Ingredient quality control method based on deep learning

Abstract

The invention discloses an ingredient quality control method based on deep learning, and the method comprises the following steps: determining the ingredient type of cigarettes and the target quality of each ingredient when an ingredient instruction is received; acquiring image information and temperature of each ingredient and pressure of a discharging pipeline; determining the viscosity characteristic of each ingredient according to the image information, temperature and viscosity prediction model of each ingredient; determining control information of each ingredient according to the viscosity characteristic of each ingredient, the target mass and the pressure of the discharging pipeline; and controlling the batching device according to the batching control information. The cigarette ingredient type and the target quality are determined by receiving the ingredient instruction, image information, temperature and discharging pipeline pressure are synchronously obtained, viscosity characteristics are predicted, control information is dynamically generated, the problem of high-viscosity raw material discharging precision can be solved, the method can dynamically adapt to complex working conditions such as equipment performance degradation and material characteristic difference, and the product quality is improved. Therefore, deviation of batching quality control is effectively avoided.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of tobacco processing control, and particularly relates to a kind of based on deep learning's ingredient quality control method. BACKGROUND

[0002] Under the current trend of high-quality development of tobacco industry, the informatization, intelligentization and automation degree of processing process have become an important indicator to measure the core competitiveness of enterprises. As a key link to realize fine processing and flexible production, spice kitchen can accurately control the blending and addition of flavoring materials according to different product requirements, which is of great significance to improve the quality of tobacco products. The accuracy of flavoring material ratio in cut tobacco directly determines the smoking characteristics of cut tobacco and the quality of final product, and slight ratio error can cause significant difference in product flavor and affect consumer experience.

[0003] However, tobacco processing link faces many difficulties in actual production. From the aspect of ingredient equipment, due to the long service life of existing equipment, long-term high-frequency use causes serious wear of valve parts, and the precision is greatly reduced. At the same time, for special raw materials with small amount and high viscosity, traditional ingredient equipment is difficult to accurately control the discharge amount, which causes deviation in raw material addition and affects the stability of cut tobacco quality. SUMMARY

[0004] Therefore, the present application provides a kind of based on deep learning's ingredient quality control method, which aims to solve the problem that it is difficult to accurately control the discharge amount of special raw materials with small amount and high viscosity in the prior art, which causes deviation in raw material addition.

[0005] The technical scheme adopted by the present application to solve the above technical problems is: The ingredient quality control method based on deep learning comprises Upon receiving an ingredient instruction, determining the ingredient type of the cigarette and the target quality of each ingredient; Obtaining image information, temperature and discharge pipeline pressure of each ingredient; According to the target quality, image information, temperature and discharge pipeline pressure of each ingredient, determining the control information of each ingredient; According to the ingredient control information, controlling the ingredient device.

[0006] As an improvement of the above technical scheme, according to the target quality, image information, temperature and discharge pipeline pressure of each ingredient, determining the control information of each ingredient, comprising According to the image information, temperature and viscosity prediction model of each ingredient, determining the viscosity characteristics of each ingredient; According to the viscosity characteristics, target quality and discharge pipeline pressure of each ingredient, determining the control information of each ingredient.

[0007] As an improvement of the above technical solution, the viscosity characteristics of each ingredient are determined according to the image information, temperature and viscosity prediction model of each ingredient, including According to the image information and identification model of each ingredient, the appearance characteristics and flow behavior characteristics of each ingredient are determined; According to the appearance characteristics, flow behavior characteristics, temperature and viscosity prediction model of each ingredient, the viscosity characteristics of each ingredient are determined.

[0008] As an improvement of the above technical solution, the appearance characteristics and flow behavior characteristics of each ingredient are determined according to the image information and identification model of each ingredient, including According to the image information and identification model of each ingredient, its particle distribution, color uniformity, colloid drawing shape and diffusion rate during stirring are determined; According to the particle distribution and color uniformity, the appearance characteristics of each ingredient are determined; According to the colloid drawing shape and diffusion rate during stirring, the flow behavior characteristics of each ingredient are determined.

