Tobacco cutting section tobacco leaf cumulative weight prediction method based on machine learning

Abstract

The invention belongs to the technical field of cigarette primary processing, and discloses a tobacco cutting section tobacco leaf cumulative weight prediction method based on machine learning, which comprises the following steps: during batch production of a tobacco cutting section, inputting factors related to tobacco leaf cumulant, climate related factors and production time period variables into a trained online prediction tobacco leaf cumulant linear regression model; the online prediction tobacco leaf cumulant linear regression model outputs the real-time online prediction belt weigher cumulant; after batch production of the tobacco cutting section is finished, factors related to the tobacco cumulant, climate related factors and production time period variables are input into the trained off-line prediction tobacco cumulant linear regression model, and the off-line prediction tobacco cumulant linear regression model outputs the off-line prediction belt scale cumulant. According to the invention, high-precision prediction of the tobacco leaf cumulative weight in the shredding section is realized.

Description

Technical Field

[0001] The invention belongs to the technical field of cigarette shredding, and in particular relates to a method for predicting the cumulative weight of tobacco leaves in a shredding section based on machine learning. Background Art

[0002] In the process of making cigarettes, from the original cigarette box leaving the warehouse to the finished tobacco entering the cabinet, many key processes are required in the middle. The entire processing process is cumbersome and the process route is long, resulting in many loss points. Accurate statistics and stable control of tobacco flow in the cutting section, as well as prediction of the cumulative weight of tobacco based on belt scales and process data fitting, are effective ways and urgent needs for factories to improve the quality of the cutting process and the output rate. In addition, the belt scale in the cutting section is a key equipment for tobacco weighing control and tobacco production quality reporting in the process of making tobacco, and its equipment status and measurement accuracy are crucial.

[0003] Therefore, how to accurately predict the cumulative weight of tobacco leaves in the shredding section and monitor the status of the belt scale equipment on this basis has become a technical problem that urgently needs to be solved in this field. Summary of the invention

[0004] The purpose of the present invention is to overcome the deficiencies in the prior art and provide a method for predicting the cumulative weight of tobacco leaves in a shredding section based on machine learning, so as to solve the problem of difficulty in predicting the cumulative weight of tobacco leaves in the shredding section.

[0005] To solve the above technical problems, the present invention is implemented by adopting the following solutions: The present invention provides a method for predicting the cumulative weight of tobacco leaves in a shredding section based on machine learning, comprising: When the shredding section is in batch production, factors related to tobacco leaf accumulation, climate-related factors and production time period variables are input into the trained online prediction of tobacco leaf accumulation linear regression model, and the online prediction of tobacco leaf accumulation linear regression model outputs real-time online prediction of belt scale accumulation; When the batch production of the shredding section is completed, the factors related to the tobacco leaf accumulation, climate-related factors and production time period variables are input into the trained offline prediction of tobacco leaf accumulation linear regression model, and the offline prediction of tobacco leaf accumulation linear regression model outputs the offline prediction of belt scale accumulation.

[0006] Optionally, factors related to the cumulative amount of tobacco leaves include the cumulative amount of water added to loose leaves, the cumulative amount of weighing before loose leaves, the loose leaves outlet temperature, the loose leaves outlet moisture, the loose leaves inlet moisture before cutting, the pre-distribution cabinet storage time, the distribution cabinet storage time and the cabinet change time from the pre-distribution cabinet to the distribution cabinet; climate-related factors include atmospheric temperature and atmospheric humidity; the production time period variable is the production time period.

