coefficient of the original grey relational analysis is replaced by and The relative entropy of .

[0066] because With process parameter sequence and quality indicator sequence The distribution of the two sequences changes, so this method has dynamic characteristics compared to the original grey relational analysis, and can change based on the similarity of the two sequence distributions, so it is called dynamic grey relational analysis. According to research, the resolution coefficient of grey relational analysis can be taken according to demand. ,therefore, The value range of ,when If it is greater than 19, it will be truncated.

[0067] f). Calculate the mean of the improved grey correlation of each sequence to obtain the comprehensive correlation score of each element, that is, the score of the process parameters and the corresponding quality indicators. , calculated as: .

[0068] Further, we can know the correlation score The value range of each element in is .

[0069] The above dynamic grey correlation analysis only analyzes the input process parameters and output quality indicators of a certain process, and it is difficult to conduct effective correlation analysis on silk-making lines with multiple processes and multiple variables. For the continuous silk-making line processing route, the input process parameters can affect the subsequent output quality indicators, and the output quality indicators will further serve as input parameters of subsequent processes to further affect the output quality indicators of subsequent processes. On this basis, a hierarchical dynamic grey correlation analysis is conducted for continuous processes. On the basis of the above method, the quality indicators of the output of the intermediate process are further input as part of the parameters of the subsequent process. The specific steps are as follows:

[0070] With intermediate process parameters , and the quality indicator output sequence of the previous process , And the subsequent process quality indicator output sequence , , according to the process parameter input matrix composed of this step method for . Further calculation and , The improved grey correlation relationship of the process is used to obtain the continuous improved grey correlation relationship score set between each input parameter and output of the process. , where the subscript "C" indicates "continuous", further forming a directed association network. For a production line with n processes, its association There are n-1 sets.

[0071] Furthermore, suppose the process parameters and quality index set from the previous process input are , the quality index output by this process According to steps 1-6, the correlation score between the process parameters input in the previous process and the output quality index can be obtained, which is recorded as .

[0072] Furthermore, set the quality index of the output of the next process , based on the quality indicators output from the previous process and the quality indicators of the next process output , further calculation and The correlation score between .because , , It forms an association chain, that is, Possible impact , and may further affect , and then form a continuous production correlation network. From step 5, we know that the correlation score ranges from 0 to 1. Combined with the correlation network, the subsequent process correlation score As a weight, it scores the correlation between it and all process parameters and quality indicators input in the previous process Point multiplication is used to further calculate the dynamic hierarchical grey relational score of the continuous process. The calculation method is as follows: , where the dot product symbol “·” represents and According to the actual situation, the method described in this step is not limited to the three-step silk thread, but can be extended to the multi-step silk thread.

[0073] It can be seen that the present invention provides a method for analyzing the correlation between the process parameters and quality indicators of tobacco silk making, which improves the traditional grey correlation analysis method. First, the relative entropy method is used to judge the distribution similarity between the input process parameters and the output quality indicators, and the resolution coefficient of the grey correlation analysis is determined according to the distribution similarity. Finally, the correlation score between the input process parameters and the output quality indicators is calculated. This method eliminates the instability of the resolution coefficient of the traditional grey correlation analysis under multi-factor conditions and improves the objectivity of the grey correlation analysis. It is no longer necessary to manually set the resolution coefficient of the grey correlation analysis, reducing the influence of subjective factors. The characteristics of the continuous production of silk making lines can be fully considered, and the influence of the previous and next processes on the quality indicators can be taken into account, rather than only considering a certain process. This method can solve the problem that the correlation between the process parameters and the quality indicators in the existing silk production depends on manual experience, and the process quality cannot be accurately regulated objectively. It can more accurately obtain more process parameters that affect the quality indicators of the silk making process, and improve the stability of the homogenized production of cigarette products.