[0009] As an improvement of the above technical solution, the control information of each ingredient is determined according to the viscosity characteristics of each ingredient, target quality and discharge pipeline pressure, including According to the viscosity characteristics of each ingredient, target quality, discharge pipeline pressure and preset mapping relationship, the valve opening and pump speed are determined as the control information of each ingredient.

[0010] The present application solves the above technical problems and also adopts the following technical solutions: The ingredient quality control device based on deep learning comprises The first target acquisition unit is used to determine the ingredient type of cigarettes and the target quality of each ingredient when receiving the ingredient instruction; The second target acquisition unit is used to acquire the image information, temperature and discharge pipeline pressure of each ingredient; The target data processing unit is used to determine the control information of each ingredient according to the target quality, image information, temperature and discharge pipeline pressure of each ingredient; The first ingredient control unit is used to control the ingredient device according to the ingredient control information.

[0011] As an improvement of the above technical solution, the target data processing unit comprises The first model prediction unit is used to determine the appearance characteristics and flow behavior characteristics of each ingredient according to the image information and identification model of each ingredient; The second model prediction unit is used to determine the viscosity characteristics of each ingredient according to the appearance characteristics, flow behavior characteristics, temperature and viscosity prediction model of each ingredient; The data mapping relationship unit is configured to determine the control information of each ingredient according to the viscosity characteristics, target quality, and discharge pipeline pressure of each ingredient.

[0012] As an improvement of the above technical solution, the first model prediction unit is specifically configured to determine the particle distribution, color uniformity, colloid stringing form, and diffusion rate during stirring of each ingredient according to the image information and recognition model of each ingredient. The appearance characteristics of each ingredient are determined according to the particle distribution and color uniformity. The flow behavior characteristics of each ingredient are determined according to the colloid stringing form and diffusion rate during stirring.

[0013] As an improvement of the above technical solution, the data mapping relationship unit is specifically configured to determine the valve opening and pump speed as the control information of each ingredient according to the viscosity characteristics, target quality, discharge pipeline pressure, and preset mapping relationship of each ingredient.

[0014] Meanwhile, the present application also provides an electronic device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the steps of the above-mentioned deep learning-based ingredient quality control method when executing the computer program.

[0015] Meanwhile, the present application also provides a computer-readable storage medium, which stores a computer program, and the computer program implements the steps of the above-mentioned deep learning-based ingredient quality control method when executed by a processor.

[0016] The present application has the following beneficial effects: The present application can determine the cigarette ingredient type and target quality by receiving the ingredient instruction, synchronously acquire image information, temperature, and discharge pipeline pressure, predict viscosity characteristics, and dynamically generate control information, thereby solving the discharge precision problem of high-viscosity raw materials, dynamically adapting to complex working conditions such as equipment performance degradation and material property differences, and effectively avoiding the deviation of ingredient quality control. BRIEF DESCRIPTION OF DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0018] Figure 1 is a control method flowchart provided by the embodiments of the present application; Figure 2 is a control device structure schematic diagram provided by the embodiments of the present application; Figure 3 Figure 1 is a structural schematic diagram of an electronic device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0019] The technical solutions in the embodiments of the present application will be clearly and completely described with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative effort belong to the scope of protection of the present application.

[0020] Unless otherwise defined, technical terms or scientific terms used in the present application shall have the common meaning understood by a person of ordinary skill in the art to which the present application belongs. The terms "first", "second", and similar terms used in the present application do not indicate any order, number, or importance, but are only used to distinguish different components. The terms "include", "contain", and similar terms mean that the components or objects before the terms encompass the components or objects listed after the terms and their equivalents, and do not exclude other components or objects. The terms "connect" or "connected" and similar terms are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. The terms "upper", "lower", "left", "right", and the like are only used to indicate relative positional relationships, and when the absolute positions of the described objects change, the relative positional relationships can also change accordingly.

[0021] The following description is for the purpose of explanation and is not intended to limit the present application, and specific details such as specific system structures, techniques, and the like are presented in order to thoroughly understand the embodiments of the present application. However, it should be clear to a person of ordinary skill in the art that the present application can be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted in order not to obscure the description of the present application with unnecessary details.