[0007] Optionally, the training of the online prediction of tobacco leaf accumulation linear regression model and the offline prediction of tobacco leaf accumulation linear regression model includes: According to the process flow chart of the slicing section and the shredding section of the silk thread making line, several groups of factors related to the accumulation of tobacco leaves were identified; Taking tobacco silk production batches as units, factors related to the cumulative amount of tobacco leaves and historical data of the cumulative amount of the corresponding weighing before cutting and moistening are derived by batch; Process the exported historical data to remove outliers; Import climate-related factors and production time period variables, use Pearson formula to conduct correlation analysis on historical data of monthly climate-related factors, production time period variables, factors related to tobacco leaf accumulation after removing discrete points, and corresponding accumulation before cutting and moistening. Variables with correlation coefficients greater than set values ​​with corresponding accumulation before cutting and moistening after removing discrete points in monthly climate-related factors, production time period variables, and factors related to tobacco leaf accumulation after removing discrete points are used as inputs of monthly linear regression models to be trained, and corresponding accumulation before cutting and moistening after removing discrete points is used as outputs of monthly linear regression models to be trained. Generate online prediction tobacco leaf accumulation linear regression models and offline prediction tobacco leaf accumulation linear regression models corresponding to monthly training according to months. Among them, the input quantity for online prediction of tobacco leaf cumulative quantity linear regression model training is the process standard moisture content of the leaf moistening inlet before cutting when the batch of shredding section is started and the historical data of the moisture content of the leaf moistening inlet detected by the moisture content detection instrument of the leaf moistening inlet during the batch production of the shredding section; the input quantity for offline prediction of tobacco leaf cumulative quantity linear regression model training is the historical data average of the moisture content of the leaf moistening inlet before cutting for the entire batch after the batch production of the shredding section is completed.

[0008] Optionally, when the shredding section starts batch production, factors related to the tobacco leaf accumulation, climate-related factors and production time period variables are input into the trained online prediction of tobacco leaf accumulation linear regression model, including: When the shredding section is in batch production, climate-related factors, production time period variables, and factors related to tobacco leaf accumulation except for the moisture in the leaf before cutting are input into the trained online prediction linear regression model of tobacco leaf accumulation; When the shredding process starts, the standard moisture content of the pre-cut leaf moistening entrance is input into the trained online prediction model for the cumulative amount of tobacco leaves. When the shredding process is in progress, multiple groups of moisture content of the pre-cut leaf moistening entrance detected stably by the moisture detection instrument are read at intervals, and the mean of the multiple groups of moisture content of the pre-cut leaf moistening entrance is used as the arithmetic mean and input into the trained online prediction model for the cumulative amount of tobacco leaves.

[0009] Optionally, after the batch production of the shredding section is completed, factors related to the tobacco leaf accumulation, climate-related factors and production time period variables are input into the trained offline prediction of tobacco leaf accumulation linear regression model, including: When the batch production of the shredding section is completed, climate-related factors, production time period variables and factors related to the tobacco leaf accumulation amount except the moisture in the leaf before cutting are input into the trained offline prediction of tobacco leaf accumulation amount linear regression model; When the batch production of the shredding section is completed, the historical data mean of the moisture content of the leaves before cutting in the entire batch is input into the trained offline prediction linear regression model for the cumulative amount of tobacco leaves.

[0010] Optionally, the method further includes: analyzing the real-time online predicted belt scale cumulative amount, the offline predicted belt scale cumulative amount, and the belt scale measured cumulative amount, and determining whether to issue an early warning based on the analysis result, including: The real-time online predicted belt scale cumulative amount recorded at intervals is converted into a time series array, and the elements in the time series array are sorted from small to large to obtain the median of the real-time online predicted belt scale cumulative amount; If the deviation between the cumulative amount measured by the belt scale and the median of the real-time online predicted cumulative amount of the belt scale is greater than the first set threshold, an early warning is issued; If the deviation between the accumulated amount measured by the belt scale and the accumulated amount predicted by the offline belt scale is greater than the second set threshold, an early warning is issued; If the deviation between the real-time online predicted belt scale cumulative median and the offline predicted belt scale cumulative median is greater than the third set threshold, an early warning is issued.

[0011] Optionally, it also includes: When an early warning is issued, first check whether the belt scale is abnormal or not; If the belt scale calibration is not invalid, then compare the difference between the median of the real-time online predicted belt scale cumulative amount, the offline predicted belt scale cumulative amount and the belt scale measured cumulative amount; If the difference between the median of the real-time online predicted belt scale cumulative amount and the belt scale measured cumulative amount is greater than the difference between the offline predicted belt scale cumulative amount and the belt scale measured cumulative amount, it is determined that the linear regression model for online prediction of tobacco leaf cumulative amount has a deviation; If the difference between the offline predicted belt scale cumulative amount and the belt scale measured cumulative amount is greater than the difference between the median of the real-time online predicted belt scale cumulative amount and the belt scale measured cumulative amount, it is determined that the offline predicted tobacco leaf cumulative amount linear regression model has a deviation.

