Tobacco-making weighing belt scale state prediction method and device, electronic terminal and storage medium

Abstract

The invention discloses a tobacco shred making weighing belt scale state prediction method and device, an electronic terminal and a storage medium in the field of tobacco shred making, and the method comprises the steps: obtaining a tobacco shred making process flow diagram, marking factors related to the material weight, inputting real-time production data into a plurality of pre-constructed belt scale material weight prediction models, and carrying out the prediction of the material weight of the belt scale. A predicted value of the accumulated weight of the materials of the weighing belt scale is obtained, an actual value of the accumulated weight of the materials is recorded, and a data form is formed; comparing the data form with preset over-limit threshold group data, and executing an abnormality judgment operation to obtain a belt weigher state prediction result after production is finished; and inputting the belt weigher state prediction result, the predicted value of the accumulated weight of the materials of the weighing belt weigher in each batch, the factors related to the material weight and the material mark into a pre-constructed support vector machine model so as to predict the belt weigher state of the materials in production in real time. Real-time monitoring and early warning can be provided for the production process of the cut tobacco production line, and the working efficiency is improved.

Description

Technical Field

[0001] The invention relates to a method, a device, an electronic terminal and a storage medium for predicting the state of a silk-making weighing belt scale, and belongs to the technical field of tobacco silk-making. Background Art

[0002] In the process of making cigarettes, from the delivery of raw tobacco leaves to the delivery of finished tobacco leaves to the cabinet, many key processes are required. This process is not only complicated and has a long process route, but also has multiple links that may cause material loss. In order to improve the competitiveness of the factory, it is very important to achieve lean production and reduce costs. Especially in the process of making cigarettes, if the key quality indicator - the yield rate - decreases, and the factory cannot accurately control each process and find the loss point in time, it will lead to huge waste and loss.

[0003] In order to realize the intelligent tobacco production line, online real-time monitoring of the cumulative weight deviation of tobacco leaves in the tobacco production line is an important link. This needs to be achieved by comparing the predicted cumulative weight of tobacco leaves with the actual cumulative weight. To do this, the premise is to have accurate belt weighing, because any abnormality of the weighing belt scale (such as unpeeled, tare fluctuation or sensor abnormality) will directly affect the accuracy of material monitoring and prediction. Therefore, ensuring the stability and accuracy of the weighing belt scale is the basis for realizing online material monitoring and material balance. Summary of the invention

[0004] The purpose of the present invention is to provide a silk weighing belt scale state prediction method, device, electronic terminal and storage medium, which can accurately provide classification prediction results to ensure the reliability of online material monitoring and material balance, and improve production efficiency and quality.

[0005] In order to solve the above technical problems, the present invention is implemented by adopting the following technical solutions.

[0006] On the one hand, the present invention provides a method for predicting the state of a silk weighing belt scale, comprising: Obtaining a silk thread production process flow chart, in which factors related to material weight are identified; Input the real-time production data including factors related to the material weight into the pre-built multiple belt scale material weight prediction models, obtain the predicted value of the cumulative weight of each batch of weighing belt scale materials, and record the actual value of the cumulative weight of each batch of weighing belt scale materials, and form a data form based on the predicted value and the actual value; The data form is compared with the preset over-limit threshold group data, and a difference judgment operation is performed to obtain the belt scale state prediction result of all batches of material data after the production is completed; The belt scale state prediction result, the predicted value of the cumulative weight of each batch of weighing belt scale materials, factors related to material weight and material brand are input into the pre-built support vector machine model to predict the belt scale state of material data in production in real time.

[0007] In combination with the first aspect, further, the factors related to the material weight include: The first relevant factor before the weighing belt scale, marked as variable group ; The factors related to the first weighing belt scale after the first weighing belt scale and before the second weighing belt scale are labeled as variable groups ; And so on, until the relevant factors after the f-1th belt scale and before the fth belt scale are marked as the variable group ; Among them, the variable group includes each weighing belt scale, material moisture measurement point, temperature measurement point, tobacco material storage cabinet time, temperature and humidity.

[0008] In combination with the first aspect, further, obtaining the predicted value of the cumulative weight of each batch of weighing belt scale materials includes: Using the gradient boosting regression tree algorithm to establish a prediction model for the cumulative weight of multiple weighing belt scale materials , … ; Export multiple batches of material data from the historical database and add material brands to the material data; The material grade, variable group Input the first belt scale material cumulative weight prediction model Get the first belt scale cumulative weight prediction value ; Material brand, variable group , the first belt scale cumulative weight prediction value , variable group Input the second belt scale material cumulative weight prediction model The second belt scale cumulative weight prediction value is obtained ; And so on, until the material brand, variable group , the first belt scale cumulative weight prediction value , variable group , The second belt scale cumulative weight prediction value … The variable group is used as the input to the f-th belt scale material cumulative weight prediction model , to get the predicted value of the accumulated weight of the fth belt scale .

