Intelligent control system and method for chocolate production process

Abstract

The invention discloses a chocolate production process intelligent control system and method, and relates to the technical food intelligent manufacturing field, and the system comprises an intelligent defoaming forming module, a sensor network, an intelligent control unit, a big data analysis platform and a human-computer interaction interface; the intelligent defoaming and forming module is used for intelligently and precisely controlling the defoaming and forming process of the chocolate paste; the sensor network is used for collecting temperature, humidity, pressure and viscosity in the chocolate production process and shape and size information in the forming process in real time; the intelligent control unit is used for receiving data transmitted by the sensor network, analyzing and processing the data and then sending out a control instruction; the big data analysis platform deploys a digital twin model and is used for simulating and predicting production data and generating an optimization strategy; and the man-machine interaction interface is used for providing an operation interface. According to the chocolate production system, the problems of control lag, insufficient precision and over-high energy consumption in the existing chocolate production process are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent food manufacturing, and in particular to an intelligent control system and method for a chocolate production process. Background Art

[0002] Traditional methods have significant shortcomings in the core production process of defoaming and molding chocolate paste. These processes typically rely on preset, fixed parameters, such as vibration frequency and the pressure and position of the molding mold. However, the properties of chocolate paste are not static. Its viscosity is affected by factors such as raw material composition and storage time, and temperature and bubble density can also vary due to fluctuations in the production environment. For example, in high-temperature and high-humidity environments, paste viscosity may decrease, increasing bubble formation. Different batches of raw materials may contain varying proportions of ingredients such as cocoa butter, leading to viscosity variations. Because traditional methods cannot sense these dynamic changes in real time and adjust control parameters accordingly, incomplete defoaming often occurs, and the shape and size of the molded chocolate are difficult to maintain, seriously affecting the product's appearance and internal quality. These issues severely hinder the intelligent upgrade of chocolate production. A full-process optimization solution that integrates multimodal sensing, data-driven decision-making, and digital twin simulation is urgently needed to overcome the bottlenecks of control accuracy and energy efficiency. Summary of the Invention

[0003] The object of the present invention is to provide an intelligent control system and method for a chocolate production process to solve the problems raised in the above background technology.

[0004] In order to solve the above technical problems, the present invention provides the following technical solutions: an intelligent control system and method for a chocolate production process, comprising an intelligent defoaming and molding module, a sensor network, an intelligent control unit, a big data analysis platform, and a human-computer interaction interface; the intelligent defoaming and molding module is used for intelligent and precise control of the defoaming and molding process of the chocolate slurry; the sensor network is used for real-time collection of temperature, humidity, pressure, viscosity during the chocolate production process, as well as shape and size information during the molding process; the intelligent control unit is used for receiving data from the sensor network, analyzing and processing the data, and then issuing control instructions; the big data analysis platform deploys a digital twin model for simulating and predicting production data and generating optimization strategies; the human-computer interaction interface is used to provide an operation interface to facilitate personnel to monitor and manage the production line, and data is transmitted between modules via high-speed data lines and wireless communications.

[0005] According to the above technical solution, the intelligent defoaming and molding module includes a viscosity sensor, a vibration rod and an infrared vision sensor. The viscosity sensor is used to measure the viscosity range of the chocolate slurry; the infrared vision sensor is used to detect the bubble density in real time and capture the shape and size information during the molding process; the vibration rod is a variable frequency vibration rod, and the vibration frequency range can be adjusted in real time according to the slurry viscosity, and the adjustment range is 100Hz to 1000Hz; the viscosity sensor is installed on the chocolate slurry delivery pipeline, and the vibration rod is installed under the molding mold. The various components are connected via high-speed data lines.

[0006] According to the above technical solution, the sensor network also includes a temperature sensor, a humidity sensor, and a pressure sensor. The temperature sensor, humidity sensor, and pressure sensor are distributed on the chocolate production line to comprehensively monitor production environment parameters. The intelligent control unit is used to calculate the optimal vibration frequency and molding control parameters based on sensor data analysis, and issue control instructions to control the vibration rod to adjust the vibration frequency and control the molding mold to adjust the position and pressure parameters.