[0012] Optionally, it also includes: If there is a deviation in the linear regression model for online prediction of tobacco leaf accumulation, the mean of the offline predicted belt scale accumulation and the belt scale measured accumulation is imported into the linear regression model for online prediction of tobacco leaf accumulation for retraining; If there is a deviation in the offline prediction of tobacco leaf accumulation linear regression model, the median of the real-time online prediction of the belt scale accumulation and the mean of the belt scale measurement accumulation are imported into the offline prediction of tobacco leaf accumulation linear regression model for retraining.

[0013] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps of the aforementioned method for predicting the cumulative weight of tobacco leaves in a shredding section based on machine learning are implemented.

[0014] The present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the aforementioned method for predicting the cumulative weight of tobacco leaves in a shredding section based on machine learning. Beneficial Effects

[0015] 1. The present invention makes full use of the characteristics of real-time online prediction and offline prediction. Based on the combination of the two models, the cumulative weight error of the material is obtained and a status warning is given to the production line, thereby playing a role in monitoring the cumulative weight status of tobacco leaves.

[0016] 2. The present invention adopts both real-time online prediction and offline prediction methods: although the real-time online prediction model cannot directly predict accurately during the production process, it can provide dynamic prediction values ​​as intermediate variables to machine learning prediction algorithms in other cigarette making processes, and provide important process intermediate values ​​for various other artificial intelligence algorithms; although the offline prediction model can only predict after the production is completed, its prediction accuracy is very high, and it can be used as an important reference indicator for monitoring the cumulative status of production line materials.

[0017] 3. The present invention solves the problem of difficulty in fitting multiple correlated factors based on a linear regression algorithm, and uses an improved prediction method based on the mean of historical batch data for modeling and the real-time mean for prediction, which solves the problem that some input factors are in continuous change and stable data cannot be obtained.

[0018] 4. The present invention performs outlier removal and correlation analysis on the corresponding input factors, not only analyzing the strong correlation and weak correlation of the data, but also eliminating the influence of outlier data on model training to the greatest extent, thereby improving the accuracy of the prediction model and ensuring the accuracy of the prediction.

[0019] 5. The present invention performs modeling according to different months. Although the number of models is increased, the correlation with the cumulative material weight Y is rationally considered from the perspective of weather factors, which improves the logical correlation of the prediction model and ensures the accuracy of the prediction.

[0020] 6. The present invention can be retrained according to the deviation results to continuously improve the prediction accuracy of the model to achieve the purpose of accurately monitoring the cumulative weight status of tobacco leaves. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 This is a flow chart of a method for predicting the cumulative weight of tobacco leaves in a shredding section based on machine learning provided in Example 1 of the present invention; Figure 2 This is a process flow chart of the tobacco thread slicing and shredding sections in a method for predicting the cumulative weight of tobacco leaves in a shredding section based on machine learning provided in Example 1 of the present invention; Figure 3 It is a schematic diagram of exporting the format of historical data in a method for predicting the cumulative weight of tobacco leaves in a shredding section based on machine learning provided in Example 1 of the present invention; Figure 4 It is a schematic diagram of an X, Y modeling algorithm based on historical data in a method for predicting the cumulative weight of tobacco leaves in a shredding stage based on machine learning provided in Example 1 of the present invention. DETAILED DESCRIPTION

[0022] The present invention will be further described below in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and cannot be used to limit the protection scope of the present invention. Example 1

[0023] This embodiment provides a method for predicting the cumulative weight of tobacco leaves in the shredding stage based on machine learning. Figure 1 As shown, the following steps are included: Step S01: Draw a process flow chart for the silk thread slicing section and the silk thread cutting section, and identify relevant factors, such as Figure 2 As shown. In the process flow chart, multiple groups of factors related to the cumulative amount of tobacco leaves are identified and defined as X variables. Among them, variable X1 is the cumulative amount of water added to loose leaves; variable X2 is the cumulative amount of loose leaves weighed before moistening; variable X3 is the outlet temperature of loose leaves; variable X4 is the outlet moisture of loose leaves; variable X5 is the inlet moisture of leaves before cutting; variable X6 is the storage time of pre-distribution cabinet; variable X7 is the storage time of distribution cabinet; variable X8 is the cabinet change time from pre-distribution cabinet to distribution cabinet; output Y is the cumulative amount of weighed before cutting.