[0009] In combination with the first aspect, further, forming a data form includes: The predicted value and actual value of the cumulative weight of each batch of materials on the weighing belt scale are recorded in the data form of the corresponding batch respectively, and the data of the preset batch number closest in time is stored in the form of a first-in-first-out data structure.

[0010] In combination with the first aspect, further, the performing of the difference determination operation includes: (1) Single batch out-of-limit data difference judgment Set a single batch of over-limit threshold group data , the single batch of over-limit threshold group data Store in the two-dimensional array A1; The absolute value of the difference between the predicted value and the actual value of the cumulative weight of each batch of weighing belt scale materials in the data form is compared with the single batch over-limit threshold group data set in the corresponding two-dimensional array A1. If the absolute value of the difference between the predicted value and the actual value is greater than the single batch exceeding limit threshold group data set in the corresponding two-dimensional array A1, , the weighing belt scale of the current batch is in an abnormal state, otherwise it is in a non-abnormal state.

[0011] (2) Determine the difference between the accumulated values ​​of multiple batches of over-limit data Set multiple batches of continuous exceeding threshold group data and the number of consecutive batches k, the consecutive batches of over-limit threshold group data Store it in another two-dimensional array A2; The accumulated value of the absolute value of the difference between the predicted value and the actual value of the cumulative weight of the weighing belt scale materials for consecutive k batches in the data form is compared with the multiple batches of over-limit threshold group data set in the corresponding two-dimensional array A2. If the accumulated absolute value of the difference between the predicted value and the actual value is greater than the continuous batch of over-limit threshold group data set in the corresponding two-dimensional array A2, , then the weighing belt scale of consecutive k batches is in an abnormal state, otherwise it is in a non-abnormal state.

[0012] (3) Multiple batches of out-of-limit difference sign judgment Set the number of consecutive batches v, and determine whether the difference between the predicted value and the actual value of the cumulative weight of the weighing belt scale material of each batch in the data form is greater than 0 or less than 0. If both are greater than 0 or less than 0, the weighing belt scale of consecutive v batches is in an abnormal state, otherwise it is in a non-abnormal state.

[0013] In combination with the first aspect, further, the expressions for performing the difference determination operation are respectively: (1) Single batch out-of-limit data difference judgment

[0014] in, Represents the predicted value of the mth belt scale and the uth variable group in the data form; Indicates the actual value of the mth belt scale and the uth variable group in the data form; Represents a single batch of over-limit threshold group data; (2) Determine the difference between the accumulated values ​​of multiple batches of over-limit data

[0015] Where n represents the number of batches; k represents the maximum number of consecutive batches; Indicates the predicted value of the bth belt scale and the cth variable group in the data form; Indicates the actual value of the bth belt scale and the cth variable group in the data form; Indicates multiple batches of continuous exceeding threshold group data; (3) Multiple batches of out-of-limit difference sign judgment U

[0016] Wherein, v represents batch; Represents the predicted value of the sth belt scale and the rth variable group in the data form; Represents the actual value of the sth belt scale and the rth variable group in the data form.

[0017] In combination with the first aspect, further, after performing the difference determination operation, the method further includes: If the single batch over-limit data difference, the cumulative value of the multi-batch over-limit data difference and the multi-batch over-limit difference symbol are not judged to be different, and the belt scale equipment and the work report data are also in normal state, then the data of the corresponding batch of weighing belt scale in non-abnormal state is imported into the belt scale material cumulative weight prediction model as the latest data , … To update and modify the existing belt scale material cumulative weight prediction model , … .

[0018] In a second aspect, a state prediction device for a silk weighing belt scale comprises: A factor identification module, used to obtain a silk thread production process flow chart, and identify factors related to material weight in the silk thread production process flow chart; The cumulative weight prediction module is used to input the real-time production data containing factors related to the material weight into the pre-built multiple belt scale material weight prediction models, obtain the predicted value of the cumulative weight of each batch of weighing belt scale materials, and record the actual value of the cumulative weight of each batch of weighing belt scale materials, and form a data form based on the predicted value and the actual value; The production end abnormality judgment module is used to compare the data form with the preset over-limit threshold group data, and perform abnormality judgment operations to obtain the belt scale status prediction results of all batches of material data after the production is completed; The in-production prediction module is used to input the belt scale status prediction results, the predicted value of the cumulative weight of each batch of weighing belt scale materials, factors related to material weight, and material brand into a pre-built support vector machine model to predict the belt scale status of material data in production in real time.

