Intelligent microwave-heat exchange composite milk sterilization system and control method thereof

Abstract

The invention relates to the technical field of milk sterilization systems, in particular to an intelligent microwave-heat exchange composite milk sterilization system and a method thereof.The system comprises a mixing module used for adjusting the flow speed and temperature of fresh milk; the microwave sterilization module is used for heating milk by using a multi-frequency magnetic field; the temperature control module comprises a microwave power regulator and a heat exchanger and is used for regulating the microwave heating time and the heat exchange rate so as to control the milk temperature; and the digital twin module optimizes the sterilization process by using deep reinforcement learning and simulates the system performance in a virtual environment, and a faster and more uniform heating process is realized through the synergistic effect of microwaves and heat exchange. The application of the multi-frequency magnetic field technology further improves the uniformity of microwave heating, and effectively solves the problems of hot spots and cold spots in traditional heat exchange sterilization. Experimental results show that compared with a traditional method, the sterilization effect is improved by about 11.5%, which means higher food safety.

Description

Technical Field

[0001] The invention relates to the technical field of milk sterilization systems, in particular to an intelligent microwave-heat exchange composite milk sterilization system and a method thereof. Background Art

[0002] As people pay more and more attention to food safety and nutritional health, milk sterilization technology has always been a hot research topic in the dairy processing industry. Traditional milk sterilization methods mainly rely on heat exchange technology, which heats milk to a specific temperature and keeps it for a certain period of time to achieve the purpose of sterilization. Although this method can effectively kill harmful microorganisms, it also has some defects that are difficult to overcome.

[0003] First, traditional heat exchange sterilization often requires a long heating time, which inevitably leads to varying degrees of loss of nutrients in milk. In particular, vitamins and certain proteins with high heat sensitivity are prone to denaturation or degradation during long-term heat treatment. Secondly, the temperature distribution of heat exchange sterilization is often not uniform enough, and hot spots or cold spots are prone to occur, which not only affects the sterilization effect, but may also cause local overheating, further exacerbating nutrient loss. In addition, the energy efficiency of traditional sterilization systems is low. Under the current increasingly severe energy and environmental pressures, more energy-saving and environmentally friendly sterilization technologies are urgently needed.

[0004] In recent years, microwave sterilization technology has attracted widespread attention due to its fast and uniform heating characteristics. However, simple microwave sterilization also faces some challenges. For example, the depth of microwave heating is limited, making it difficult to ensure uniform heating of large volumes of milk; in addition, precise control of microwave heating is also a technical problem, which can easily lead to local overheating.

[0005] On the other hand, with the rapid development of artificial intelligence and Internet of Things technologies, the concept of intelligent manufacturing is gradually penetrating into the food processing industry. However, most of the milk sterilization systems on the market are still at the simple PID control stage, which makes it difficult to adapt to the complex and changing production environment, and it is even more impossible to achieve continuous optimization of the sterilization process.

[0006] In view of this, there is an urgent need for a new milk sterilization system that can comprehensively utilize the advantages of microwave and heat exchange technologies and combine them with intelligent control strategies. The system should be able to ensure the sterilization effect while maximizing the protection of milk nutrients, improving energy efficiency, and having self-learning and optimization capabilities. Summary of the invention

[0007] The present invention is an innovative solution to the above technical problems. By organically combining microwave sterilization with heat exchange technology and introducing deep reinforcement learning algorithms and digital twin technology, the intelligent microwave-heat exchange composite milk sterilization system of the present invention has achieved multiple breakthroughs in sterilization effect, nutritional protection, energy efficiency and intelligent control.

[0008] The present invention proposes an intelligent microwave-heat exchange composite milk sterilization system, which comprises: A mixing module to regulate the flow rate and temperature of raw milk; Microwave sterilization module, which uses multi-frequency magnetic fields to heat milk; A temperature control module, including a microwave power regulator and a heat exchanger, for adjusting microwave heating time and heat exchange rate to control milk temperature; A digital twin module that uses deep reinforcement learning to optimize sterilization processes and simulate system performance in a virtual environment; Among them, the core formula of the deep reinforcement learning optimization algorithm is:

[0009] In the formula, is the state-action value function, s is the current state, a is the current action, α is the learning rate and 0<α≤1, is the immediate reward, γ is the discount factor and 0≤ <1, For the next state, For the next action.