[0007] According to the above technical solution, the big data analysis platform is used to conduct in-depth analysis of production data, explore potential problems and optimization points in the production process, and provide a basis for system optimization; the human-computer interaction interface provides an intuitive and easy-to-use operation interface, supporting personnel to conduct real-time monitoring, parameter adjustment and optimization management of the production line.

[0008] According to the above technical solution, the intelligent control method for a chocolate production process of the intelligent control system for a chocolate production process includes the following steps:

[0009] Step S1: The sensor network collects various parameters of the chocolate production process in real time, including slurry viscosity, temperature, humidity, pressure, and shape and size information during the molding process;

[0010] Step S2: The intelligent control unit receives the data collected by the sensor network, analyzes and processes it through a preset intelligent control algorithm, and calculates the optimal vibration frequency and molding control parameters;

[0011] Step S3: The intelligent control unit issues a control instruction based on the calculation result, controls the vibration rod to adjust the vibration frequency, and controls the forming mold to adjust the position and pressure parameters;

[0012] Step S4: The production process is monitored in real time through the human-computer interaction interface. Operators can adjust and optimize control parameters according to actual conditions. At the same time, the big data analysis platform stores, analyzes and mines production data to provide support for continuous optimization of the system.

[0013] According to the above technical solution, the vibration frequency adjustment calculation expression in the intelligent control algorithm is:

[0014]

[0015] Wherein, μ is the viscosity of the slurry, the unit is "cP", which is used to measure the fluidity of the chocolate slurry. The higher the viscosity, the worse the fluidity, and the vibration frequency needs to be reduced to avoid structural resonance; ρ is the bubble density detected by the infrared vision sensor, the unit is "cells / cm 2 ", used to reflect the density of bubbles in the slurry. The higher the density, the higher the vibration frequency needs to be. Prevent high-frequency vibration from consuming excessive energy; T is the slurry temperature, unit "℃", temperature affects the slurry viscosity, and rising temperature usually reduces the viscosity; T0 is the slurry reference temperature, which is used to standardize the effect of temperature on viscosity; μ0 is the slurry viscosity threshold, when the actual viscosity exceeds this value, the Sigmoid function in the denominator suppresses high-frequency vibration to avoid equipment overload; f is the vibration frequency, unit Hz, which is used to control the operating frequency of the vibration rod. The higher the frequency, the faster the bubble elimination speed, but the energy consumption increases; k1, k2, and k3 are the weights of adjusting the influence of viscosity, bubble density, and temperature on the vibration frequency, which are obtained by the big data analysis platform through historical data optimization. α is the adjustment factor, which is used to control the slope of the Sigmoid function and determines the steepness of the frequency change near the viscosity threshold.

[0016] According to the above technical solution, the forming control algorithm in the intelligent control algorithm includes the following steps:

[0017] Step A: Obtain the chocolate outline image through the infrared vision sensor, use the improved U-Net model to segment the outline, and output the boundary point coordinate set {(x i ,y i )}, perform contour extraction;

[0018] Step B: Calculate the shape error by defining a shape error function, where the shape error function is defined as: Where ∈ is the shape error, in units of "mm", which is used to quantify the deviation between the actual contour and the standard contour. The smaller the value, the higher the shape accuracy. N is the number of contour points, which represents the total number of sampling points extracted from the chocolate contour. i ,y i is the coordinate of the actual contour point, obtained by segmentation using the infrared vision sensor and the U-Net model. The coordinates of the preset ideal chocolate shape are used as the basis for error calculation;

[0019] Step C: Dynamically adjust the mold pressure P according to the error ∈, where P = P0 + γ·tanh(∈ / ∈0); P is the target mold pressure, in MPa, which controls the pressure applied by the molding mold to the slurry. The greater the pressure, the faster the molding speed, but too high a pressure will cause structural damage; P0 is the reference pressure, γ is the error gain coefficient, which is used to control the sensitivity of the shape error to the pressure adjustment. The larger the value, the more significant the effect of the error on the pressure. ∈0 is the error normalization factor, which is used to scale the actual error ∈ to a reasonable range to prevent calculation overflow caused by excessive values; in the formula, the pressure is dynamically adjusted according to the shape error: the larger the error, the higher the pressure increase, but the maximum increase is limited by the tanh function.

