Method and device for predicting cigarette weight of cigarette making machine

Abstract

The invention provides a cigarette making machine cigarette weight prediction method and device. The method comprises the following steps: acquiring key index data in a cigarette production process; inputting the key index data into a trained neural network model, and obtaining a cigarette weight prediction value output by the neural network model; the neural network model is obtained by training multiple pieces of sample data in advance, and each piece of sample data comprises key index data and a cigarette weight label. According to the cigarette making machine cigarette weight prediction method and device, the neural network model is used for predicting the cigarette weight in advance, the situation that the number of unqualified cigarettes is large due to post detection is avoided, and the production efficiency and the resource utilization rate are improved.

Description

Technical Field

[0001] The invention relates to the technical field of cigarette manufacturing, and in particular to a method and device for predicting the weight of cigarettes in a cigarette making machine. Background Art

[0002] Cigarette weight is one of the core indicators in cigarette manufacturing, and the rationality of its design has a decisive influence on ensuring the physical characteristics and sensory quality of the finished cigarettes. If the cigarette weight is too light, there may be a phenomenon of empty cigarette heads, uneven bamboo joints, and a significant reduction in the resistance to smoking, which will directly affect the consumer's experience. On the contrary, if the cigarette weight is too heavy, the resistance to smoking may become too large, increasing the difficulty of smoking and the production cost.

[0003] Currently, the cigarette industry generally lacks an effective means to predict the trend of cigarette weight changes in advance, and most cigarette factories rely on post-monitoring strategies to manage cigarette weight. This method mainly uses a comprehensive test bench to conduct real-time testing during the cigarette production process, and once it is found that the cigarette weight exceeds the standard range, corresponding adjustments are made.

[0004] However, this post-detection and adjustment method has obvious limitations, because it cannot avoid the losses caused by weight deviations that have already occurred in the production process, including the production of defective products, reduced production efficiency and waste of resources. Summary of the invention

[0005] The present invention provides a method and device for predicting the weight of cigarettes of a cigarette making machine, which are used to solve the technical problems of a large number of unqualified cigarettes, low production efficiency and waste of resources in the prior art.

[0006] The present invention provides a method for predicting cigarette weight of a cigarette making machine, comprising: Obtain key indicator data in the cigarette production process; Inputting the key indicator data into a trained neural network model to obtain a predicted value of cigarette weight output by the neural network model; The neural network model is obtained after being trained in advance with a plurality of sample data, each sample data including key indicator data and a cigarette weight label.

[0007] In some embodiments, the method further comprises: Comprehensively collect relevant indicator data in the cigarette production process, as well as the weight of cigarettes produced under each relevant indicator data; The relevant indicator data are screened to determine the indicator data that has a greater impact on the weight of cigarettes as the key indicator data.

[0008] In some embodiments, the screening of the relevant indicator data to determine the indicator data having a greater impact on the cigarette weight includes: The random forest model is used to screen the relevant indicator data to determine the indicator data that has a greater impact on cigarette weight.

[0009] In some embodiments, the key indicator data includes one or more of the following data: Filament rate; Medium silk rate; Short silk rate; Broken silk rate; Cigarette making machine parameters.

[0010] In some embodiments, the method further comprises: When the predicted cigarette weight exceeds the reasonable cigarette weight range, a pre-built experience database is queried to determine a combination of control parameters; Based on the control parameter combination, the equipment parameters in the cigarette production process are adjusted.

[0011] In some embodiments, the method further comprises: Collect and analyze data before and after adjustment, and continuously optimize the neural network model and the experience library.

[0012] The present invention also provides a cigarette weight prediction device for a cigarette making machine, comprising: An acquisition module is used to obtain key indicator data in the cigarette production process; A prediction module, used to input the key indicator data into a trained neural network model to obtain a predicted value of cigarette weight output by the neural network model; The neural network model is obtained after being trained in advance with a plurality of sample data, each sample data including key indicator data and a cigarette weight label.

[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 computer program, any of the above-mentioned methods for predicting cigarette weights of cigarette making machines is implemented.

[0014] The present invention also provides a non-transitory computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, any of the above-mentioned methods for predicting cigarette weights of cigarette making machines is implemented.

[0015] The present invention also provides a computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the computer program implements any of the above-mentioned methods for predicting cigarette weights of cigarette making machines.