[0010] Preferably, the mixing module comprises: Flow rate regulator, which adjusts the milk flow by changing the flow rate; Thermostats, which raise or lower the temperature by cooling or heating water; Flow sensor, used to measure flow rate; Temperature sensor, used to measure temperature; A flow controller for controlling the pump drive; A mixing chamber to balance flow rate and temperature changes; Wherein, the flow rate regulator adopts the following control algorithm:

[0011] In the formula, is the adjusted flow rate, is the initial flow rate, is the error between the target flow rate and the actual flow rate, , , are the proportional, integral, and differential coefficients respectively.

[0012] Preferably, the microwave sterilization module comprises: A microwave generator, used for generating microwaves; A microwave power regulator, used to control microwave power; Power meter, used to measure microwave power; a microwave cavity for containing the milk to be sterilized; Wherein, the microwave power regulator adopts the following adaptive control algorithm:

[0013] In the formula, is the microwave power at time t, is the initial microwave power, is the maximum power adjustment, τ is the time constant, K is the temperature feedback coefficient, is the target temperature, is the actual temperature at time t.

[0014] Preferably, the microwave power regulator of the temperature control module works in coordination with the heat exchanger to accurately adjust the temperature of the milk by controlling the microwave heating time and the heat exchange rate, and the coordinated control adopts the following optimization algorithm:

[0015] Subject to:

[0016]

[0017]

[0018]

[0019] In the formula, To optimize the objective function, is the temperature at time t, is the target temperature, is the microwave power at time t, is the heat exchange rate at time t, , , is the weight coefficient, is the temperature change rate function, and For temperature limit, is the maximum microwave power, is the maximum heat exchange rate.

[0020] Preferably, the digital twin module uses the following deep reinforcement learning method to optimize the sterilization process:

[0021] Among them, θ is the policy network parameter, α is the learning rate, is the policy gradient, and:

[0022] In the formula, is the parameterized policy function, γ is the discount factor, is the reward at time t.

[0023] The control method of the intelligent microwave-heat exchange composite milk sterilization system comprises the following steps: a) regulating the flow rate and temperature of the fresh milk through the mixing module; b) heating the milk using a multi-frequency magnetic field of a microwave sterilization module; c) cooperatively controlling the temperature of the milk through the microwave power regulator and the heat exchanger of the temperature control module; d) Optimize sterilization process parameters using deep reinforcement learning methods of digital twin modules; Wherein, step d) adopts the following double Q network algorithm:

[0024] In the formula, Q is the main network, Q is the target Q network, and is the network parameter, r is the immediate reward, γ is the discount factor, s and a are the current state and action, s' and a' are the next state and action, and the target network is updated every fixed number of steps: ← .

[0025] Preferably, the method further comprises the following steps: e) Use neural network model to analyze the nutritional components in fresh milk and provide reference for microwave sterilization time; f) Based on the Kalman filter algorithm, predict the nutritional components of milk at each moment; g) Use the predicted nutrients to match the optimal sterilization time and achieve precise sterilization; Among them, the neural network model adopts the following attention mechanism:

[0026] Where Q is the query matrix, K is the key matrix, and V is the value matrix. is the dimension of the key vector.

[0027] Preferably, the method further comprises the following steps: h) Control the mixing module, microwave sterilization module and temperature control module through the edge computing unit; i) Use Kalman filter algorithm to optimize control signals and improve system accuracy and efficiency; The edge computing unit adopts the following distributed optimization algorithm:

[0028] st In the formula, is the global objective function, is the local objective function of the ith edge node, is the decision variable of the ith edge node, and c is the global constraint.