[0019] In a third aspect, an electronic terminal is provided, characterized in that it comprises a processor and a memory connected to the processor, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, the steps of the method described in the first aspect are performed.

[0020] In a fourth aspect, a computer-readable storage medium stores a computer program thereon, wherein the program, when executed by a processor, implements the steps of the method described in the first aspect.

[0021] Compared with the prior art, the present invention has the following beneficial effects: The present invention obtains the silk thread process flow chart and identifies the factors related to the material weight. The method can accurately capture the key variables that affect the material weight and provide an accurate data basis for subsequent prediction and monitoring. The real-time production data of these key factors are input into the pre-built multiple belt scale material weight prediction models, which can not only obtain the predicted value of the cumulative weight of each batch of materials, but also record the actual value, thereby forming a detailed data form. This step helps to compare the predicted value and the actual value, evaluate the accuracy of the prediction model, and provide data support for subsequent analysis. By comparing the data form with the preset over-limit threshold group data and performing the judgment operation, the method can provide the belt scale state prediction results of all batches of material data after the production is completed, and timely discover potential abnormal states, thereby improving the monitoring and control capabilities of the production process. Finally, the prediction results, predicted values, related factors and material grades are input into the pre-built support vector machine model. The method can predict the belt scale state of material data in production in real time, provide real-time monitoring and early warning for the production process of silk thread, enhance the transparency and responsiveness of the production process, help optimize production efficiency and product quality, and ensure production safety and stability. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 It is a flow chart of a method for predicting the state of a silk weighing belt scale provided by an embodiment of the present invention; FIG2 is a schematic diagram showing the positions of multiple weighing belt scales in a silk making process flow chart provided by an embodiment of the present invention; FIG. 3 is a schematic diagram of a prediction model for the cumulative weight of materials on a single belt scale provided by an embodiment of the present invention. FIG. 4 is a schematic diagram of a correlation matrix provided by an embodiment of the present invention FIG5 is a schematic diagram of a data form provided by an embodiment of the present invention; Figure 6 It is a schematic diagram of predicting abnormal or non-abnormal state of a belt scale using the SVM model provided by an embodiment of the present invention; Figure 7 It is a schematic diagram of predicting the status of a belt scale during and after production provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0023] The technical solution of the present invention is described in detail below through the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present invention and the specific features in the embodiments are detailed descriptions of the technical solution of the present invention, rather than limitations on the technical solution of the present invention. The embodiments of the present invention and the technical features in the embodiments may be combined with each other unless there is a conflict.

[0024] The term "and / or" is only a description of the association relationship between related objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " generally indicates that the related objects are in an "or" relationship. Example 1

[0025] See also Figure 1 The present invention introduces a method for predicting the state of a silk weighing belt scale, comprising: Step S1, obtaining a silk thread production process flow chart, and identifying factors related to material weight in the silk thread production process flow chart; Here, in order to visualize each link in the entire silk making process, the silk thread making process flow chart shows the entire processing process from silk thread making, through various steps, to the output of finished products, and in the flow chart, factors related to the material weight need to be specially identified.

[0026] Step S2, inputting the real-time production data including the factors related to the material weight into the pre-built multiple belt scale material weight prediction models, obtaining the predicted value of the cumulative weight of each batch of weighing belt scale materials, and recording the actual value of the cumulative weight of each batch of weighing belt scale materials, and forming a data form according to the predicted value and the actual value; Here, data forms provide basic data for subsequent analysis and prediction.

[0027] Step S3, comparing the data form with the preset over-limit threshold group data, and performing a difference judgment operation to obtain the belt scale state prediction result of all batches of material data after the production is completed; Step S4, input the belt scale status prediction result, the predicted value of the cumulative weight of each batch of weighing belt scale materials, factors related to material weight and material brand into the pre-built support vector machine model to predict the belt scale status of material data in production in real time.

[0028] Through the above steps, the present invention can realize accurate prediction of the state of the silk weighing belt scale, thereby improving the monitoring efficiency and accuracy of the silk line and ensuring the stability of the production process and the product quality. Example 2

[0029] This embodiment further describes in detail the application of the method for predicting the state of a silk weighing belt scale, including the following steps: Step S1, obtain a silk thread production process flow chart, and identify factors related to material weight in the silk thread production process flow chart, the factors mainly include each weighing belt scale, material moisture measurement point, temperature measurement point, tobacco material storage cabinet time, temperature and humidity.