[0024] Step S02: Taking tobacco shred production batches as units, derive the batch mean or maximum value of X1-X8 and Y by batch. The cumulative amount and time are both the maximum values ​​within the batch, while the temperature and moisture are both the batch mean values. The production batch number is used as the identifier of the data row to derive historical data.

[0025] Step S03, historical data processing, outlier removal: According to past production records, remove the data of fault shutdown batches and test production batches, and then draw a historical trend chart for X1-X8, Y to remove the data of outlier batches.

[0026] Step S04: climate-related factors are added to the linear regression model, and variable X9 is the atmospheric temperature; 10 The variable is atmospheric humidity, which respectively takes the average temperature and average humidity of the batch of materials during the slice production time. Secondly, add the production time period variable, X 11 The variable is the production time period, and the data attribute is a discrete quantity. The attribute value of most materials in the batch during the slice production time is taken. If the production is from 7 am to 2 pm, then X 11 The variable is labeled 1 (representing the daytime warming period), and if production is from 2 pm to 7 pm, then X 11 The variable is labeled 2 (representing the daytime cooling stage). If production is from 7 pm to 7 am on the second line, then X 11 The variable is labeled as 3 (representing the night phase). The format of exporting historical data is as follows Figure 3 shown.

[0027] Step S05: Modeling by month, training different linear regression models for different months. Perform correlation analysis by month. For each month, climate-related factors, production time period variables, and factors related to the cumulative amount of tobacco leaves after removing discrete points, i.e., variables X1-X 11 The historical data of the cumulative amount of the scale before cutting and moistening, i.e. Y, corresponding to the discrete points after removal are used for correlation analysis using the Pearson formula, and the correlation matrix is ​​drawn. 11 The variables with correlation coefficient with Y less than 0.2 will be input into variables X1-X 11 The variables with correlation coefficient with Y greater than 0.2 are used as input variables of the linear regression algorithm.

[0028] Step S06: Based on the historical data of variables X and Y after correlation analysis, train the linear regression model algorithm. The X variable with a monthly Pearson correlation coefficient greater than 0.2 in step S05 is used as the input of the monthly linear regression model to be trained, and the corresponding Y is used as the output of the monthly linear regression model to be trained. The online prediction of tobacco leaf accumulation linear regression model and the offline prediction of tobacco leaf accumulation linear regression model are generated by monthly training, such as Figure 4As shown. Among them, the X5 variable in the input quantity during the training of the linear regression model for predicting the cumulative amount of tobacco leaves online is the historical data of the standard moisture content of the inlet of pre-cutting moistening leaves when the batch of the shredding section is started and the moisture content of the inlet of pre-cutting moistening leaves detected by the moisture detection instrument during the batch production of the shredding section; the X5 variable in the input quantity during the training of the linear regression model for predicting the cumulative amount of tobacco leaves offline is the historical data mean of the moisture content of the inlet of pre-cutting moistening leaves of the entire batch after the batch production of the shredding section is completed (for details, please refer to the relevant part of step S07).

[0029] Step S07: Real-time online prediction of belt scale cumulative volume. The slitting section starts batch production, including X1-X4 and X6-X 11 All of them are known values. X1-X4 and X6-X8 can be obtained based on the historical reporting data of the batch in the slicing section. 11 The weather and time related values ​​are as follows: As long as the subsequent production of the slitting section is not stopped, X9-X 11 It can also be roughly estimated. The X5 variable is the moisture value of the leaf before cutting, which cannot be directly obtained through historical data, nor can it be estimated through the nature of continuous production. Therefore, the real-time online prediction of the belt scale accumulation needs to dynamically consider the X5 variable as a changing value. The processing method is as follows: (1) When the shredding process is started, the moisture detection instrument at the entrance of the pre-cutting moistening leaves cannot detect the material, so the process standard moisture value at the entrance of the pre-cutting moistening leaves is substituted into the X5 variable; (2) When the moisture detection instrument at the entrance of the leaf moistening before cutting detects the material and the value is stable, the data is read at the specified interval and stored in the data stack. Finally, the mean of all the data in the data stack is obtained and substituted into the X5 variable as the arithmetic mean.