[0022] Referring to Figure 1 The first embodiment of the present application relates to a deep learning-based ingredient quality control method, comprising Upon receiving an ingredient instruction, determining the type of the cigarette and the target quality of each ingredient; Obtaining image information, temperature, and discharge pipeline pressure of each ingredient; Determining control information of each ingredient according to the target quality, image information, temperature, and discharge pipeline pressure of each ingredient; Controlling the ingredient device according to the ingredient control information.

[0023] In the embodiment of the present application, in the tobacco processing production process, when the control system of the spice kitchen receives the ingredient instruction, the first task is to determine the ingredient type of the cigarette and the target mass of each ingredient. The specific implementation process is as follows: First, the ingredient instruction is generated and issued by the production execution system, which contains rich and key information. The “cigarette_type” field clearly indicates the cigarette type, such as “flue-cured type-A level”, which directly determines the type and ratio of the required ingredients; the “formula_version” field records the formula version, such as “V3.2”, and the system can call the corresponding version of the precise formula data according to this; the “ingredients” array lists each ingredient information in detail, including “ingredient_code” (ingredient code, such as “SP-001”), “ingredient_name” (ingredient name, such as “fruit flavor essence”), “target_mass” (target mass, usually in kg, such as 5.2 kg), and “property” (material property label, used to identify whether the ingredient is a high-viscosity material “high_viscosity” or a small-quantity material “small_quantity”).

[0024] After obtaining the instruction, the system will immediately start a strict production plan verification process. On the one hand, the formula version is verified, and the formula version number in the instruction is compared with the valid formula version stored in the database. Only when the latest and correct version is matched, can the ingredient process follow the current optimal process standard; on the other hand, the material availability check is carried out, according to the ingredient list in the instruction, the raw material inventory system is queried, and it is confirmed whether the inventory quantity of each ingredient meets the production demand, and whether the batch of corresponding material in the warehouse is within the shelf life, and whether the storage condition meets the requirements. If it is found that the inventory of a certain ingredient is insufficient or there is a quality risk, the system will issue a warning in time, and the ingredient process will be suspended to avoid production interruption or quality accident.

[0025] After verification, the system will further refine the ingredient type in the ingredient instruction. For different types of cigarettes, the ingredient type is significantly different. For example, flue-cured cigarettes may focus on adding various fruit and caramel flavors to highlight unique flavors, while mixed cigarettes may need to adjust different proportions of natural and synthetic flavors. The system determines the role and characteristics of each ingredient in the entire formula system according to the cigarette type, combined with historical production data and process requirements, to provide a basis for subsequent precise control.

[0026] In the embodiment of the present application, by receiving the ingredient instruction to determine the cigarette ingredient type and the target quality, synchronously acquiring the image information, temperature and discharge pipeline pressure to predict the viscosity characteristics and dynamically generate the control information, the problem of high-viscosity raw material discharge precision can be solved, and the complex working conditions such as equipment performance degradation and material property difference can be dynamically adapted, so as to effectively avoid the deviation of the ingredient quality control.

[0027] As a possible implementation manner, according to the target quality, image information, temperature and discharge pipeline pressure of each ingredient, the control information of each ingredient is determined, including According to the image information, temperature and viscosity prediction model of each ingredient, the viscosity characteristics of each ingredient are determined. According to the viscosity characteristics, target quality and discharge pipeline pressure of each ingredient, the control information of each ingredient is determined.

[0028] In the embodiment of the present application, first, the real-time image information of each ingredient is collected by a high-resolution industrial camera. The content of the image capture covers multiple key dimensions: at the particle level, the particle size distribution, shape regularity and agglomeration state can be analyzed; at the color level, the color uniformity and color difference range can be quantified; for colloidal materials, the image can also record the filament morphology (such as filament length and fracture characteristics); in the dynamic process, the diffusion rate and mixing uniformity during stirring can be analyzed through continuous image sequence. At the same time, a high-precision temperature sensor monitors the ingredient temperature in real time, which is accurate to 0.1℃ and covers the change of the environment temperature from the raw material storage temperature to the conveying process.