[0030] See also Figure 2 In the embodiment of the present invention, it is assumed that there are f material weighing belt scales (f 5) In the silk-making process, there are multiple factors that affect the cumulative amount of materials. These factors are identified and grouped according to their positions in the silk-making line. The specific grouping is as follows: Mark the relevant factors before the first weighing belt scale as variable group ; Label the related factors after the first weighing belt scale and before the second heavy belt scale as variable group ; And so on, until the relevant factors after the f-1th belt scale and before the fth belt scale are marked as the variable group , then the grouping ends.

[0031] By grouping, it is possible to more systematically analyze the impact of each stage on the cumulative amount of material, and incorporate these factors into the machine learning model to predict and monitor the changes in material weight during the silk making process.

[0032] Among them, for the variable group , which includes the cumulative amount of water added to loose leaves , the cumulative value of the belt scale before loosening and moistening the leaves , Loose leaf outlet temperature value , Loose and moist leaves outlet moisture value , Moisture value of leaf inlet before cutting , Locker time , Cabinet change time , atmospheric temperature , atmospheric humidity Other factors may be deleted or added based on on-site conditions.

[0033] For variable groups , which includes the cumulative value of adding materials and water at one time , The accumulated value of the belt scale before adding material , one-time feeding outlet temperature value , Moisture value at the outlet of the first feed , Moisture value at the primary feed inlet , Locker time , the time to change cabinets from pre-distribution cabinets to distribution cabinets , atmospheric temperature , atmospheric humidity Other factors may be deleted or added based on on-site conditions.

[0034] Similarly, the variable group , variable group …variable group The factor selection method of is consistent with the first two groups, and other factors can be deleted or added based on the on-site situation.

[0035] Furthermore, on the silk production line, multiple weighing belt scales are arranged in sequence and defined as , , … , which correspond to the cumulative weight values ​​of the materials measured by the first, second, third…fth weighing belt scale respectively.

[0036] It should be noted that , , … And variable groups , variable group , variable group …variable group All have been obtained and are the only actual values.

[0037] Step S2, inputting the real-time production data including the factors related to the material weight into the pre-built multiple belt scale material weight prediction models, obtaining the predicted value of the cumulative weight of each batch of weighing belt scale materials, and recording the actual value of the cumulative weight of each batch of weighing belt scale materials, and forming a data form according to the predicted value and the actual value; Specifically, obtaining the predicted value of the cumulative weight of each batch of weighing belt scale materials includes the following steps: Step S21: Establish a prediction model for the cumulative weight of multiple weighing belt scale materials using a gradient boosting regression tree algorithm , … ; Step S22, exporting multiple batches of material data from the historical database, and adding material brands to the material data; See also Figure 3 In the embodiment of the present invention, at least 2000 batches of data are exported from the historical database in units of one production batch, and a new input variable, namely the material brand, is added to the data set. In addition, production failure batch data and experimental production batch data are removed from the collected data. These data may contain outliers, which are not conducive to model learning. Batches with outliers in the data are deleted to prevent the problem of poor prediction accuracy during the model fitting process.

[0038] Step S23: the material grade, variable group Input the first belt scale material cumulative weight prediction model , get the first belt scale cumulative weight prediction value ; Step S24: Material brand, variable group , the first belt scale cumulative weight prediction value , variable group Input the second belt scale material cumulative weight prediction model , get the second belt scale cumulative weight prediction value ; Step S25: According to this logic, fit is performed until the material brand and variable group , the first belt scale cumulative weight prediction value , variable group , The second belt scale cumulative weight prediction value … The variable group is used as the input end and put into the f-th belt scale material cumulative weight prediction model To obtain the predicted value of the accumulated weight of the fth belt scale .

[0039] In summary, each model uses the cumulative weight of the previous belt scale as one of the input variables, which can capture the continuous changes in the material weight during the silk making process. In this way, each model can predict the cumulative weight of the current belt scale based on the prediction results of the previous belt scale and all relevant factors of the current belt scale.

[0040] In addition, for the gradient boosted regression tree algorithm (GBRT), assuming that the h-th belt scale (h f) Fitting prediction model, the input of the model can ignore the material brand and the accumulated weight of the belt scale (j=1,2,3…f), only analyze variable groups , variable group , …, variable group The variable group in , and the accumulated weight of the belt scale The correlation of i=1,2,3,…,f-1; u=1,2,3,…,N See also Figure 4 , by drawing the correlation matrix, calculate the cumulative weight of the belt scale With variable group In the embodiment of the present invention, first, consider the cumulative weight of the belt scale Variable groups with correlation coefficients greater than 0.5 , remove and belt scale cumulative weight Variable groups with correlation coefficients less than 0.5 , thereby filtering out the target variable belt scale cumulative weight Groups of variables that have a strong linear relationship.

[0041] Secondly, we further eliminated variables with correlation coefficients between variable groups greater than 0.74 in order to reduce multicollinearity between variables and avoid overfitting.