[0020] According to the above technical solution, the system continuously collects and analyzes data during the production process and updates the dynamic weight coefficients k1, k2, k3 through an adaptive learning algorithm. The adaptive learning algorithm adopts the DRL framework and specifically includes the following steps:

[0021] Step u: Define a multi-objective reward function R = ω1·(1-defect rate)-ω1·energy consumption+ω1·production efficiency; where ω1, ω2, and ω3 are weight coefficients, representing the optimization priorities of defect rate, energy consumption, and production efficiency, respectively, and ω1+ω2+ω3=1, where defect rate is the percentage of unqualified chocolate in total production, energy consumption is the electricity consumption per unit mass of chocolate, and production efficiency is the output per unit time;

[0022] Step v: Build the Actor-Critic network:

[0023] Actor network: inputs the real-time production state parameter μ,ρ,T,∈,P, and outputs the adjustment of dynamic weight coefficients Δk1, Δk2, Δk3;

[0024] Critic network: evaluates the long-term reward value R of the current coefficient combination and generates policy gradients to optimize the Actor network parameters;

[0025] Step w: Dynamically update the formula Among them, R is the reward value, which is used to comprehensively evaluate the quality of production status and guide the optimization direction of the DRL model. The larger the value, the better the overall performance of the system; η is the learning rate, which is used to control the parameter update step size. The weight coefficient k for the reward function i The partial derivative of .

[0026] Step x: When the chocolate surface defect rate is ≤0.5% and the energy consumption fluctuation range is ≤±5% in 10 consecutive batches of production, the parameters are determined to be converged and the coefficients k1, k2, and k3 are locked.

[0027] Compared with the existing technology, the beneficial effects achieved by the present invention are as follows: the present invention breaks through the limitations of traditional technology by constructing a multimodal sensor network and an intelligent control unit. It can utilize multiple parameters such as fusion viscosity, bubble density, and temperature to dynamically calculate the optimal vibration frequency through a nonlinear function, avoiding the one-sidedness of single parameter control; based on the improved U-Net model and shape error function, it realizes pixel-level recognition of contour deviation and dynamic adjustment of mold pressure, significantly improving molding accuracy; introduces a deep reinforcement learning (DRL) framework, continuously optimizes control parameters through a multi-objective reward function, and balances defect rate, energy consumption, and production efficiency; deploys a digital twin simulation model to perform real-time analysis and prediction of production data, providing data-driven decision support for process optimization. This effectively solves the problems of control lag, insufficient precision, and excessive energy consumption in the existing chocolate production process, and provides key technical support for the intelligent and green upgrade of the industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention.

[0029] In the attached figure:

[0030] Figure 1 This is a schematic diagram of the structure of an intelligent control system for chocolate production process according to the present invention;

[0031] Figure 2 This is a schematic diagram of an intelligent control method for chocolate production process according to the present invention. DETAILED DESCRIPTION

[0032] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0033] Example 1

[0034] See also Figure 1The present invention provides a technical solution: an intelligent control system for a chocolate production process, comprising an intelligent defoaming and molding module, a sensor network, an intelligent control unit, a big data analysis platform, and a human-computer interaction interface; the intelligent defoaming and molding module is used for intelligent and precise control of the defoaming and molding process of the chocolate slurry; the sensor network is used to collect temperature, humidity, pressure, viscosity, and shape and size information of the chocolate production process in real time; the intelligent control unit is used to receive data from the sensor network, analyze and process the data, and then issue control instructions; the big data analysis platform deploys a digital twin model for simulating and predicting production data and generating optimization strategies; the human-computer interaction interface is used to provide an operation interface to facilitate personnel to monitor and manage the production line, and data is transmitted between modules via high-speed data lines and wireless communications.

[0035] The intelligent defoaming and molding module includes a viscosity sensor, a vibration rod and an infrared vision sensor. The viscosity sensor is used to measure the viscosity range of the chocolate slurry; the infrared vision sensor is used to detect the bubble density in real time and capture the shape and size information during the molding process; the vibration rod is a variable frequency vibration rod, and the vibration frequency range can be adjusted in real time according to the slurry viscosity, with an adjustment range of 100Hz to 1000Hz; the viscosity sensor is installed on the chocolate slurry delivery pipeline, and the vibration rod is installed under the molding mold. The various components are connected via high-speed data lines.