[0029] As a possible implementation manner, according to the image information, temperature and viscosity prediction model of each ingredient, the viscosity characteristics of each ingredient are determined, including According to the image information and identification model of each ingredient, the particle distribution, color uniformity, colloidal filament morphology and diffusion rate during stirring are determined. According to the particle distribution and color uniformity, the appearance characteristics of each ingredient are determined. According to the colloidal filament morphology and diffusion rate during stirring, the flow behavior characteristics of each ingredient are determined. According to the appearance characteristics, flow behavior characteristics, temperature and viscosity prediction model of each ingredient, the viscosity characteristics of each ingredient are determined.

[0030] In the embodiment of the present application, the recognition model adopts a convolutional neural network architecture and has accurate recognition capability for various image features. Before use, the recognition model is trained using labeled ingredient image datasets, which contain various samples under different viscosity and temperature conditions, so that the model can accurately recognize key features such as particle distribution, color uniformity, and colloid string shape. After the input image is processed by the model, the output is particle distribution parameters (such as average particle size and particle size distribution variance), color uniformity indicators (such as RGB mean deviation), colloid string quantification features (such as maximum string length and fracture frequency), and diffusion rate values (such as unit time diffusion area change). Based on these data, the appearance characteristics of the ingredients are determined according to the uniformity of the particle distribution (the smaller the particle size distribution variance, the higher the uniformity), the agglomeration index (the percentage of clumping area or the statistical distance between particles), and the color difference threshold (the standard deviation of RGB values) and the stratification coefficient (the percentage of stratification area). The flow behavior characteristics of the ingredients are determined according to the toughness of the colloid string (the longer the maximum string length and the lower the fracture frequency, the higher the viscosity), the fracture morphology (the smoothness of the fracture surface reflects the cohesion), and the coefficient of stirring diffusion (the diffusion area change) and the mixing time constant (the time required to reach a specified uniformity).

[0031] Then, the determined appearance characteristics and flow behavior characteristics, together with the temperature data obtained by the high-precision temperature sensor, are input into a viscosity prediction model, such as a long short-term memory neural network (LSTM). The model is trained with a large amount of data to fully learn the mapping relationship between the ingredient characteristics and the viscosity, and then outputs key viscosity-related parameters such as dynamic viscosity (reflecting the internal friction of the ingredient flow), apparent viscosity (a comprehensive indicator considering the non-Newtonian fluid characteristics), and viscosity-temperature coefficient (the sensitivity of viscosity to temperature change), thereby forming a vector as the viscosity characteristic. This non-contact real-time monitoring method effectively avoids the pollution or disturbance caused by traditional contact measurement of the viscometer; the deep fusion of multi-dimensional features greatly improves the accuracy of viscosity prediction, and at the same time, it can also track the dynamic changes of the ingredient characteristics in the production process, such as the decrease of viscosity with the increase of temperature during stirring, providing reliable dynamic basis for subsequent precise ingredient control based on viscosity, such as valve opening adjustment and pump speed setting, thereby fundamentally solving the problem of difficult control of high-viscosity raw material discharge accuracy.

[0032] As a possible implementation manner, the control information of each ingredient is determined according to the viscosity characteristics of each ingredient, the target quality, and the discharge pipeline pressure, including According to the viscosity characteristics of each ingredient, the target quality, the discharge pipeline pressure, and the preset mapping relationship, the valve opening and the pump speed are determined as the control information of each ingredient.

[0033] In the embodiment of the present application, in the ingredient dispensing scenario of the tobacco flavor kitchen, there is a technical background that the valve is worn out due to long-term high-frequency use of the equipment, and it is difficult to accurately dispense the high-viscosity essence flavor. The process of determining the ingredient control information is particularly critical. Taking the production of flue-cured tobacco cigarettes as an example, after receiving the ingredient instruction, the system determines that the target mass of the fruity essence is 5 kg. Through the high-resolution industrial camera, temperature sensor and pressure sensor, the image information, current temperature and discharge pipeline pressure of the essence are obtained, which are 28 ℃ and 0.3 MPa respectively, and the viscosity characteristics thereof are high viscosity and poor flowability.