[0042] For example: If the variable group , variable group , variable group Accumulated weight with belt scale The correlation coefficient is greater than 0.5, but the variable group , variable group , variable group The correlation coefficient between the two variables is greater than 0.75, indicating that there is a high correlation between these variable groups. If all of them are fitted into the prediction model, the model will be overfitted. Therefore, only one of the three variable groups can be retained (select the variable that is consistent with the cumulative weight of the belt scale). The variable group with the largest correlation coefficient).

[0043] By using the above method of drawing a correlation matrix, variables that have a strong correlation with the prediction target but are not highly correlated with each other are selected to build a prediction model that is both accurate and not overfitted.

[0044] Furthermore, the gradient boosting regression tree parameters are set. For the first belt scale material cumulative weight prediction model ,set up The number of model trees is 10, and the learning rate is assigned to 0.1; for the second belt scale material cumulative weight prediction model , set the number of model trees to 10+2=12, and assign the learning rate to 01.-0.005=0.0095; ...; for the f-th belt scale material cumulative weight prediction model , set the number of model trees to 10+2*f, the learning rate to 0.1-0.005*f, and the minimum learning rate to 0.05. If the calculated learning rate is less than 0.05, the learning rate is set to the minimum value of 0.05.

[0045] Furthermore, the loss function of the gradient boosting regression tree was set to least squares regression, the Friedman mean square error was set as the indicator for evaluating the splitting quality, the minimum split sample lower limit was 2, the minimum number of samples contained in the leaf node was 1, the minimum node weight coefficient was 0, and the maximum depth was 4 to prevent the model from being too complicated. The model was fitted according to 70% of the data, and the remaining 30% of the data was used to verify the fitting accuracy of the prediction model.

[0046] See also Figure 5 , input the real-time generated data into the fitted multiple belt scale material weight prediction models, obtain the predicted value of the cumulative weight of each batch of weighing belt scale materials, and record the predicted value of the cumulative weight of each batch of weighing belt scale materials, count the predicted value and actual value of each batch in the corresponding data form, and record multiple groups of data in a stack, in the form of a first-in-first-out data structure, and store the data of the preset batch number closest in time. The embodiment of the present invention only saves the data of the most recent 10 batches for analysis and monitoring.

[0047] Step S3, comparing the data form with the preset over-limit threshold group data, and performing a difference judgment operation to obtain the belt scale state prediction result of all batches of material data after the production is completed; Specifically, the three methods of judging abnormalities after production are as follows: Step S31: single batch over-limit data difference judgment Set a single batch of over-limit threshold group data , the single batch of over-limit threshold group data Stored in the two-dimensional array A1, the expression of A1 is as follows:

[0048] Where A1 represents a two-dimensional array; Indicates the threshold value of the first variable group of the first belt scale; Indicates the threshold value of the second variable group of the first belt scale; Indicates the threshold value of the third variable group of the first belt scale; Indicates the threshold value of the Nth variable group of the first belt scale; Indicates the threshold value of the first variable group of the second belt scale; Indicates the threshold value of the second variable group of the second belt scale; Indicates the threshold value of the second variable group of the second belt scale; Indicates the threshold value of the second variable group of the second belt scale; It represents the threshold value of the third variable group of the second belt scale; Indicates the threshold value of the Nth variable group of the second belt scale; It represents the threshold value of the first variable group of the third belt scale; It represents the threshold value of the second variable group of the third belt scale; It represents the threshold value of the third variable group of the third belt scale; Indicates the threshold value of the Nth variable group of the third belt scale; represents the threshold value of the first variable group of the f-1th belt scale; represents the threshold value of the second variable group of the f-1th belt scale; represents the threshold value of the third variable group of the f-1th belt scale; Represents the threshold value of the Nth variable group of the f-1th belt scale.

[0049] Compare the absolute value of the difference between the predicted value and the actual value of the cumulative weight of each batch of weighing belt scale materials in the data form with the single batch over-limit threshold group data set in the corresponding two-dimensional array A1 If the absolute value of the difference between the predicted value and the actual value is greater than the single batch exceeding limit threshold group data set in the corresponding two-dimensional array A1, , the weighing belt scale of the current batch is in an abnormal state, otherwise it is in a non-abnormal state. The expression for judging the difference of the over-limit data of the single batch is:

[0050] in, Represents the predicted value of the mth belt scale and the uth variable group in the data form; Indicates the actual value of the mth belt scale and the uth variable group in the data form; Represents a single batch of over-limit threshold group data.