[0036] The sensor network also includes temperature sensors, humidity sensors and pressure sensors, which are distributed on the chocolate production line to comprehensively monitor production environment parameters; the intelligent control unit is used to calculate the optimal vibration frequency and molding control parameters based on sensor data analysis, and issue control instructions to control the vibration rod to adjust the vibration frequency and the molding mold to adjust the position and pressure parameters.

[0037] The big data analysis platform is used to conduct in-depth analysis of production data, explore potential problems and optimization points in the production process, and provide a basis for system optimization; the human-computer interaction interface provides an intuitive and easy-to-use operation interface, supporting personnel to conduct real-time monitoring, parameter adjustment and optimization management of the production line. Operators can intervene and adjust the production process through the human-computer interaction interface based on the analysis results provided by the big data analysis platform, thereby realizing dynamic optimization of the production process.

[0038] Example 2

[0039] See also Figure 2 This embodiment discloses an intelligent control method for a chocolate production process, comprising the following steps:

[0040] Step S1: The sensor network collects various parameters of the chocolate production process in real time, including slurry viscosity, temperature, humidity, pressure, and shape and size information during the molding process;

[0041] Step S2: The intelligent control unit receives the data collected by the sensor network, analyzes and processes it through a preset intelligent control algorithm, and calculates the optimal vibration frequency and molding control parameters;

[0042] Step S3: The intelligent control unit issues a control instruction based on the calculation result, controls the vibration rod to adjust the vibration frequency, and controls the forming mold to adjust the position and pressure parameters;

[0043] Step S4: The production process is monitored in real time through the human-computer interaction interface. Operators can adjust and optimize control parameters according to actual conditions. At the same time, the big data analysis platform stores, analyzes and mines production data to provide support for continuous optimization of the system.

[0044] The calculation expression for vibration frequency adjustment in the intelligent control algorithm is:

[0045]

[0046] Wherein, μ is the viscosity of the slurry, the unit is "cP", which is used to measure the fluidity of the chocolate slurry. The higher the viscosity, the worse the fluidity, and the vibration frequency needs to be reduced to avoid structural resonance; ρ is the bubble density detected by the infrared vision sensor, the unit is "cells / cm 2 ", used to reflect the density of bubbles in the slurry. The higher the density, the higher the vibration frequency needs to be. Prevent excessive energy consumption by high-frequency vibration; T is the slurry temperature, unit "℃". Temperature affects the slurry viscosity. Rising temperature usually reduces the viscosity. T0 is the slurry base temperature, which is used to standardize the effect of temperature on viscosity. μ0 is the slurry viscosity threshold. When the actual viscosity exceeds this value, the Sigmoid function in the denominator suppresses high-frequency vibration to avoid equipment overload. f is the vibration frequency, unit Hz, which is used to control the operating frequency of the vibration rod. The higher the frequency, the faster the bubble elimination speed, but the energy consumption increases. k1, k2, and k3 are the weights of adjusting the influence of viscosity, bubble density, and temperature on the vibration frequency, respectively. They are obtained by the big data analysis platform through historical data optimization. α is the adjustment factor, which is used to control the slope of the Sigmoid function and determines the steepness of the frequency change near the viscosity threshold.

[0047] In the formula, viscosity, bubble density, and temperature parameters are introduced through multimodal data fusion, and coupled through inverse, square root, and logarithmic functions, so as to better fit the laws of fluid mechanics. Viscosity threshold segmented control is achieved through temperature compensation, which effectively avoids the problems of incomplete defoaming and excessive energy consumption caused by traditional vibration defoaming control that only relies on a single parameter of viscosity and ignores the spatial distribution of bubble density and the influence of temperature on slurry fluidity.