[0016] The cigarette weight prediction method and device for a cigarette making machine provided by the present invention use a neural network model to predict the weight of cigarettes in advance, thereby avoiding a large number of unqualified cigarettes caused by subsequent detection and improving production efficiency and resource utilization. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0018] Figure 1 It is a schematic flow chart of the method for predicting cigarette weight of a cigarette making machine provided by the present invention.

[0019] Figure 2 It is a logical flow diagram of cigarette weight prediction and control of a cigarette making machine provided by the present invention.

[0020] Figure 3 It is a schematic diagram of the cigarette weight prediction model provided by the present invention.

[0021] Figure 4 It is a schematic diagram of cigarette weight detection data of the comprehensive test bench provided by the present invention.

[0022] Figure 5 It is a schematic diagram comparing the actual value and predicted value of cigarette weight provided by the present invention.

[0023] Figure 6 It is a schematic diagram of the cigarette weight warning provided by the present invention.

[0024] Figure 7 It is a structural schematic diagram of a cigarette weight prediction device for a cigarette making machine provided by the present invention.

[0025] Figure 8 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0026] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0027] The weight of cigarettes is an important indicator in the cigarette quality evaluation system. At the same time, the weight of cigarettes is closely related to the resistance to draw and hardness. If the resistance to draw is too large, the aroma permeability will be poor, making it difficult for smokers to inhale. If the resistance to draw is too small, the aroma will be too bland and the smoke will be diluted by the air. At present, tobacco manufacturers mainly rely on manually operated weight control plug-in equipment to manually adjust the weight of cigarettes in the management and control of cigarette weight. This practice exposes several significant problems: first, the feedback mechanism is slow, resulting in untimely discovery of problems; second, the problem investigation process is chaotic and lacks systematicity; third, it is inefficient, affecting production speed; fourth, the accuracy is insufficient, making it difficult to meet high-quality production requirements; fifth, the lack of standardized processes leads to inconsistencies and increased difficulty in management.

[0028] In order to solve the technical problems in the prior art, the present invention introduces artificial intelligence technology, aiming to achieve advance prediction of cigarette weight. By building an intelligent prediction model, the project can monitor key parameters in the production process in real time, and use machine learning algorithms to analyze these data to identify trends in cigarette weight deviations in advance. Once it is predicted that the weight is about to deviate from the standard range, the system will immediately trigger an early warning mechanism, find a solution by querying the experience library method, and adjust the various parameters of the cigarette machine in time to ensure that the cigarette weight is always kept within a controllable and standard range.

[0029] This artificial intelligence-based advance prediction and real-time adjustment strategy can not only effectively reduce weight deviation problems in the production process and improve product quality and consistency, but also significantly improve production efficiency and resource utilization, bringing significant economic and social benefits to cigarette companies.

[0030] Figure 1 FIG. 1 is a flow chart of a method for predicting cigarette weight of a cigarette making machine provided by the present invention. Figure 1 As shown, the method includes the following: Step 101, obtaining key indicator data in the cigarette production process; Step 102: input the key indicator data into the trained neural network model to obtain the cigarette weight prediction value output by the neural network model; The neural network model is obtained after being trained in advance with a plurality of sample data, each sample data including key indicator data and a cigarette weight label.

[0031] In some embodiments, the method further comprises: Comprehensively collect relevant indicator data in the cigarette production process, as well as the weight of cigarettes produced under each relevant indicator data; The relevant indicator data are screened to determine the indicator data that has a greater impact on the weight of cigarettes as the key indicator data.

[0032] Figure 2 FIG. 1 is a schematic diagram of a logical flow of cigarette weight prediction and control of a cigarette making machine provided by the present invention, such as Figure 2 As shown, specifically, before cigarette production, relevant data will be fully collected first, including raw material data, production process parameters, environmental parameters, etc., so as to pre-calculate the weight of cigarettes, and compare the calculated cigarette weight with the standard weight value set based on the specifications of different brands of cigarettes. When the error value is within the preset range (for example, within 5%), production can be carried out.

[0033] The relevant indicator data in the cigarette production process include cigarette machine parameters, cigarette raw material data, environmental parameters and process parameter data, etc.

[0034] Cigarette making machine parameters include cut wire width, leveler groove width, air chamber positive and negative pressure, wire return amount, secondary stem removal amount, etc.