[0030] Since the X5 variable in the real-time online prediction model changes dynamically, Y also changes with the production of the slitting section, but its changing trend tends to the actual value of the accumulated amount of the belt scale. 11 The variables are input into the linear regression model for online prediction of tobacco leaf accumulation, and Y is calculated, which is the real-time online prediction of belt scale accumulation, denoted as Although the real-time online prediction of material accumulation cannot be directly accurate, the dynamic prediction value can be provided as an intermediate variable to the machine learning prediction algorithm in other cigarette making processes, providing important process intermediate values ​​for various other artificial intelligence algorithms.

[0031] Step S08, offline prediction of belt scale cumulative amount. After the batch production of the shredding section is completed, the variable X5 changes from dynamic to static, and the unique mean can be calculated based on the moisture data of the leaf inlet before cutting of the entire batch. 11 The variable is input into the offline prediction of tobacco leaf accumulation linear regression model, and Y is calculated, which is the offline prediction of belt scale accumulation, denoted as .

[0032] Step S09: Calculation and early warning of material cumulative weight error. Real-time online prediction of belt scale cumulative weight Offline prediction of belt scale accumulation All are based on historical data fitting and are close to the actual material cumulative weight The actual cumulative weight measured by the belt scale is the cumulative value of online measurement, and there are also certain measurement errors and system errors. Assuming that the cumulative amount measured by the belt scale is , there is a certain deviation between the above four related Y. Since the actual material cumulative weight It is impossible to know forever, but based on the knowledge of probability and logic, we can conclude that , , At least one value is very close Therefore, based on the high probability event rate close to 100%, the following error calculation can be done: Real-time online prediction of belt scale accumulation For multiple data, record at interval T , Convert to time series array . Set the array elements Sort from small to large, the median is .

[0033] If the belt scale measures the cumulative amount Real-time online prediction of the median of belt scale cumulative volume If the deviation is greater than the set threshold 1, it is considered and There is one with A deviation occurs, so the production system gives an early warning.

[0034] The expression formula is: (1) If the belt scale measures the cumulative amount Offline prediction of belt scale accumulation If the deviation is greater than the set threshold 2, it is considered and There is one with A deviation occurs, so the production system gives an early warning.

[0035] (2) If the median of the belt scale cumulative volume is predicted online in real time Offline prediction of belt scale accumulation If the deviation is greater than the set threshold 3, it is considered and There is one with A deviation occurs, so the production system gives an early warning.

[0036] (3) In formula 1-3, , , The thresholds 1, 2 and 3 set for the method of the present invention are set manually and can be modified at any time.

[0037] When the production system gives an early warning, it is determined by humans or other artificial intelligence algorithms that there is a deviation in the belt scale or the online linear regression model for predicting the cumulative amount of tobacco leaves or the offline linear regression model for predicting the cumulative amount of tobacco leaves; specifically, as follows: First, check whether the belt scale is abnormally invalid; if the belt scale is not invalid after verification, compare the difference between the median of the real-time online predicted belt scale cumulative amount, the offline predicted belt scale cumulative amount and the belt scale measured cumulative amount; as follows: If the difference between the median of the real-time online predicted belt scale cumulative amount and the belt scale measured cumulative amount is greater than the difference between the offline predicted belt scale cumulative amount and the belt scale measured cumulative amount, it is determined that the linear regression model for online prediction of tobacco leaf cumulative amount has a deviation; If the difference between the offline predicted belt scale cumulative amount and the belt scale measured cumulative amount is greater than the difference between the median of the real-time online predicted belt scale cumulative amount and the belt scale measured cumulative amount, it is determined that the offline predicted tobacco leaf cumulative amount linear regression model has a deviation.

[0038] Step S10, marking results and retraining. If it is determined through the above method that the online prediction of tobacco leaf accumulation linear regression model or the offline prediction of tobacco leaf accumulation linear regression model has a prediction deviation, the batch data is calibrated and re-imported into the online prediction of tobacco leaf accumulation linear regression model or the offline prediction of tobacco leaf accumulation linear regression model for retraining to improve the model prediction accuracy. The specific method is described as follows: If there is a deviation in the linear regression model for online prediction of tobacco leaf accumulation, the offline prediction of belt scale accumulation will be and belt scale to measure cumulative volume The mean of is used as Y to import into the online prediction of tobacco leaf accumulation linear regression model for retraining. The formula is: (4) If there is a deviation in the linear regression model for offline prediction of tobacco leaf accumulation, the median of the belt scale accumulation will be predicted online in real time. and belt scale to measure cumulative volume The mean of is used as Y to import the offline prediction of tobacco leaf accumulation linear regression model for retraining. The formula is: (5) In formula 4-5, The output Y is imported as the online linear regression model for predicting tobacco leaf accumulation or the offline linear regression model for predicting tobacco leaf accumulation, and the model training is performed again. Example 2