[0034] Based on the analysis of a large amount of production data of the tobacco flavor kitchen in the early stage, the system establishes a preset mapping relationship. The relationship records in detail the valve opening and pump speed parameters under different viscosity, target mass and discharge pipeline pressure combinations according to the characteristics of tobacco ingredients. When the system substitutes the viscosity characteristics of the fruity essence, 5 kg target mass and 0.3 MPa discharge pipeline pressure data into the mapping relationship, through retrieval and calculation, it is concluded that in order to ensure the accurate dispensing of the high-viscosity essence, the first flow control valve opening needs to be set to 30%, and at the same time, the conveying pump with a pump speed of 80 r / min needs to be started. This control information will drive the ingredient device to first carry out preliminary flow control through the first flow control valve, and then combine the error between the actual discharge mass and the target mass to carry out fine adjustment by using the high-precision second flow control valve, so as to overcome the problems caused by the decline of equipment precision and the difference of material characteristics, realize the accurate proportioning of essence flavors, and guarantee the smoke taste characteristics of tobacco and the stability of product quality.

[0035] The ingredient mass control method described in the embodiment further includes collecting the actual mass of each ingredient; determining adjustment information according to the actual mass and the target mass; controlling the ingredient device according to the adjustment information.

[0036] As a possible implementation manner, the ingredient device includes a first flow control valve and a second flow control valve corresponding to each ingredient; the precision of the second flow control valve is greater than that of the first flow control valve. Therefore, the control of the ingredient device includes controlling the first flow control valve according to the ingredient control information; controlling the second flow control valve according to the adjustment information.

[0037] As a possible implementation manner, determining the adjustment information according to the actual mass and the target mass includes calculating the error between the actual mass and the target mass; adopting an adaptive PID algorithm to dynamically adjust the control information according to the error to obtain the adjustment information; Among them, the adaptive PID algorithm adjusts the proportional coefficient, integral coefficient and differential coefficient according to the viscosity characteristics of each ingredient and the discharge pipeline pressure.

[0038] In some embodiments of the actual production of the tobacco flavor kitchen, to ensure that the ingredient accuracy meets the high standard requirements, a real-time feedback and dynamic adjustment mechanism is introduced in addition to the basic process of the deep learning-based ingredient quality control method. First, the actual discharge quality of each ingredient is collected in real time through high-precision weighing sensors or mass flow meters and other devices. These sensors are deployed at the discharge port of the ingredient device and can acquire quality data at a very high frequency (such as multiple times per second), ensuring that subtle changes in the ingredient process are captured. After obtaining the actual quality data, it is strictly compared with the target quality determined when the ingredient instruction is received. Specifically, the error between the actual quality and the target quality is first calculated, for example, the target is to add 5 kg of fruit flavor essence, and if the actual collected quality is 4.8 kg, the error is 0.2 kg. Subsequently, the system uses an adaptive PID algorithm to dynamically adjust the control information to determine the adjustment information. This algorithm considers factors such as the viscosity characteristics of the ingredient obtained in advance and the current discharge pipeline pressure to intelligently adjust the proportional coefficient, integral coefficient and differential coefficient. For example, if the viscosity of the ingredient is high, the algorithm will appropriately increase the proportional coefficient to speed up the adjustment; if the discharge pipeline pressure is unstable, the differential coefficient is adjusted to enhance the system's anti-interference ability to pressure fluctuations. Through algorithm operation, the final adjustment information such as valve opening degree and pump speed is obtained.

[0039] Finally, according to the generated adjustment information, the ingredient device is precisely controlled. The ingredient device includes a first flow control valve and a second flow control valve, with the second flow control valve having higher precision. The system will first use the first flow control valve to preliminarily control the flow according to the control information determined in advance, to quickly approach the target quality; and after obtaining the adjustment information, the second flow control valve is used for fine-tuning to ensure that the actual addition quality of each ingredient is highly consistent with the target quality. Taking the addition of fruit flavor essence as an example, if it is calculated that the discharge amount needs to be increased, the system will appropriately increase the opening degree of the second flow control valve and adjust the pump speed to accurately supplement the difference in subsequent discharge amount, thereby effectively solving the ingredient deviation problem caused by equipment aging and material property differences, and ensuring the stability and consistency of the tobacco quality.