[0029] Preferably, the method further comprises the following steps: j) Collect data generated by the mixing module, microwave sterilization module, and temperature control module; k) Analyze and store the collected data through the processor of the digital twin module; l) Training sterilization process parameters in a virtual environment; m) passing the optimized sterilization process parameters to the digital twin model for further optimization; The digital twin model adopts the following generative adversarial network (GAN) structure:

[0030] In the formula, D is the discriminator, G is the generator, is the real data distribution, is the noise distribution, x is the real data sample, and z is the noise sample.

[0031] Preferably, the method further comprises the following steps: n) Use deep reinforcement learning methods to combine and control the mixing module, microwave sterilization module, and temperature control module; o) Continuously optimize the sterilization process in online production to achieve intelligent sterilization; The deep reinforcement learning method adopts the following hierarchical reinforcement learning algorithm:

[0032] In the formula, is the total Q value function, is the high-level Q-value function used to select the sub-goal g, It is a low-level Q-value function used to select a specific action a, s is the current state, g is the sub-goal, and a is the specific action.

[0033] The beneficial effects of the present invention are mainly reflected in the following aspects: First, in terms of sterilization effect, the present invention achieves a faster and more uniform heating process through the synergistic effect of microwaves and heat exchange. The application of multi-frequency magnetic field technology further improves the uniformity of microwave heating and effectively solves the hot and cold spot problems existing in traditional heat exchange sterilization. The experimental results show that compared with the traditional method, the sterilization effect of the present invention is improved by about 11.5%, which means higher food safety.

[0034] Secondly, in terms of nutritional protection, the intelligent control system of the present invention can accurately control the heating time and temperature, minimizing the loss of heat-sensitive nutrients. Experimental data show that the protein denaturation rate of the present invention is reduced by about 39.5% compared with the traditional method, and the vitamin C retention rate is increased by about 8.2%. This achievement is of great significance for the production of high-quality, high-nutrition value milk products.

[0035] In terms of energy efficiency, the system of the present invention significantly improves energy utilization through the optimized combination of microwaves and heat exchange, as well as the application of intelligent control strategies. Compared with traditional methods, it can save about 19% of energy consumption for every 1000L of milk processed. This not only reduces production costs, but also complies with the national policy orientation of energy conservation and emission reduction.

[0036] In terms of production efficiency, the present invention shortens the sterilization time by 37.5%, greatly improving the processing capacity of the production line, which is of great significance for meeting the growing market demand and improving the competitiveness of enterprises.

[0037] More importantly, the digital twin technology and deep reinforcement learning algorithm introduced in this invention give the system the ability to continuously optimize itself. By continuously simulating and optimizing the sterilization process in a virtual environment, the system can adapt to the changes in the characteristics of different batches of raw materials and always maintain the best sterilization effect and energy efficiency. This intelligent and adaptive control method marks an important step forward in the development of milk sterilization technology towards intelligent manufacturing.

[0038] In summary, the present invention has not only achieved a number of breakthroughs in terms of technology, but also integrated the innovative achievements of multidisciplinary knowledge such as microwave physics, thermodynamics, control engineering, and artificial intelligence. It provides a new intelligent solution for the dairy processing industry, which is expected to promote the overall improvement of the industry's technical level and provide consumers with safer, more nutritious, and healthier milk products. At the same time, its energy-saving and environmentally friendly characteristics will also make an important contribution to the sustainable development of the industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is the overall logic block diagram of the system of the present invention.

[0040] Figure 2 It is a logic block diagram of the hybrid module of the present invention.

[0041] Figure 3 It is a logic block diagram of the microwave sterilization module of the present invention.

[0042] Figure 4 It is a logic block diagram of the temperature control module of the present invention.

[0043] Figure 5 This is the edge computing control process of the present invention. DETAILED DESCRIPTION

[0044] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the specific implementation methods, structures, features and effects thereof are described in detail below in conjunction with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0045] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0046] See also Figure 1-5 The present invention relates to an intelligent microwave-heat exchange composite milk sterilization system and a control method thereof. The present invention will be described in detail below in conjunction with specific implementation methods.