[0051] Step S32: Multiple batches of over-limit data difference accumulation value judgment Set multiple batches of continuous exceeding threshold group data And the number of consecutive batches k, multiple consecutive batches of over-limit threshold group data Stored in another two-dimensional array A2, the expression of A2 is as follows:

[0052] Where A2 represents another two-dimensional array; Indicates the multi-batch threshold of the first variable group of the first belt scale; Indicates the multi-batch threshold of the second variable group of the first belt scale; Indicates the multi-batch threshold of the third variable group of the first belt scale; Indicates the multi-batch threshold of the Nth variable group of the first belt scale; Indicates the multi-batch threshold of the first variable group of the second belt scale; Indicates the multi-batch threshold of the second variable group of the second belt scale; Indicates the multi-batch threshold of the second variable group of the second belt scale; Indicates the multi-batch threshold of the second variable group of the second belt scale; It represents the multi-batch threshold value of the third variable group of the second belt scale; Indicates the multi-batch threshold value of the Nth variable group of the second belt scale; It represents the multi-batch threshold value of the first variable group of the third belt scale; It represents the multi-batch threshold value of the second variable group of the third belt scale; It represents the multi-batch threshold value of the third variable group of the third belt scale; Indicates the multi-batch threshold of the Nth variable group of the third belt scale; It represents the multi-batch threshold of the first variable group of the f-1th belt scale; It represents the multi-batch threshold of the second variable group of the f-1th belt scale; It represents the multi-batch threshold value of the third variable group of the f-1th belt scale; Represents the multi-batch threshold of the Nth variable group of the f-1th belt scale.

[0053] The accumulated value of the absolute value of the difference between the predicted value and the actual value of the cumulative weight of the weighing belt scale materials for consecutive k batches in the data form is compared with the single batch over-limit threshold group data set in the corresponding two-dimensional array A2. If the accumulated absolute value of the difference between the predicted value and the actual value is greater than the continuous batch of over-limit threshold group data set in the corresponding two-dimensional array A2, , then the weighing belt scale of consecutive k batches is in an abnormal state, otherwise it is in a non-abnormal state. The abnormal judgment expression of the cumulative value of the difference of the multiple batches of over-limit data is:

[0054] Where n represents the number of batches; k represents the maximum number of consecutive batches; Indicates the predicted value of the bth belt scale and the cth variable group in the data form; Indicates the actual value of the bth belt scale and the cth variable group in the data form; Indicates multiple batches of continuous exceeding threshold group data; Step S33: Multiple batches of over-limit difference sign judgment Set the number of consecutive batches v, and determine whether the difference between the predicted value and the actual value of the cumulative weight of the weighing belt scale materials of each batch in the data form is greater than 0 or less than 0. If both are greater than 0 or less than 0, the weighing belt scale of consecutive v batches is in an abnormal state, otherwise it is in a non-abnormal state. The expression for judging the sign of the multi-batch over-limit difference is as follows: U

[0055] Wherein, v represents batch; Represents the predicted value of the sth belt scale and the rth variable group in the data form; Represents the actual value of the sth belt scale and the rth variable group in the data form.

[0056] For a batch of data, if the single batch over-limit data difference in step S31, the cumulative value of the multiple batches of over-limit data difference in step S32, and the multiple batches of over-limit difference symbols in step S33 are not judged as abnormal, that is, the difference between the predicted value and the actual value is within the acceptable range, and the belt scale equipment and reporting data are also in normal state, then the batch data (the data of the corresponding batch of weighing belt scale in non-abnormal state) is imported into the belt scale material cumulative weight prediction model as the latest data , … To update and modify the existing belt scale material cumulative weight prediction model , … . , thereby improving the accuracy and predictive ability of the prediction model.

[0057] Step S4, input the belt scale state prediction result, the predicted value of the cumulative weight of each batch of weighing belt scale materials, factors related to material weight, and material brand into a pre-built support vector machine model (SVM) to predict the belt scale state of material data in production in real time.

[0058] See also Figure 6 , record all batch data of abnormal and non-abnormal states in step S3, collect at least 1000 records, and use them as historical data of the SVM model for model fitting, so that it can predict the belt scale status of material data in production based on the input data.

[0059] Among them, the input variables of the SVM model are material brand, Y 1 Belt scale cumulative weight prediction value, X 1N The key variable in the variable group (G 1X used in the prediction model 1N variables), Y 2 Belt scale cumulative weight prediction value, X 2N The key variable in the variable group (G 2 X used in the prediction model 2N variable)…Y f Belt scale cumulative weight prediction value, X fN The key variable in the variable group (G f X used in the prediction model fN variable).