[0040] Taking the addition of a certain high-viscosity essence as an example, after receiving the instruction, the first flow control valve quickly acts to adjust the valve opening degree to the corresponding position, and cooperates with the set pump speed to output the essence at a flow close to the target quality, completing most of the ingredient quantity in a short time and achieving rapid approximation to the target quality.

[0041] During the batching process, the system continuously collects the actual mass of each ingredient through high-precision sensors and compares it with the target mass in real time. When an error is calculated between the actual mass and the target mass, the adaptive PID algorithm is triggered to dynamically optimize the control information. This algorithm takes the error as the core input, and combines key parameters such as the viscosity characteristics of the ingredient and the discharge pipeline pressure to intelligently adjust the proportional coefficient, integral coefficient, and differential coefficient. For high-viscosity ingredients, due to their poor flowability and the hysteresis of the discharge, the algorithm will appropriately increase the proportional coefficient to speed up the control response, allowing the system to more quickly respond to flow deviations. If the discharge pipeline pressure fluctuates greatly, the algorithm will adjust the differential coefficient to enhance the system's ability to predict and resist pressure changes, avoiding unstable flow caused by pressure fluctuations. Through iterative optimization of the control parameters by the algorithm, precise adjustment information is ultimately generated.

[0042] After the adjustment information is generated, the system sends it to the second flow control valve. The second flow control valve, with its high-precision characteristics, adjusts the valve opening according to the adjustment information, making small and precise adjustments to the ingredient flow for the final fine-tuning. For example, when the actual addition of a certain flavor is detected to be 0.1 kg different from the target amount, the second flow control valve can precisely control the valve opening change to adjust the actual mass to the target value with a very small flow increment or decrement, effectively compensating for ingredient deviations caused by factors such as equipment wear and tear and material property differences, ensuring the high precision and stability of tobacco batching, and providing reliable protection for the consistency of tobacco quality.

[0043] It should be understood that the description of the order of the steps in the above embodiments does not mean the order of execution, and the execution order of each process should be determined according to its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0044] Figure 2 is a structural schematic diagram of a batching quality control device based on deep learning provided by a second embodiment of the present application. The batching quality control device comprises: a first target acquisition unit configured to determine the type of tobacco batching and the target mass of each ingredient when receiving a batching instruction; a second target acquisition unit configured to acquire image information, temperature, and discharge pipeline pressure of each ingredient; a target data processing unit configured to determine control information for each ingredient based on the target mass, image information, temperature, and discharge pipeline pressure of each ingredient; a first batching control unit configured to control the batching device based on the batching control information.

[0045] As a possible implementation, the target data processing unit comprises The first model prediction unit is configured to determine the appearance feature and the flow behavior feature of each ingredient according to the image information of each ingredient and the recognition model; The second model prediction unit is configured to determine the viscosity feature of each ingredient according to the appearance feature, the flow behavior feature, the temperature and the viscosity prediction model of each ingredient; The data mapping relationship unit is configured to determine the control information of each ingredient according to the viscosity feature of each ingredient, the target quality and the discharge pipeline pressure.

[0046] As a possible implementation manner, the first model prediction unit is specifically configured to determine the particle distribution, the color uniformity, the colloid drawing shape and the diffusion rate during stirring of each ingredient according to the image information of each ingredient and the recognition model. The appearance feature of each ingredient is determined according to the particle distribution and the color uniformity. The flow behavior feature of each ingredient is determined according to the colloid drawing shape and the diffusion rate during stirring.

[0047] As a possible implementation manner, the data mapping relationship unit is specifically configured to determine the valve opening degree and the pump speed as the control information of each ingredient according to the viscosity feature of each ingredient, the target quality, the discharge pipeline pressure and the preset mapping relationship.

[0048] In the embodiment of the present application, the control device described above further comprises a third target acquisition unit configured to collect the actual quality of each ingredient, and determine the adjustment information according to the actual quality and the target quality. The second ingredient control unit is configured to control the ingredient device according to the adjustment information.