[0047] First, the present invention provides an intelligent microwave-heat exchange composite milk sterilization system, which includes a mixing module 1, a microwave sterilization module 2, a temperature control module 3 and a digital twin module 4. The mixing module 1 is used to adjust the flow rate and temperature of fresh milk; the microwave sterilization module 2 uses a multi-frequency magnetic field to heat the milk; the temperature control module 3 includes a microwave power regulator 31 and a heat exchanger 32, which are used to adjust the microwave heating time and the heat exchange rate to control the milk temperature; the digital twin module 4 uses deep reinforcement learning to optimize the sterilization process and simulate the system performance in a virtual environment.

[0048] In a preferred embodiment of the present invention, the core formula of the deep reinforcement learning optimization algorithm adopted by the digital twin module 4 is:

[0049] in, is the state-action value function, s is the current state (such as milk temperature, flow rate, etc.), a is the current action (such as adjusting microwave power, heat exchange rate, etc.), α is the learning rate and 0 < α ≤ 1, r is the immediate reward (such as sterilization effect, energy consumption, etc.), γ is the discount factor and 0 ≤ γ < 1, For the next state, This algorithm enables the system to continuously learn and optimize the sterilization process, improving sterilization effect and energy efficiency.

[0050] Secondly, the present invention further optimizes the mixing module 1. The mixing module 1 includes a flow rate regulator 11, a thermostat 12, a flow sensor 13, a temperature sensor 14, a flow controller 15 and a mixing chamber 16. The flow rate regulator 11 adjusts the milk flow by changing the flow rate; the thermostat 12 raises or lowers the temperature by water cooling or heating; the flow sensor 13 is used to measure the flow rate; the temperature sensor 14 is used to measure the temperature; the flow controller 15 is used to control the pump drive; and the mixing chamber 16 is used to balance the flow rate and temperature changes.

[0051] In another embodiment of the present invention, the flow rate regulator 11 adopts the following control algorithm:

[0052] in, is the adjusted flow rate, is the initial flow rate, is the error between the target flow rate and the actual flow rate, , , The PID control algorithm can achieve accurate flow rate regulation and ensure that the milk reaches the optimal flow rate and temperature conditions before entering the microwave sterilization module 2.

[0053] Again, the present invention has made an innovative design to the microwave sterilization module 2. The microwave sterilization module 2 includes a microwave generator 21, a microwave power regulator 22, a power meter 23 and a microwave cavity 24. The microwave generator 21 is used to generate microwaves; the microwave power regulator 22 is used to control the microwave power; the power meter 23 is used to measure the microwave power; and the microwave cavity 24 is used to contain the milk to be sterilized.

[0054] In a preferred embodiment of the present invention, the microwave power regulator 22 adopts the following adaptive control algorithm:

[0055] in, is the microwave power at time t, is the initial microwave power (usually set to 300W), is the maximum power adjustment (can be set to 200W), τ is the time constant (can be set to 10s), K is the temperature feedback coefficient (can be set to 5W / ℃), is the target temperature (usually 72°C), is the actual temperature at time t. This adaptive control algorithm can dynamically adjust the microwave power according to the actual temperature change of milk, ensuring the sterilization effect while avoiding nutritional loss caused by overheating.

[0056] Next, the present invention conducts an in-depth study on the coordinated control of the temperature control module 3. The microwave power regulator 31 of the temperature control module 3 works in coordination with the heat exchanger 32 to accurately adjust the milk temperature by controlling the microwave heating time and the heat exchange rate. The coordinated control adopts the following optimization algorithm:

[0057] subject to:

[0058]

[0059]

[0060]

[0061] in, To optimize the objective function, is the temperature at time t, is the target temperature (usually 72°C), is the microwave power at time t, is the heat exchange rate at time t, , , is the weight coefficient (can be set to 1, 0.01, 0.01 respectively), is the temperature change rate function, and is the temperature limit (can be set to 70℃ and 75℃), is the maximum microwave power (can be set to 500W), is the maximum heat exchange rate (can be set to 100L / min). This optimization algorithm can minimize energy consumption and nutrient loss while ensuring the sterilization effect.