[0035] Cigarette raw material data include parameters of silk bundle, parameters of cigarette wrapping paper and parameters of tobacco shreds, etc.

[0036] Among them, the parameters of the tow include single-filament linear density, tow linear density, etc. The parameters of cigarette wrapping paper include wrapping paper width, basis weight, whiteness, tensile strength, ash content, moisture content, etc. The parameters of shredded tobacco include moisture content, long filament rate, medium filament rate, short filament rate, and broken filament rate, etc.

[0037] Environmental parameters include temperature, humidity, pressure difference, illumination, cleanliness, etc. in the production workshop.

[0038] The process parameter data include the process standards of cigarette making machines, alarm records of cigarette weight exceeding the standard, and the optimal process parameters for cigarettes of different specifications.

[0039] First, the reasonable weight of cigarettes is determined in advance to avoid blindness in the production process, reduce the waste of raw and auxiliary materials, and ensure product quality.

[0040] The estimated calculation formula for the weight N of a single cigarette is as follows: N=N1+N2+N3+N4

[0041] Among them, N1 represents the filter rod weight calculated according to the filter rod specifications, N2 represents the cigarette wrapping paper weight calculated according to the cigarette wrapping paper length, width and quantity, N3 represents the tobacco weight calculated according to the cigarette specifications and tobacco parameters, and N4 represents the tipping paper weight.

[0042] In order to determine whether a cigarette is suitable for trial production, the calculated weight of a single cigarette (denoted as N) needs to be compared with the standard weight value set based on the cigarette specifications. For example, the processing steps are as follows: 1. Calculate the percentage difference between the weight N of a single cigarette and the standard value.

[0043] 2. If the difference percentage is ≤5%, trial production is allowed.

[0044] 3. If the difference percentage is >5%, adjust the characteristics and specifications of the raw and auxiliary materials of the cigarette and recalculate until the conditions are met.

[0045] The neural network model in the present invention is obtained after being trained in advance with a plurality of sample data, and each sample data includes key indicator data and a cigarette weight label.

[0046] Figure 3 Schematic diagram of the cigarette weight prediction model provided by the present invention, such as Figure 3 As shown, a prediction model for cigarette weight, namely a neural network model, is constructed.

[0047] The data sources for building (training) the neural network model mainly include four parts: cigarette machine operating parameter data, physical inspection data of incoming cigarette materials, real-time cigarette inspection data on the comprehensive test bench, and environmental data of cigarette production (workshop temperature, humidity, etc.).

[0048] In some embodiments, the data may also be pre-processed.

[0049] Data preprocessing mainly deals with the connection between the two parts, connecting the data tables through the shortest time difference.

[0050] Part 1: Associate the physical inspection data of incoming cigarette materials with the operating parameter data of the cigarette making machine to form a model input variable table.

[0051] Part 2: Input the model successfully merged in Part 1 into the variable table and associate it with the quality inspection equipment data to obtain the processed data.

[0052] In some embodiments, special data processing may also be performed on the data, for example, including the following processing: Data merging: By correlating the data detection time and the cigarette making machine data in the two data sets, the corresponding real-time collected data of the detected cigarettes are found and merged into a unified data set.

[0053] Missing value processing: Process the missing values ​​in a unified data set and remove data records with missing values.

[0054] Outlier detection and processing: In order to reflect the abnormal production conditions in the actual production process, outlier detection and processing are performed on the records of abnormal values ​​of cigarette index items.

[0055] In some embodiments, the screening of the relevant indicator data to determine the indicator data having a greater impact on the cigarette weight includes: The random forest model is used to screen the relevant indicator data to determine the indicator data that has a greater impact on cigarette weight.

[0056] Specifically, variable screening can identify the variables with the greatest impact and the most predictive power on the target variable from a large number of possible variables. Through variable screening, the number of variables in the model can be significantly reduced, thereby reducing computational complexity, speeding up model training, and reducing the risk of overfitting.

[0057] The random forest classifier is a powerful ensemble learning algorithm that combines the power of multiple decision trees to improve the prediction accuracy and robustness of the model. Random forests have the characteristics of high accuracy, strong robustness, ability to handle high-dimensional data, and ability to evaluate the importance of features.