[0048] The forming control algorithm in the intelligent control algorithm includes the following steps:

[0049] Step A: Obtain the chocolate outline image through the infrared vision sensor, use the improved U-Net model to segment the outline, and output the boundary point coordinate set {(x i ,y i )}, perform contour extraction;

[0050] Step B: Calculate the shape error by defining a shape error function, where the shape error function is defined as: Where ∈ is the shape error, in units of "mm", which is used to quantify the deviation between the actual contour and the standard contour. The smaller the value, the higher the shape accuracy. N is the number of contour points, which represents the total number of sampling points extracted from the chocolate contour. i ,y i is the coordinate of the actual contour point, obtained by segmentation using the infrared vision sensor and the U-Net model. The coordinates of the preset ideal chocolate shape are used as the basis for error calculation. By comparing the coordinates of the actual contour with the standard contour point by point and accumulating the Euclidean distance of all points, the shape accuracy is comprehensively evaluated. The improved U-Net algorithm adds a channel attention mechanism to improve the accuracy of small-scale defect recognition and effectively avoid the problem of traditional simple contour matching that cannot quantify local deformation.

[0051] Step C: Dynamically adjust the mold pressure P according to the error ∈, where P = P0 + γ·tanh(∈ / ∈0); where P is the target mold pressure, in units of "MPa", which controls the pressure applied by the molding mold to the slurry. The greater the pressure, the faster the molding speed, but too high a pressure will cause structural damage; P0 is the reference pressure, γ is the error gain coefficient, which is used to control the sensitivity of the shape error to the pressure adjustment. The larger the value, the more significant the impact of the error on the pressure. ∈0 is the error normalization factor, which is used to scale the actual error ∈ to a reasonable range to prevent the calculation overflow caused by excessive values; in the formula, the pressure is dynamically adjusted according to the shape error: the larger the error, the higher the pressure increase, but the maximum increase is limited by the tanh function; in the formula, the hyperbolic tangent function constraint is used to map the error ∈ to the [-1,1] interval, limit the pressure adjustment amplitude, prevent overshoot, and balance accuracy and energy consumption through digital twin simulation calibration, so that the pressure adjustment and deformation degree are dynamically related to avoid excessive pressure causing mold wear or slurry splashing.

[0052] The system continuously collects and analyzes data during the production process and updates the dynamic weight coefficients k1, k2, and k3 through an adaptive learning algorithm. The adaptive learning algorithm uses the DRL framework and specifically includes the following steps:

[0053] Step u: Define a multi-objective reward function R = ω1·(1-defect rate)-ω1·energy consumption+ω1·production efficiency; where ω1, ω2, and ω3 are weight coefficients, representing the optimization priorities of defect rate, energy consumption, and production efficiency, respectively, and ω1+ω2+ω3=1, where defect rate is the percentage of unqualified chocolate in total production, energy consumption is the electricity consumption per unit mass of chocolate, and production efficiency is the output per unit time;

[0054] Step v: Build the Actor-Critic network:

[0055] Actor network: inputs the real-time production state parameter μ,ρ,T,∈,P, and outputs the adjustment of dynamic weight coefficients Δk1, Δk2, Δk3;

[0056] Critic network: evaluates the long-term reward value R of the current coefficient combination and generates policy gradients to optimize the Actor network parameters;

[0057] Step w: Dynamically update the formula Among them, R is the reward value, which is used to comprehensively evaluate the quality of production status and guide the optimization direction of the DRL model. The larger the value, the better the overall performance of the system; η is the learning rate, which is used to control the parameter update step size. The weight coefficient k for the reward function i The partial derivative of .

[0058] Step x: When the chocolate surface defect rate is ≤0.5% and the energy consumption fluctuation range is ≤±5% in 10 consecutive batches of production, the parameters are determined to be convergent and the coefficients k1, k2, and k3 are locked;

[0059] This application breaks through the limitations of traditional technology by constructing a multimodal sensor network and an intelligent control unit. It can utilize multiple parameters such as fusion viscosity, bubble density, and temperature to dynamically calculate the optimal vibration frequency through a nonlinear function, thus avoiding the one-sidedness of single parameter control; based on the improved U-Net model and shape error function, it realizes pixel-level recognition of contour deviation and dynamic adjustment of mold pressure, significantly improving molding accuracy; introduces a deep reinforcement learning (DRL) framework, continuously optimizes control parameters through a multi-objective reward function, and balances defect rate, energy consumption, and production efficiency; deploys a digital twin simulation model to perform real-time analysis and prediction of production data, providing data-driven decision support for process optimization. This effectively solves the problems of control lag, insufficient precision, and excessive energy consumption in the existing chocolate production process, and provides key technical support for the intelligent and green upgrade of the industry.