[0062] Then, the present invention innovatively designs the deep reinforcement learning method of the digital twin module 4. The digital twin module 4 uses the following deep reinforcement learning method to optimize the sterilization process:

[0063] Among them, θ is the policy network parameter, α is the learning rate (can be set to 0.001), is the policy gradient, and:

[0064] in, is a parameterized policy function, γ is a discount factor (can be set to 0.99), is the reward at time t (such as the weighted sum of sterilization effect, energy consumption, and nutrient retention rate). This deep reinforcement learning method can continuously optimize the sterilization process parameters in a virtual environment and provide optimization suggestions for actual production.

[0065] In addition, the present invention also provides a control method for an intelligent microwave-heat exchange composite milk sterilization system. The method comprises the following steps: a. adjusting the flow rate and temperature of fresh milk through a mixing module 1; b. heating the milk using a multi-frequency magnetic field of a microwave sterilization module 2; c. collaboratively controlling the temperature of the milk through a microwave power regulator 31 and a heat exchanger 32 of a temperature control module 3; d. optimizing the sterilization process parameters using a deep reinforcement learning method of a digital twin module 4.

[0066] In a preferred embodiment of the present invention, step d) adopts the following double Q network algorithm:

[0067] in, For the main Q network, is the target Q network, and are network parameters, r is the immediate reward, γ is the discount factor (can be set to 0.99), s and a are the current state and action, s' and a' are the next state and action, and the target network is updated every fixed number of steps (such as 1000 steps): ← This dual Q-network algorithm can improve the stability and convergence speed of deep reinforcement learning.

[0068] Next, the present invention also introduces a neural network model to analyze the nutritional components in fresh milk and provide a reference for the microwave sterilization time. The method includes the following steps: e. Analyzing the nutritional components in fresh milk using a neural network model; f. Predicting the nutritional components of milk at each moment based on the Kalman filter algorithm; g. Using the predicted nutritional components, matching the optimal sterilization time to achieve precise sterilization.

[0069] In another embodiment of the present invention, the neural network model adopts the following attention mechanism:

[0070] Among them, Q is the query matrix, K is the key matrix, and V is the value matrix. is the dimension of the key vector. This attention mechanism can better capture the relationship between milk components and improve the accuracy of nutritional analysis.

[0071] In addition, the present invention also introduces edge computing technology to improve the real-time performance and reliability of the system. The method includes the following steps: h Controlling the mixing module 1, the microwave sterilization module 2 and the temperature control module 3 through the edge computing unit; i Using the Kalman filter algorithm to optimize the control signal to improve the system accuracy and efficiency.

[0072] In a preferred embodiment of the present invention, the edge computing unit adopts the following distributed optimization algorithm:

[0073] st in, is the global objective function, is the local objective function of the ith edge node, is the decision variable of the ith edge node, and c is the global constraint. This distributed optimization algorithm can reduce network communication overhead and improve the real-time performance of the system while ensuring global optimality.

[0074] Finally, the present invention also makes an innovative design for the digital twin model. The method includes the following steps: j Collecting data generated by the mixing module 1, the microwave sterilization module 2, and the temperature control module 3; k Analyzing and storing the collected data through the processor of the digital twin module 4; l Training the sterilization process parameters in a virtual environment; m Passing the optimized sterilization process parameters to the digital twin model for further optimization.

[0075] In another embodiment of the present invention, the digital twin model adopts the following generative adversarial network (GAN) structure:

[0076] Among them, D is the discriminator, G is the generator, is the real data distribution, is the noise distribution, x is the real data sample, and z is the noise sample. This GAN structure can generate a more realistic virtual environment and improve the simulation accuracy of the digital twin model.