[0058] The present invention uses a random forest model for variable screening, and the random forest can perform importance analysis on feature variables. The OOB error evaluation feature variables are analyzed by random forest, and high-dimensional data samples are screened to obtain the importance of each feature variable, and several with higher importance are selected as input variables of the prediction model.

[0059] The specific implementation steps are as follows: 1. Assuming that the initial training set is N, resampling is performed through the bootstrap method, and multiple decision trees are constructed together with the point random splitting technique. When each tree is constructed, only a part of the randomly selected features is considered for splitting, which increases the diversity and robustness of the model. The unselected samples generated by the bootstrap method constitute the out-of-bag data (OOB).

[0060] 2. Assuming that there are M attributes (indicators) in the samples that affect the weight of cigarettes, when each node of the decision tree needs to be split, m attributes are randomly selected from the M attributes (m<M), and then the information gain is taken from the m attributes to select the optimal attribute as the classification variable of the node.

[0061] 3. Each node in the decision tree splitting process is processed according to step 2 until it can no longer be split.

[0062] 4. Random forest uses multiple decision trees to identify and classify new data, and finally decides the classification result through voting of each tree.

[0063] In some embodiments, the key indicator data includes one or more of the following data: Filament rate; Medium silk rate; Short silk rate; Broken silk rate; Cigarette making machine parameters.

[0064] During the training process of the present invention, the data set is divided into a training set and a test set in a ratio of 8:2. The training set is a data set used to train the model, while the test set is used to evaluate the performance of the model.

[0065] In some embodiments, a BP neural network is used as a prediction model.

[0066] The BP neural network has been expanded in structure and function to form a network model with stronger characteristics. This extended version of the BP neural network adds an extra intermediate layer between the hidden layer and the output layer in the traditional three-layer architecture (input layer, hidden layer, output layer). This layer serves as a delay operator (also called a memory layer or context layer) to store and process time series information or historical data in the data.

[0067] After a long period of research and analysis, the present invention finally decided to use BP neural network as the algorithm of the prediction model. BP neural network is trained according to the error back propagation algorithm and is mainly composed of an input layer, an output layer and one or more hidden layers.

[0068] The input of BP neural network is x i ;w ij is the weight of the input layer and the hidden layer; ф is the activation function of the hidden layer; w id is the weight between the hidden layer and the output layer; the output layer activation function is Ψ; θ i ,θ k are the thresholds of the hidden layer and the output layer respectively; θ k is the output of the neural network.

[0069] Specifically, in the input layer of the BP neural network, the characteristic variable values ​​affecting cigarette weight, such as long thread rate, medium thread rate, short thread rate, broken thread rate and cigarette making machine parameters, are input after the importance evaluation by random forest, and the cigarette weight is used as the network output of the BP neural network model.

[0070] For example, the neural network uses the Relu function from the input layer to the hidden layer, the Linear function is used in the output layer, the learning rate is 0.0001, and the number of training times is 3000. The number of hidden layer nodes is determined by an empirical formula, and the calculation formula is as follows:

[0071] In the formula, h is the number of neurons in the hidden layer nodes, m and n are the number of neurons in the input layer and output layer nodes respectively, and a is a regulation constant between 1 and 10. According to the formula, the hidden layer is determined to be between 5 and 10. Based on the comparison of training results of different numbers of hidden layer nodes, the present invention selects the number of hidden layer nodes to be 10.

[0072] The training process of BP neural network mainly includes the following steps: 1. In order to keep the dimension of the data consistent, the data can be normalized to [0, 1]. The normalization formula is as follows:

[0073] Among them, x is the initial data; x min 、x max are the minimum and maximum values ​​of the initial data respectively; x i The data are normalized.

[0074] 2. Determine the best set of feature variables by performing OOB analysis of random forest on the feature variables to optimize the model.

[0075] 3. Input the best feature variable set into the BP neural network, determine the network structure and the number of hidden layers for training.

[0076] 4. Test and verify the BP neural network to check whether the training effect and prediction accuracy meet expectations. If the maximum number of iterations and accuracy are reached, stop the BP neural network training and obtain the output results.

[0077] In some embodiments, after the model training is completed, the model can also be evaluated.

[0078] Specifically, after the modeling is completed, the model is evaluated. The main task of model evaluation is to evaluate the generalization ability, accuracy, stability and other aspects of the model.