[0039] The present embodiment provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the steps of the method for predicting the cumulative weight of tobacco leaves in a shredding section based on machine learning described in Example 1 are implemented. Example 3

[0040] This embodiment provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the steps of the method for predicting the cumulative weight of tobacco leaves in the shredding stage based on machine learning described in Embodiment 1 are implemented.

[0041] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0042] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0043] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1A function specified in one or more boxes.

[0044] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0045] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for predicting the cumulative weight of tobacco leaves in the shredding section based on machine learning, characterized in that: include: When the shredding section is in batch production, factors related to tobacco leaf accumulation, climate-related factors and production time period variables are input into the trained online prediction of tobacco leaf accumulation linear regression model, and the online prediction of tobacco leaf accumulation linear regression model outputs real-time online prediction of belt scale accumulation; When the batch production of the shredding section is completed, the factors related to the tobacco leaf accumulation, climate-related factors and production time period variables are input into the trained offline prediction of tobacco leaf accumulation linear regression model, and the offline prediction of tobacco leaf accumulation linear regression model outputs the offline prediction of belt scale accumulation.

2. The method for predicting the cumulative weight of tobacco leaves in the shredding section based on machine learning according to claim 1 is characterized in that: Factors related to the cumulative amount of tobacco leaves include the cumulative amount of water added for loose moistening, the cumulative amount of weighing before loose moistening, the outlet temperature of loose moistening leaves, the outlet moisture of loose moistening leaves, the inlet moisture of leaves before cutting, the storage time of the pre-distribution cabinet, the storage time of the distribution cabinet and the cabinet change time from the pre-distribution cabinet to the distribution cabinet; climate-related factors include atmospheric temperature and atmospheric humidity; the production time period variable is the production time period.

3. The method for predicting the cumulative weight of tobacco leaves in the shredding section based on machine learning according to claim 2 is characterized in that: The training of the online prediction of tobacco leaf accumulation linear regression model and the offline prediction of tobacco leaf accumulation linear regression model includes: According to the process flow chart of the slicing section and the shredding section of the silk thread making line, several groups of factors related to the accumulation of tobacco leaves were identified; Taking tobacco silk production batches as units, factors related to the cumulative amount of tobacco leaves and historical data of the cumulative amount of the corresponding weighing before cutting and moistening are derived by batch; Process the exported historical data to remove outliers; Import climate-related factors and production time period variables, use Pearson formula to conduct correlation analysis on historical data of monthly climate-related factors, production time period variables, factors related to tobacco leaf accumulation after removing discrete points, and corresponding accumulation before cutting and moistening. Variables with correlation coefficients greater than set values ​​with corresponding accumulation before cutting and moistening after removing discrete points in monthly climate-related factors, production time period variables, and factors related to tobacco leaf accumulation after removing discrete points are used as inputs of monthly linear regression models to be trained, and corresponding accumulation before cutting and moistening after removing discrete points is used as outputs of monthly linear regression models to be trained. Generate online prediction tobacco leaf accumulation linear regression models and offline prediction tobacco leaf accumulation linear regression models corresponding to monthly training according to months. Among them, the input quantity for online prediction of tobacco leaf cumulative quantity linear regression model training is the process standard moisture content of the leaf moistening inlet before cutting when the batch of shredding section is started and the historical data of the moisture content of the leaf moistening inlet detected by the moisture content detection instrument of the leaf moistening inlet during the batch production of the shredding section; the input quantity for offline prediction of tobacco leaf cumulative quantity linear regression model training is the historical data average of the moisture content of the leaf moistening inlet before cutting for the entire batch after the batch production of the shredding section is completed.