[0077] The method further comprises the following steps: n) Using deep reinforcement learning method to perform combined control on mixing module 1, microwave sterilization module 2 and temperature control module 3; o) Continuously optimize the sterilization process in online production to achieve intelligent sterilization; The deep reinforcement learning method adopts the following hierarchical reinforcement learning algorithm:

[0078] In the formula, is the total Q value function, is the high-level Q-value function used to select the sub-goal g, It is a low-level Q-value function used to select a specific action a, s is the current state, g is the sub-goal, and a is the specific action.

[0079] In summary, the intelligent microwave-heat exchange composite milk sterilization system and control method provided by the present invention realizes precise control and continuous optimization of the milk sterilization process through innovative algorithm design and module collaboration. The system can not only ensure the sterilization effect, but also protect the nutritional components of milk to the maximum extent, while significantly improving energy efficiency. This intelligent sterilization system is expected to promote technological progress in the dairy processing industry and provide consumers with safer and more nutritious milk products.

[0080] In order to verify the superiority of the intelligent microwave-heat exchange composite milk sterilization system of the present invention, the present invention has carried out a series of comparative experiments. The specific experimental process and result analysis of the embodiment and comparative example will be described in detail below.

[0081] Embodiment 1: In this embodiment, the intelligent microwave-heat exchange composite milk sterilization system of the present invention is used. The system includes a mixing module 1, a microwave sterilization module 2, a temperature control module 3 and a digital twin module 4. We use a deep reinforcement learning algorithm to optimize the sterilization process and achieve real-time control through an edge computing unit.

[0082] The specific operation is as follows: First, the temperature of the fresh milk is adjusted to 10°C through the mixing module, and the flow rate is set to 100L / h. Then, the milk is heated using the multi-frequency magnetic field of the microwave sterilization module, and the initial microwave power is set to 300W. The microwave power regulator and heat exchanger of the temperature control module work together to accurately control the temperature of the milk at 72°C for 15 seconds. Finally, the milk is quickly cooled to 4°C through the heat exchanger. The entire process is monitored and optimized in real time by the digital twin module.

[0083] Comparative Example 1: In the comparative example, a conventional plate heat exchange sterilization system is used, which sterilizes using heat exchange only, without microwave heating and intelligent control unit.

[0084] The specific operation is as follows: the temperature of the fresh milk is adjusted to 10°C and the flow rate is set to 100L / h. The milk is heated to 72°C through a plate heat exchanger and maintained for 15 seconds. The milk is then quickly cooled to 4°C through another heat exchanger. The whole process is temperature regulated by a PID controller.

[0085] Detection indicators and methods: 1. Bactericidal effect: The total bacterial count was determined by the standard plate count method and expressed as logarithmic reduction.

[0086] 2. Protein denaturation rate: Use the Kjeldahl method to determine the soluble protein content and calculate the denaturation rate.

[0087] 3. Vitamin C retention rate: High performance liquid chromatography (HPLC) was used to determine the vitamin C content.

[0088] 4. Energy consumption: Use a precision electric energy meter to measure the power consumption of the entire sterilization process.

[0089] 5. Sterilization time: Record the total time from the start of heating to the completion of cooling.

[0090] The test results are shown in the following table: Detection indicators Example 1 Comparative Example 1 Bactericidal effect (log reduction) 5.8 5.2 Protein denaturation rate 2.3% 3.8% Vitamin C retention rate 92% 85% Energy consumption (kWh / 1000L) 28.5 35.2 Sterilization time (min) 2.5 4.0 It can be seen from the above test results that the intelligent microwave-heat exchange composite milk sterilization system of the present invention shows obvious advantages in various indicators.

[0091] First, in terms of sterilization effect, Example 1 achieved 5.8 log reduction, while Comparative Example 1 only achieved 5.2 log reduction. This shows that the system of the present invention can more effectively kill microorganisms in milk and improve product safety. This is mainly due to the multi-frequency magnetic field heating technology of the microwave sterilization module, which can heat milk more evenly and avoid sterilization dead corners.

[0092] In terms of protein denaturation rate and vitamin C retention rate, Example 1 was 2.3% and 92% respectively, which was significantly better than 3.8% and 85% of Comparative Example 1. This shows that the system of the present invention can better protect the nutrients in milk. This is because the intelligent control system can accurately control the heating time and temperature to avoid nutritional loss caused by overheating. At the same time, microwave heating can quickly heat up, reduce the heat treatment time, and further reduce nutritional loss.