[0079] The present invention uses mean absolute percentage error (MAPE), root mean square error (RMSE) and mean absolute error (MAE) as evaluation indicators to compare the performance of the prediction model. The calculation formula is as follows:

[0080] In the formula, p i is the predicted value of cigarette weight, y i is the actual value. The smaller the values ​​of MAPE, RMSE and MAE, the smaller the deviation between the predicted value and the actual value, and the better the prediction performance. Otherwise, the worse the prediction performance.

[0081] After the above training work is completed, the model needs to be deployed, which is mainly completed by operation and maintenance personnel and algorithm engineers. When deploying the model, we should pay attention to issues such as threads and computing power, and we need to improve the computing power and storage resource capabilities to the optimal state.

[0082] In some embodiments, the method further comprises: When the predicted cigarette weight exceeds the reasonable cigarette weight range, a pre-built experience database is queried to determine a combination of control parameters; Based on the control parameter combination, the equipment parameters in the cigarette production process are adjusted.

[0083] Specifically, the experience database contains a series of valuable experience accumulation from equipment fault diagnosis, maintenance process to post-maintenance reflection, which is of great significance for improving equipment maintenance efficiency, shortening fault recovery time, ensuring production continuity and cultivating maintenance personnel skills. The experience database includes the following contents: 1. Basic equipment information.

[0084] 2. Fault diagnosis and troubleshooting.

[0085] 3. Optimal process parameters.

[0086] The optimal process parameters are recorded in detail, and the optimal key process parameter configuration of the cigarette making machine is determined under conditions of different cigarette brands, specifications, raw materials from different manufacturers, etc.

[0087] 4. Alarm records of cigarette index items.

[0088] The cigarette index alarm record records in detail each time a cigarette index exceeds the standard and the processing process after the cigarette index exceeds the standard warning and alarm, including the record of the cigarette machine's key process parameters, environmental parameters, raw material parameters, and parameter records after maintenance.

[0089] 5. Case sharing.

[0090] 6. Data statistical analysis.

[0091] 7. Technical support and training.

[0092] The experience database can also include process parameter management, equipment management, cigarette brand management, cigarette raw material management, environmental parameter management, real-time optimization of process parameters, etc.

[0093] Figure 4 Schematic diagram of cigarette weight detection data of the comprehensive test bench provided by the present invention, such as Figure 4 As shown, the real-time monitoring of cigarette weight data displays the single value, mean, and SD value of the test cigarette data in real time, the real-time maximum, minimum, number of unqualified, mean, and SD value changes of the physical measurement indicators in a group of samples, as well as the monitoring of the warning information of cigarettes exceeding the standard.

[0094] Figure 5 : is a schematic diagram comparing the actual value and the predicted value of the cigarette weight provided by the present invention, such as Figure 5As shown in the figure, after comprehensively considering the above indicators, the system has completed the cigarette index correlation analysis model. The model collects the cigarette machine equipment status data in real time, combines the data of cigarette index items monitored in real time by the comprehensive test bench, and predicts the current development trend of cigarette weight based on the model established by artificial intelligence.

[0095] Figure 6 Schematic diagram of the cigarette weight warning provided by the present invention, such as Figure 6 As shown, the weight of each cigarette brand has an upper limit, a center value and a lower limit. If the indicator item meets the eight abnormality judgment criteria of SPC (statistical process control), the system will issue a timely warning.

[0096] The eight SPC criteria include: 1. One point falls outside area A.

[0097] 2. Nine consecutive points fall on the same side of the center line.

[0098] 3. Increase or decrease by 6 points in a row.

[0099] 4. Among the 14 consecutive points, the adjacent points alternate up and down.

[0100] 5. Two of the three consecutive points fall outside area B on the same side of the center line.

[0101] 6. 4 out of 5 consecutive points fall outside zone C on the same side of the center line.

[0102] 7. Fifteen consecutive points fall within zone C on both sides of the center line.

[0103] 8. Eight consecutive points fall on both sides of the center line and none of them are in area C.

[0104] Analysis of reasons why cigarette weight does not meet the eight SPC criteria.

[0105] When a cigarette indicator issues a warning, the system immediately queries the experience database and finds the main reason affecting the cigarette weight warning based on past experience and the parameters of the current cigarette machine's key indicators.

[0106] Send instructions to the cigarette making machine based on the early warning information.