[0060] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0061] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0062] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 The steps for the function specified in one or more boxes.

[0063] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present invention, ordinary technicians in this field can also make many forms without departing from the scope of protection of the purpose of the present invention and the claims, which are all protected by the present invention.

Claims

1. An intelligent control system for chocolate production process, characterized by: It includes an intelligent defoaming and molding module, a sensor network, an intelligent control unit, a big data analysis platform and a human-computer interaction interface; the intelligent defoaming and molding module is used for intelligent and precise control of the defoaming and molding process of chocolate slurry; the sensor network is used to collect temperature, humidity, pressure, viscosity during the chocolate production process and shape and size information during the molding process in real time; the intelligent control unit is used to receive data from the sensor network, analyze and process it, and then issue control instructions; the big data analysis platform deploys a digital twin model to simulate and predict production data and generate optimization strategies; the human-computer interaction interface is used to provide an operation interface to facilitate personnel to monitor and manage the production line, and data is transmitted between modules through high-speed data lines and wireless communications.

2. The intelligent control system for chocolate production process according to claim 1, characterized in that: The intelligent defoaming and molding module includes a viscosity sensor, a vibration rod, and an infrared vision sensor. The viscosity sensor is used to measure the viscosity range of the chocolate slurry; the infrared vision sensor is used to detect bubble density in real time and capture shape and size information during the molding process; the vibration rod is a variable-frequency vibration rod, and the vibration frequency range can be adjusted in real time according to the slurry viscosity, with an adjustment range of 100Hz to 1000Hz. The viscosity sensor is installed on the chocolate slurry delivery pipeline, and the vibration rod is installed below the molding mold. All components are connected via high-speed data lines.

3. The intelligent control system for chocolate production process according to claim 1, characterized in that: The sensor network also includes temperature sensors, humidity sensors, and pressure sensors, which are distributed throughout the chocolate production line to comprehensively monitor production environment parameters. The intelligent control unit is used to analyze and calculate the optimal vibration frequency and molding control parameters based on sensor data, and issue control instructions to control the vibration rod to adjust the vibration frequency and the molding mold to adjust the position and pressure parameters.

4. The intelligent control system for chocolate production process according to claim 1, characterized in that: The big data analysis platform is used to conduct in-depth analysis of production data, explore potential problems and optimization points in the production process, and provide a basis for system optimization; the human-computer interaction interface provides an intuitive and easy-to-use operation interface, supporting personnel to conduct real-time monitoring, parameter adjustment and optimization management of the production line.

5. An intelligent control method for a chocolate production process based on the intelligent control system for a chocolate production process according to any one of claims 1 to 4, characterized in that: The following steps are involved: Step S1: The sensor network collects various parameters of the chocolate production process in real time, including slurry viscosity, temperature, humidity, pressure, and shape and size information during the molding process; Step S2: The intelligent control unit receives the data collected by the sensor network, analyzes and processes it through a preset intelligent control algorithm, and calculates the optimal vibration frequency and molding control parameters; Step S3: The intelligent control unit issues a control instruction based on the calculation result, controls the vibration rod to adjust the vibration frequency, and controls the forming mold to adjust the position and pressure parameters; Step S4: The production process is monitored in real time through the human-computer interaction interface. Operators can adjust and optimize control parameters according to actual conditions. At the same time, the big data analysis platform stores, analyzes and mines production data to provide support for continuous optimization of the system.