[0074] The above describes in detail the structure, features and effects of the present invention based on the embodiments shown in the drawings. The above is only a preferred embodiment of the present invention, but the present invention is not limited to the scope of implementation shown in the drawings. Any changes made in accordance with the concept of the present invention, or modifications to equivalent embodiments with equivalent changes, which still do not exceed the spirit covered by the description and drawings, should be within the protection scope of the present invention.

Claims

1. A method for correlation analysis between process parameters and quality indicators of tobacco shreds, characterized in that: include: Obtaining the process parameters and quality indicators corresponding to each process of tobacco silk making, and forming the corresponding process parameter data matrix and quality indicator data matrix; Normalize all process parameters and quality indicators; Calculate the relative entropy of each process parameter and quality index; Calculate the absolute difference between each process parameter and each quality index element and the two extreme differences within the absolute difference sequence; The grey relational degree is calculated according to the relative entropy, the absolute difference and the two extreme differences; According to the grey correlation degree, a comprehensive correlation score is calculated for each process parameter and the corresponding quality index.

2. The method for correlation analysis between process parameters and quality indicators of tobacco shred according to claim 1, characterized in that: The process parameters include: flow rate of each process, water addition ratio, valve opening, hot air temperature, and water addition flow rate.

3. The method for correlation analysis of process parameters and quality indicators of tobacco shred according to claim 2, characterized in that: The quality indicators include: discharge temperature and discharge moisture content of each process.

4. The method for correlation analysis between process parameters and quality indicators of tobacco shredding according to claim 3, characterized in that: The forming of the corresponding process parameter data matrix and quality indicator data matrix includes: The various process parameters included in the process are marked as , , … , where i represents the number of species; The process parameters corresponding to each process are marked as , j represents the number of data contained, is the jth process parameter included in the i-th process parameter; The process parameter data matrix is: , where n is the number of types corresponding to the process parameters, and m is the number of data corresponding to each group.

5. The method for correlation analysis between process parameters and quality indicators of tobacco shred according to claim 4, characterized in that: The forming of the corresponding process parameter data matrix and quality index data matrix also includes: The various quality indicators of the process output are marked as , , … , and mark the quality index corresponding to each process as , is the jth quality indicator contained in the i-th quality indicator; The quality indicator data matrix is: , where h is the number of types of quality indicators.

6. The method for correlation analysis between process parameters and quality indicators of tobacco shred according to claim 5, characterized in that: All process parameters and quality indicators are normalized, including: right All elements in are normalized using the formula: , represents the normalized process parameters; right All elements in are normalized using the formula: , Represents the normalized quality index.

7. The method for correlation analysis between process parameters and quality indicators of tobacco shred according to claim 6, characterized in that: The calculation of the relative entropy of each process parameter and quality index includes: The calculation formula is as follows: , where Represents a sequence of process parameters Quality indicator sequence The relative entropy of , l is the number of data corresponding to each group.

8. The method for correlation analysis between process parameters and quality indicators of tobacco shreds according to claim 7, characterized in that: The calculation of the absolute difference between each process parameter and each quality indicator element and the two extreme differences in the absolute difference sequence includes: For any process parameter sequence and quality indicator sequence , corresponding to the absolute difference The calculation formula is: ; By calculating the absolute difference between all process parameters and the specified quality index, an array consisting of absolute difference sequences is obtained. , and get the The extreme difference of internal elements and ,in express The global minimum value in express The global maximum value in .

9. The method for correlation analysis between process parameters and quality indicators of tobacco shred according to claim 8, characterized in that: The calculating the correlation degree according to the relative entropy, the absolute difference and the two-pole difference comprises: Relevance value The calculation formula is: .

10. The method for correlation analysis between process parameters and quality indicators of tobacco shreds according to claim 9, characterized in that: Scoring each process parameter and the corresponding quality indicator according to the correlation degree includes: The formula is used to calculate the average correlation of each sequence to obtain the comprehensive correlation score of each element. The calculation formula is: .

Images