[0107] After the system finds the reasons that affect the warning of excessive cigarette weight by searching the experience database, it will store the control parameter data in JSON format and use the HTTP protocol for network transmission. The JSON data is placed in the body of the HTTP request, and then the data is sent to the cigarette making machine and other equipment through the HTTP POST method.

[0108] The control parameters include: needle roller speed, large fan pressure, small fan pressure, wire return amount and air chamber negative pressure, etc.

[0109] After receiving the instruction, the cigarette making machine and other equipment will immediately parse the JSON data and adjust the key process parameters. During the adjustment process, the system will continuously monitor the warning information of the cigarette index prediction. If the warning system still alarms after adjusting the cigarette making machine parameters, the system will search the experience database again, find a solution, and resend the parameter adjustment information to the cigarette making machine until the cigarette weight prediction value reaches the preset standard.

[0110] During the production process, the present invention constructs a prediction model based on BP neural network based on the comprehensive collected relevant data and the cigarette weight value tested in real time by the comprehensive test bench to predict the change trend of cigarette weight in the next period of time. At the same time, an experience library containing historical production experience, successful cases and troubleshooting solutions is established to provide support for production optimization. During the production process, by real-time monitoring of cigarette index items, once the predicted cigarette weight does not meet the eight SPC criteria, the early warning mechanism is immediately triggered to send a signal to adjust the parameters to the cigarette machine to ensure that cigarette production meets quality requirements and maintains the stability of the production process. After the adjustment is completed, the system continues to monitor and predict the weight of cigarettes in real time to ensure that the weight value in the production process continues to meet production requirements. At the same time, the system collects and analyzes the data before and after the adjustment, continuously optimizes the prediction model and experience library, and improves the overall control effect and intelligence level.

[0111] At present, the industry has not yet developed a technology that can effectively predict excessive cigarette weight. The existing method is that after the comprehensive test bench detects that the cigarette index data exceeds the standard, the operator will find the cause and adjust the process parameters of the cigarette machine. This requires the operator to have rich work experience and skills to find the best solution, and the operator needs to spend a lot of time and energy. Due to the different experience and skills of the operators, the effect of handling the problem will also vary, with strong uncertainty, and it cannot meet the company's demand for early warning of quality.

[0112] The present invention comprehensively considers the real-time production status of the cigarette machine, the quality parameters of incoming tobacco, and the production environment parameters, and realizes real-time monitoring and early warning of cigarette weight. The system predicts that the cigarette index items will exceed the standard in the future, and will search the experience database for the first time to find the cause of the problem, and then issue a process parameter adjustment instruction to the cigarette machine. During the process of process parameter adjustment, the system will continuously monitor the cigarette production data and compare it with the preset indicators until the predicted cigarette index item data returns to normal or reaches the preset standard, and the warning of the cigarette index item exceeding the standard ends. This method can help enterprises quickly respond to and solve quality problems, improve production efficiency and quality, and reduce manual intervention and errors. At the same time, by analyzing and early warning the production data, it is also possible to achieve automatic control and optimization of the production process.

[0113] Based on the above analysis, the present invention has the following advantages: high early warning accuracy, fast response speed, less human intervention, high raw material utilization efficiency, and can save energy and increase efficiency for enterprises. Compared with existing products and technologies, the present invention has better application prospects.

[0114] The cigarette weight prediction device for a cigarette making machine provided by the present invention is described below. The cigarette weight prediction device for a cigarette making machine described below and the cigarette weight prediction method for a cigarette making machine described above can be referred to each other.

[0115] Figure 7 FIG. 1 is a schematic diagram of the structure of the cigarette weight prediction device for a cigarette making machine provided by the present invention. Figure 7 As shown, the cigarette weight prediction device for a cigarette making machine provided by the present invention comprises: The acquisition module 701 is used to acquire key indicator data in the cigarette production process; The prediction module 702 is used to input the key indicator data into the trained neural network model to obtain the cigarette weight prediction value output by the neural network model; The neural network model is obtained after being trained in advance with a plurality of sample data, each sample data including key indicator data and a cigarette weight label.

[0116] Specifically, the above-mentioned cigarette weight prediction device for a cigarette making machine provided in the embodiment of the present application can implement all the method steps implemented in the above-mentioned cigarette weight prediction method embodiment for a cigarette making machine, and can achieve the same technical effects. The parts and beneficial effects of this embodiment that are the same as those of the method embodiment will not be described in detail herein.