4. The method for predicting the cumulative weight of tobacco leaves in the shredding section based on machine learning according to claim 2 is characterized in that: When the shredding section starts batch production, the factors related to tobacco leaf accumulation, climate-related factors and production time period variables are input into the trained online prediction of tobacco leaf accumulation linear regression model, including: When the shredding section is in batch production, climate-related factors, production time period variables, and factors related to tobacco leaf accumulation except for the moisture in the leaf before cutting are input into the trained online prediction linear regression model of tobacco leaf accumulation; When the shredding process starts, the standard moisture content of the pre-cut leaf moistening entrance is input into the trained online prediction model for the cumulative amount of tobacco leaves. When the shredding process is in progress, multiple groups of moisture content of the pre-cut leaf moistening entrance detected stably by the moisture detection instrument are read at intervals, and the mean of the multiple groups of moisture content of the pre-cut leaf moistening entrance is used as the arithmetic mean and input into the trained online prediction model for the cumulative amount of tobacco leaves.

5. The method for predicting the cumulative weight of tobacco leaves in the shredding section based on machine learning according to claim 2 is characterized in that: When the batch production of the shredding section is completed, the factors related to the tobacco leaf accumulation, climate-related factors and production time period variables are input into the trained offline prediction of tobacco leaf accumulation linear regression model, including: When the batch production of the shredding section is completed, climate-related factors, production time period variables and factors related to the tobacco leaf accumulation amount except the moisture in the leaf before cutting are input into the trained offline prediction of tobacco leaf accumulation amount linear regression model; When the batch production of the shredding section is completed, the historical data mean of the moisture content of the leaves before cutting in the entire batch is input into the trained offline prediction linear regression model for the cumulative amount of tobacco leaves.

6. The method for predicting the cumulative weight of tobacco leaves in the shredding section based on machine learning according to claim 1, characterized in that: It also includes: analyzing the real-time online predicted belt scale cumulative amount, the offline predicted belt scale cumulative amount and the belt scale measured cumulative amount, and determining whether to issue an early warning based on the analysis results, including: The real-time online predicted belt scale cumulative amount recorded at intervals is converted into a time series array, and the elements in the time series array are sorted from small to large to obtain the median of the real-time online predicted belt scale cumulative amount; If the deviation between the cumulative amount measured by the belt scale and the median of the real-time online predicted cumulative amount of the belt scale is greater than the first set threshold, an early warning is issued; If the deviation between the accumulated amount measured by the belt scale and the accumulated amount predicted by the offline belt scale is greater than the second set threshold, an early warning is issued; If the deviation between the real-time online predicted belt scale cumulative median and the offline predicted belt scale cumulative median is greater than the third set threshold, an early warning is issued.

7. The method for predicting the cumulative weight of tobacco leaves in the shredding section based on machine learning according to claim 6 is characterized in that: Also includes: When an early warning is issued, first check whether the belt scale is abnormal or not; If the belt scale calibration is not invalid, then compare the difference between the median of the real-time online predicted belt scale cumulative amount, the offline predicted belt scale cumulative amount and the belt scale measured cumulative amount; If the difference between the median of the real-time online predicted belt scale cumulative amount and the belt scale measured cumulative amount is greater than the difference between the offline predicted belt scale cumulative amount and the belt scale measured cumulative amount, it is determined that the linear regression model for online prediction of tobacco leaf cumulative amount has a deviation; If the difference between the offline predicted belt scale cumulative amount and the belt scale measured cumulative amount is greater than the difference between the median of the real-time online predicted belt scale cumulative amount and the belt scale measured cumulative amount, it is determined that the offline predicted tobacco leaf cumulative amount linear regression model has a deviation.

8. The method for predicting the cumulative weight of tobacco leaves in the shredding section based on machine learning according to claim 7 is characterized in that: Also includes: If there is a deviation in the linear regression model for online prediction of tobacco leaf accumulation, the mean of the offline predicted belt scale accumulation and the belt scale measured accumulation is imported into the linear regression model for online prediction of tobacco leaf accumulation for retraining; If there is a deviation in the offline prediction of tobacco leaf accumulation linear regression model, the median of the real-time online prediction of the belt scale accumulation and the mean of the belt scale measurement accumulation are imported into the offline prediction of tobacco leaf accumulation linear regression model for retraining.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps of the method for predicting the cumulative weight of tobacco leaves in the shredding section based on machine learning as described in any one of claims 1 to 8 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for predicting the cumulative weight of tobacco leaves in the shredding stage based on machine learning as described in any one of claims 1 to 8 are implemented.

Images