[0060] In summary, see Figure 7 , adopt the in-production prediction strategy and the post-production prediction strategy to update and correct the prediction model in real time, that is, if a new production batch is in the production process, the SVM model that has been fitted in step S45 can be used. This model will calculate the real-time abnormal state or real-time non-abnormal state of multiple belt scales based on the cumulative weight prediction value of a single belt scale and the key variables in the variable group; if the new batch is just finished, the three abnormality judgment methods in steps S31 to S33 can be used to generate a belt scale state prediction.

[0061] Finally, the prediction conclusion is output and the abnormal status of the belt scale is displayed.

[0062] During the production process, the SVM model is used for prediction and real-time prediction conclusions are output; after production is completed, the single belt scale material cumulative weight model is used for prediction and offline prediction conclusions are output, that is, the prediction results of the belt scale after the production is completed; finally, the abnormal status of the belt scale obtained from the SVM model and three abnormality judgment methods is displayed and output. Example 3

[0063] A state prediction device for a silk weighing belt scale, comprising: A factor identification module, used to obtain a silk thread production process flow chart, and identify factors related to material weight in the silk thread production process flow chart; The cumulative weight prediction module is used to input the real-time production data containing factors related to the material weight into the pre-built multiple belt scale material weight prediction models, obtain the predicted value of the cumulative weight of each batch of weighing belt scale materials, and record the actual value of the cumulative weight of each batch of weighing belt scale materials, and form a data form based on the predicted value and the actual value; The production end abnormality judgment module is used to compare the data form with the preset over-limit threshold group data, and perform abnormality judgment operations to obtain the belt scale status prediction results of all batches of material data after the production is completed; The in-production prediction module is used to input the belt scale status prediction results, the predicted value of the cumulative weight of each batch of weighing belt scale materials, factors related to material weight, and material brand into a pre-built support vector machine model to predict the belt scale status of material data in production in real time. Example 4

[0064] An electronic terminal, characterized in that it includes a processor and a memory connected to the processor, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, the steps of the method described in Example 1 and Example 2 are performed. Example 5

[0065] A computer-readable storage medium having a computer program stored thereon, characterized in that when the program is executed by a processor, the steps of the method described in Example 1 and Example 2 are implemented.

[0066] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0067] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks 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 produce 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.

[0068] 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 1 A function specified in one or more boxes.

[0069] 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 in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0070] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the enlightenment of the present invention, ordinary technicians in this field can also make many forms without departing from the scope of protection of the purpose of the present invention and the claims, which all fall within the protection of the present invention.

Claims

1. A method for predicting the state of a silk weighing belt scale, characterized in that: include: Obtaining a silk thread production process flow chart, in which factors related to material weight are identified; Input the real-time production data including factors related to the material weight into the pre-built multiple belt scale material weight prediction models, obtain the predicted value of the cumulative weight of each batch of weighing belt scale materials, and record the actual value of the cumulative weight of each batch of weighing belt scale materials, and form a data form based on the predicted value and the actual value; The data form is compared with the preset over-limit threshold group data, and a difference judgment operation is performed to obtain the belt scale state prediction result of all batches of material data after the production is completed; The belt scale state prediction result, the predicted value of the cumulative weight of each batch of weighing belt scale materials, factors related to material weight and material brand are input into the pre-built support vector machine model to predict the belt scale state of material data in production in real time.

2. The weighing belt scale state prediction method according to claim 1, characterized in that: The factors related to material weight include: The first relevant factor before the weighing belt scale, marked as variable group ; The factors related to the first weighing belt scale after the first weighing belt scale and before the second weighing belt scale are labeled as variable groups ; And so on, until the relevant factors after the f-1th belt scale and before the fth belt scale are marked as the variable group ; Among them, the variable group includes each weighing belt scale, material moisture measurement point, temperature measurement point, tobacco material storage cabinet time, temperature and humidity.

3. The weighing belt scale state prediction method according to claim 2, characterized in that: Obtain the predicted value of the cumulative weight of each batch of weighing belt scale materials, including: Using the gradient boosting regression tree algorithm to establish a prediction model for the cumulative weight of multiple weighing belt scale materials , … ; Export multiple batches of material data from the historical database and add material brands to the material data; The material grade, variable group Input the first belt scale material cumulative weight prediction model Get the first belt scale cumulative weight prediction value ; Material brand, variable group , the first belt scale cumulative weight prediction value , variable group Input the second belt scale material cumulative weight prediction model The second belt scale cumulative weight prediction value is obtained ; And so on, until the material brand, variable group , the first belt scale cumulative weight prediction value , variable group , The second belt scale cumulative weight prediction value …variable group As the input, it is input into the f-th belt scale material cumulative weight prediction model , to get the predicted value of the accumulated weight of the fth belt scale .