[0117] Figure 8 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 8As shown, the electronic device may include: a processor 810, a communications interface 820, a memory 830 and a communication bus 840, wherein the processor 810, the communications interface 820 and the memory 830 communicate with each other via the communication bus 840. The processor 810 may call the logic instructions in the memory 830 to execute the cigarette weight prediction method of the cigarette making machine, and the method includes: Obtain key indicator data in the cigarette production process; Inputting the key indicator data into a trained neural network model to obtain a predicted value of cigarette weight output by the neural network model; The neural network model is obtained after being trained in advance with a plurality of sample data, each sample data including key indicator data and a cigarette weight label.

[0118] In addition, the logic instructions in the above-mentioned memory 830 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or the part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0119] On the other hand, the present invention further provides a computer program product, the computer program product comprising a computer program, the computer program can be stored in a non-transitory computer-readable storage medium, when the computer program is executed by a processor, the computer can execute the cigarette weight prediction method of the cigarette making machine provided by the above methods, the method comprising: Obtain key indicator data in the cigarette production process; Inputting the key indicator data into a trained neural network model to obtain a predicted value of cigarette weight output by the neural network model; The neural network model is obtained after being trained in advance with a plurality of sample data, each sample data including key indicator data and a cigarette weight label.

[0120] In another aspect, the present invention further provides a non-transitory computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the method for predicting cigarette weight of a cigarette making machine provided by the above methods is implemented, and the method comprises: Obtain key indicator data in the cigarette production process; Inputting the key indicator data into a trained neural network model to obtain a predicted value of cigarette weight output by the neural network model; The neural network model is obtained after being trained in advance with a plurality of sample data, each sample data including key indicator data and a cigarette weight label.

[0121] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0122] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0123] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for predicting cigarette weight of a cigarette making machine, characterized in that: include: Obtain key indicator data in the cigarette production process; Inputting the key indicator data into a trained neural network model to obtain a cigarette weight prediction value output by the neural network model; The neural network model is obtained after being trained in advance with a plurality of sample data, each sample data including key indicator data and a cigarette weight label.

2. The method for predicting cigarette weight of a cigarette making machine according to claim 1, characterized in that: The method further comprises: Comprehensively collect relevant indicator data in the cigarette production process, as well as the weight of cigarettes produced under each relevant indicator data; The relevant indicator data are screened to determine the indicator data that has a greater impact on the weight of cigarettes as the key indicator data.

3. The method for predicting cigarette weight of a cigarette making machine according to claim 2, characterized in that: The screening of the relevant index data to determine the index data that has a greater impact on the weight of the cigarette includes: The random forest model is used to screen the relevant indicator data to determine the indicator data that has a greater impact on cigarette weight.

4. The method for predicting cigarette weight of a cigarette making machine according to claim 1, characterized in that: The key indicator data includes one or more of the following data: Filament rate; Medium silk rate; Short silk rate; Broken silk rate; Cigarette making machine parameters.

5. The method for predicting cigarette weight of a cigarette making machine according to claim 1, characterized in that: The method further comprises: When the predicted cigarette weight exceeds the reasonable cigarette weight range, a pre-built experience database is queried to determine a combination of control parameters; Based on the control parameter combination, the equipment parameters in the cigarette production process are adjusted.

6. The method for predicting cigarette weight of a cigarette making machine according to claim 5, characterized in that: The method further comprises: Collect and analyze data before and after adjustment, and continuously optimize the neural network model and the experience library.

7. A cigarette weight prediction device for a cigarette making machine, characterized in that: include: An acquisition module is used to obtain key indicator data in the cigarette production process; A prediction module, used to input the key indicator data into a trained neural network model to obtain a predicted value of cigarette weight output by the neural network model; The neural network model is obtained after being trained in advance with a plurality of sample data, each sample data including key indicator data and a cigarette weight label.

8. 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 computer program, the method for predicting cigarette weight of a cigarette making machine as claimed in any one of claims 1 to 6 is implemented.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for predicting cigarette weight of a cigarette making machine as claimed in any one of claims 1 to 6 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method for predicting cigarette weight of a cigarette making machine as claimed in any one of claims 1 to 6 is implemented.

Images