6. The method for intelligent control of chocolate production process according to claim 5, characterized in that: The vibration frequency adjustment calculation expression in the intelligent control algorithm is: Where μ is the viscosity of the slurry, in cP, which is used to measure the fluidity of the chocolate slurry. The higher the viscosity, the worse the fluidity, and the lower the vibration frequency needs to be to avoid structural resonance. ρ is the bubble density detected by the infrared vision sensor, in cells / cm 2 ", used to reflect the density of bubbles in the slurry. The higher the density, the higher the vibration frequency needs to be. Prevent excessive energy consumption by high-frequency vibration; T is the slurry temperature, unit "℃". Temperature affects the slurry viscosity. Rising temperature usually reduces the viscosity; T0 is the slurry base temperature, used to standardize the effect of temperature on viscosity; μ0 is the slurry viscosity threshold. When the actual viscosity exceeds this value, the Sigmoid function in the denominator suppresses high-frequency vibration to avoid equipment overload; f is the vibration frequency, unit Hz, used to control the operating frequency of the vibration rod. The higher the frequency, the faster the bubble elimination speed, but the energy consumption increases; k1, k2, and k3 are the weights of adjusting the influence of viscosity, bubble density, and temperature on the vibration frequency, respectively. They are obtained by the big data analysis platform through historical data optimization. α is the adjustment factor, used to control the slope of the Sigmoid function and determine the steepness of the frequency change near the viscosity threshold.

7. The method for intelligent control of chocolate production process according to claim 5, characterized in that: The forming control algorithm in the intelligent control algorithm includes the following steps: Step A: Obtain the chocolate outline image through the infrared vision sensor, use the improved U-Net model to segment the outline, and output the boundary point coordinate set {(x i ,y i )}, perform contour extraction; Step B: Calculate the shape error by defining a shape error function, where the shape error function is defined as: Where ∈ is the shape error, in "mm", which is used to quantify the deviation between the actual contour and the standard contour. The smaller the value, the higher the shape accuracy. N is the number of contour points, which represents the total number of sampling points extracted from the chocolate contour. i ,y i is the coordinate of the actual contour point, obtained by segmentation using the infrared vision sensor and the U-Net model. The coordinates of the preset ideal chocolate shape are used as the basis for error calculation; Step C: Dynamically adjust the mold pressure P according to the error ∈, where P = P0 + γ·tanh(∈ / ∈0); P is the target mold pressure, in MPa, which controls the pressure applied by the molding mold to the slurry. The higher the pressure, the faster the molding speed, but too high a pressure can cause structural damage. P0 is the reference pressure, and γ is the error gain coefficient, which is used to control the sensitivity of the shape error to pressure adjustment. A larger value indicates a more significant impact of the error on the pressure. ∈0 is the error normalization factor, which is used to scale the actual error ∈ to a reasonable range to prevent calculation overflow caused by excessive values. In the formula, the pressure is dynamically adjusted according to the shape error: the larger the error, the higher the pressure increase, but the maximum increase is limited by the tanh function.

8. The method for intelligent control of chocolate production process according to claim 6, characterized in that: The system continuously collects and analyzes data during the production process and updates the dynamic weight coefficients k1, k2, and k3 through an adaptive learning algorithm. The adaptive learning algorithm adopts the DRL framework and specifically includes the following steps: Step u: Define a multi-objective reward function R = ω1·(1-defect rate)-ω1·energy consumption+ω1·production efficiency; where ω1, ω2, and ω3 are weight coefficients, representing the optimization priorities of defect rate, energy consumption, and production efficiency, respectively, and ω1+ω2+ω3=1, where defect rate is the percentage of unqualified chocolate in total production, energy consumption is the electricity consumption per unit mass of chocolate, and production efficiency is the output per unit time; Step v: Build the Actor-Critic network: Actor network: inputs the real-time production state parameter μ,ρ,T,∈,P, and outputs the adjustment of dynamic weight coefficients Δk1, Δk2, Δk3; Critic network: evaluates the long-term reward value R of the current coefficient combination and generates policy gradients to optimize the Actor network parameters; Step w: Dynamically update the formula Among them, R is the reward value, which is used to comprehensively evaluate the quality of production status and guide the optimization direction of the DRL model. The larger the value, the better the overall performance of the system; η is the learning rate, which is used to control the parameter update step size. The weight coefficient k for the reward function i The partial derivative of Step x: When the chocolate surface defect rate is ≤0.5% and the energy consumption fluctuation range is ≤±5% in 10 consecutive batches of production, the parameters are determined to be converged and the coefficients k1, k2, and k3 are locked.

Images