[0049] As a possible implementation manner, the ingredient device comprises a first flow control valve and a second flow control valve corresponding to each ingredient; the precision of the second flow control valve is greater than that of the first flow control valve. The first ingredient control unit is configured to control the first flow control valve according to the ingredient control information. The second ingredient control unit is configured to control the second flow control valve according to the adjustment information.

[0050] As a possible implementation manner, the third target acquisition unit comprises the following steps for determining the adjustment information: calculating the error of the actual quality and the target quality; adopting the adaptive PID algorithm to dynamically adjust the control information according to the error to obtain the adjustment information; The adaptive PID algorithm is used to adjust the proportional coefficient, the integral coefficient and the differential coefficient according to the viscosity feature of each ingredient and the discharge pipeline pressure.

[0051] Figure 3is a schematic diagram of an electronic device 30 provided by an embodiment of the present application. As shown in Figure 3 The electronic device 30 of this embodiment includes a processor 31, a memory 32, and a computer program 33, such as a batch quality control program, stored in the memory 32 and executable on the processor 31. The processor 31 implements the steps in each of the above batch quality control method embodiments when executing the computer program 33, such as the steps shown in Figure 1 Alternatively, the processor 31 implements the functions of each module / unit in each of the above batch quality control apparatus embodiments when executing the computer program 33, such as the modules / units shown in Figure 2

[0052] For example, the computer program 33 can be divided into one or more modules / units, one or more of which are stored in the memory 32 and executed by the processor 31 to complete the present application. One or more modules / units can be a series of computer program instruction segments capable of completing a specific function, which are used to describe the execution process of the computer program 33 in the electronic device 30.

[0053] The electronic device 30 can be a desktop computer, a notebook computer, a palm computer, a cloud server, or the like. The electronic device 30 can include, but is not limited to, the processor 31 and the memory 32. Those skilled in the art can understand that Figure 3 The electronic device 30 is merely an example and does not constitute a limitation on the electronic device 30, which can include more or fewer components than shown, or combine certain components, or include different components, such as an input / output device, a network access device, a bus, and the like.

[0054] The processor 31 can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic components, discrete hardware components, or the like. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor.

[0055] ​The memory 32 can be an internal storage unit of the electronic device 30, for example, a hard disk or a memory of the electronic device 30. The memory 32 can also be an external storage device of the electronic device 30, for example, a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) card, a Flash Card, or the like equipped on the electronic device 30. Further, the memory 32 can include both the internal storage unit and the external storage device of the electronic device 30. The memory 32 is used to store a computer program and other programs and data required by the electronic device 30. The memory 32 can also be used to temporarily store data that has been output or will be output.

[0056] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the division of the above functional units and modules is exemplified, and in actual application, the above functions can be completed by different functional units and modules according to needs, that is, the internal structure of the device is divided into different functional units or modules to complete all or part of the functions described above. Each functional unit and module in the embodiment can be integrated in one processing unit, or each unit can exist physically, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of software functional unit. In addition, the specific names of each functional unit and module are only for easy distinction, and do not limit the protection scope of the application. The specific working process of the units and modules in the above system can refer to the corresponding process in the foregoing method embodiments, which will not be described here.

[0057] In the above embodiments, the description of each embodiment has its own emphasis, and the parts not described or recorded in detail in a certain embodiment can be referred to the relevant description of other embodiments.

[0058] Those of ordinary skill in the art can appreciate that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether the functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. A person skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0059] In the embodiments of the present application, it should be understood that the disclosed apparatus / equipment and method can be implemented in other manners. For example, the embodiments of the apparatus / equipment described above are merely schematic, and the division of the modules or units can be different, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections can be indirect couplings or communication connections through some interfaces, devices or units, and can be in electrical, mechanical or other forms.

[0060] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they may be located in one place, or distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the embodiments.

[0061] In addition, the functional units in each embodiment of the present application can be integrated in one processing unit, or each unit can be physically present separately, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of a software functional unit.