4. The weighing belt scale state prediction method according to claim 1, characterized in that: The forming of the data form comprises: The predicted value and actual value of the cumulative weight of each batch of materials on the weighing belt scale are recorded in the data form of the corresponding batch respectively, and the data of the preset batch number closest in time is stored in the form of a first-in-first-out data structure.

5. The weighing belt scale state prediction method according to claim 1, characterized in that: The performing of the difference determination operation includes: (1) Single batch out-of-limit data difference judgment Set a single batch of over-limit threshold group data , the single batch of over-limit threshold group data Store in the two-dimensional array A1; The absolute value of the difference between the predicted value and the actual value of the cumulative weight of each batch of weighing belt scale materials in the data form is compared with the single batch over-limit threshold group data set in the corresponding two-dimensional array A1. If the absolute value of the difference between the predicted value and the actual value is greater than the single batch exceeding limit threshold group data set in the corresponding two-dimensional array A1, , then the weighing belt scale of the current batch is in an abnormal state, otherwise it is in a normal state; (2) Determine the difference between the accumulated values ​​of multiple batches of over-limit data Set multiple batches of continuous exceeding threshold group data and the number of consecutive batches k, the consecutive batches of over-limit threshold group data Store it in another two-dimensional array A2; The accumulated value of the absolute value of the difference between the predicted value and the actual value of the cumulative weight of the weighing belt scale materials for consecutive k batches in the data form is compared with the multiple batches of over-limit threshold group data set in the corresponding two-dimensional array A2. If the accumulated absolute value of the difference between the predicted value and the actual value is greater than the continuous batch of over-limit threshold group data set in the corresponding two-dimensional array A2, , then the weighing belt scale of k consecutive batches is in an abnormal state, otherwise it is in a non-abnormal state; (3) Multiple batches of out-of-limit difference sign judgment Set the number of consecutive batches v, and determine whether the difference between the predicted value and the actual value of the cumulative weight of the weighing belt scale material of each batch in the data form is greater than 0 or less than 0. If both are greater than 0 or less than 0, the weighing belt scale of consecutive v batches is in an abnormal state, otherwise it is in a non-abnormal state.

6. The weighing belt scale state prediction method according to claim 5, characterized in that: The expressions for performing the difference determination operation are respectively: (1) Single batch out-of-limit data difference judgment ； in, Represents the predicted value of the mth belt scale and the uth variable group in the data form; Indicates the actual value of the mth belt scale and the uth variable group in the data form; Represents a single batch of over-limit threshold group data; (2) Determine the difference between the accumulated values ​​of multiple batches of over-limit data ； Where n represents the number of batches; k represents the maximum number of consecutive batches; Indicates the predicted value of the bth belt scale and the cth variable group in the data form; Indicates the actual value of the bth belt scale and the cth variable group in the data form; Indicates multiple batches of continuous exceeding threshold group data; (3) Multiple batches of out-of-limit difference sign judgment U ； Wherein, v represents batch; Represents the predicted value of the sth belt scale and the rth variable group in the data form; Represents the actual value of the sth belt scale and the rth variable group in the data form.

7. The weighing belt scale state prediction method according to claim 5, characterized in that: After the difference determination operation is performed, the following steps are included: If the single batch over-limit data difference, the cumulative value of the multi-batch over-limit data difference and the multi-batch over-limit difference symbol are not judged to be different, and the belt scale equipment and the work report data are also in normal state, then the data of the corresponding batch of weighing belt scale in non-abnormal state is imported into the belt scale material cumulative weight prediction model as the latest data , … To update and modify the existing belt scale material cumulative weight prediction model , … .

8. A state prediction device for a silk weighing belt scale, characterized in that: include: A factor identification module, used to obtain a silk thread production process flow chart, and identify factors related to material weight in the silk thread production process flow chart; The cumulative weight prediction module is used to input the real-time production data containing factors related to the material weight into the pre-built multiple belt scale material weight prediction models, obtain the predicted value of the cumulative weight of each batch of weighing belt scale materials, and record the actual value of the cumulative weight of each batch of weighing belt scale materials, and form a data form based on the predicted value and the actual value; The production end abnormality judgment module is used to compare the data form with the preset over-limit threshold group data, and perform abnormality judgment operations to obtain the belt scale status prediction results of all batches of material data after the production is completed; The in-production prediction module is used to input the belt scale status prediction results, the predicted value of the cumulative weight of each batch of weighing belt scale materials, factors related to material weight, and material brand into a pre-built support vector machine model to predict the belt scale status of material data in production in real time.

9. An electronic terminal, characterized in that: The method comprises a processor and a memory connected to the processor, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, the steps of the method according to any one of claims 1 to 7 are performed.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the method described in any one of claims 1 to 7 are implemented.

Images