[0062] If the integrated module / unit is realized in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer readable storage medium. Based on this understanding, all or part of the processes in the above-mentioned embodiment methods can also be completed by computer programs instructing related hardware, and the computer programs can be stored in a computer readable storage medium. When the processor executes the computer programs, the steps of each method embodiment described above can be implemented. The computer program includes computer program code, which can be in the form of source code, object code, executable file or some intermediate form. The computer readable medium can include any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal and software distribution medium, etc.

[0063] The above examples are only used to illustrate the technical solutions of the present application, but not limit the present application; although the present application has been described in detail with reference to the foregoing examples, those ordinarily skilled in the art should understand: the technical solutions recorded in the foregoing examples can still be modified, or some technical features can be replaced equivalently; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application.

Claims

1. Ingredient quality control method based on deep learning, characterized by: include Upon receiving the ingredient instruction, determining the ingredient type of the cigarette and the target mass of each ingredient; Obtain image information, temperature and discharge pipeline pressure of each ingredient; Determine the viscosity characteristics of each ingredient based on its image information, temperature, and viscosity prediction model; Determine the control information for each ingredient based on its viscosity characteristics, target quality, and discharge line pressure; The batching device is controlled according to the batching control information.

2. The method for controlling the quality of ingredients according to claim 1, wherein: Determine the viscosity characteristics of each ingredient based on its image information, temperature, and viscosity prediction model, including Determine the appearance characteristics and flow behavior characteristics of each ingredient based on the image information and recognition model of each ingredient; The viscosity characteristics of each ingredient are determined based on its appearance characteristics, flow behavior characteristics, temperature and viscosity prediction model.

3. The method for controlling the quality of ingredients according to claim 2, wherein: Based on the image information and recognition model of each ingredient, the appearance characteristics and flow behavior characteristics of each ingredient are determined, including Based on the image information and recognition model of each ingredient, determine its particle distribution, color uniformity, colloid drawing morphology, and diffusion rate during stirring; Determine the appearance characteristics of each ingredient based on particle distribution and color uniformity; The flow behavior characteristics of each ingredient were determined based on the colloid stringing morphology and the diffusion rate during stirring.

4. The method for controlling the quality of ingredients according to claim 1, wherein: Determine the control information of each ingredient based on its viscosity characteristics, target quality and discharge pipeline pressure, including According to the viscosity characteristics, target quality, discharge pipeline pressure and preset mapping relationship of each ingredient, the valve opening and pump speed are determined as the control information of each ingredient.

5. The method for controlling the quality of ingredients according to claim 1, wherein: The control method further includes Collect the actual mass of each ingredient; Determine adjustment information based on actual quality and target quality; The batching device is controlled according to the adjustment information.

6. The method for controlling ingredient quality according to claim 5, wherein: The batching device includes a first flow control valve and a second flow control valve corresponding to each ingredient; the second flow control valve has a higher precision than the first flow control valve; Control of batching equipment, including Controlling the first flow control valve according to the batching control information; The second flow control valve is controlled according to the adjustment information.

7. Ingredient quality control device based on deep learning, characterized by: The control device is used to implement the ingredient quality control method according to any one of claims 1 to 6, comprising a first target acquisition unit, configured to determine the type of ingredients of the cigarette and the target mass of each ingredient upon receiving an ingredient instruction; The second target acquisition unit is used to obtain the image information, temperature and discharge pipeline pressure of each ingredient; A target data processing unit is used to determine the control information of each ingredient based on the target quality, image information, temperature and discharge pipeline pressure of each ingredient; The first batching control unit is used to control the batching device according to the batching control information.

8. The ingredient quality control device according to claim 7, characterized in that: The target data processing unit includes A first model prediction unit is used to determine the appearance characteristics and flow behavior characteristics of each ingredient based on the image information of each ingredient and the recognition model; a second model prediction unit, for determining the viscosity characteristics of each ingredient based on the appearance characteristics, flow behavior characteristics, temperature and viscosity prediction model of each ingredient; The data mapping unit is used to determine the control information of each ingredient according to the viscosity characteristics, target quality and discharge pipeline pressure of each ingredient.

9. 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 steps of the ingredient quality control method according to any one of claims 1 to 6 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the ingredient quality control method according to any one of claims 1 to 6 are implemented.

Images