[0093] In terms of energy consumption, Example 1 only requires 28.5 kWh to process 1000L of milk, which saves about 19% of energy compared with 35.2 kWh in Comparative Example 1. This is mainly due to the efficient heating method and intelligent control strategy of the system of the present invention. Microwave heating can directly act on milk molecules to improve energy utilization efficiency; while the digital twin module and deep reinforcement learning algorithm can continuously optimize sterilization parameters and achieve precise control of energy consumption.

[0094] Finally, in terms of sterilization time, Example 1 only takes 2.5 minutes to complete the entire sterilization process, which is 37.5% shorter than the 4.0 minutes of Comparative Example 1. This greatly improves production efficiency, which means a significant increase in production capacity for large-scale milk processing companies.

[0095] In summary, the intelligent microwave-heat exchange composite milk sterilization system of the present invention has shown significant advantages in sterilization effect, nutritional protection, energy efficiency and production efficiency. This innovative sterilization technology can not only ensure the safety of milk, but also retain its nutritional value to the maximum extent, while significantly reducing production costs and improving production efficiency. This is of great significance to the technological progress and sustainable development of the dairy processing industry.

[0096] It is worth noting that the advantages of the system of the present invention are not limited to the above aspects. Through the digital twin module and deep reinforcement learning algorithm, the system also has the ability of self-learning and optimization. This means that as the running time increases, the system performance will be further improved.

[0097] It should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. Intelligent microwave-heat exchange composite milk sterilization system, characterized in that: The system includes: A mixing module to regulate the flow rate and temperature of raw milk; Microwave sterilization module, which uses multi-frequency magnetic fields to heat milk; A temperature control module, including a microwave power regulator and a heat exchanger, for adjusting microwave heating time and heat exchange rate to control milk temperature; A digital twin module that uses deep reinforcement learning to optimize sterilization processes and simulate system performance in a virtual environment; Among them, the core formula of the deep reinforcement learning optimization algorithm is: In the formula, is the state-action value function, s is the current state, a is the current action, α is the learning rate and 0<α≤1, is the immediate reward, γ is the discount factor and 0≤ <1, For the next state, For the next action.

2. The intelligent microwave-heat exchange composite milk sterilization system according to claim 1 is characterized in that: The mixing module comprises: Flow rate regulator, which adjusts the milk flow by changing the flow rate; Thermostats, which raise or lower the temperature by cooling or heating water; Flow sensor, used to measure flow rate; Temperature sensor, used to measure temperature; A flow controller for controlling the pump drive; A mixing chamber to balance flow rate and temperature changes; Wherein, the flow rate regulator adopts the following control algorithm: In the formula, is the adjusted flow rate, is the initial flow rate, is the error between the target flow rate and the actual flow rate, , , are the proportional, integral and differential coefficients respectively.

3. The intelligent microwave-heat exchange composite milk sterilization system according to claim 1 is characterized in that: The microwave sterilization module comprises: A microwave generator, used for generating microwaves; A microwave power regulator, used to control microwave power; Power meter, used to measure microwave power; a microwave cavity for containing the milk to be sterilized; Wherein, the microwave power regulator adopts the following adaptive control algorithm: In the formula, is the microwave power at time t, is the initial microwave power, is the maximum power adjustment, τ is the time constant, K is the temperature feedback coefficient, is the target temperature, is the actual temperature at time t.

4. The intelligent microwave-heat exchange composite milk sterilization system according to claim 1 is characterized in that: The microwave power regulator of the temperature control module works in coordination with the heat exchanger to precisely adjust the milk temperature by controlling the microwave heating time and heat exchange rate. The coordinated control adopts the following optimization algorithm: Subject to: In the formula, To optimize the objective function, is the temperature at time t, is the target temperature, is the microwave power at time t, is the heat exchange rate at time t, , , is the weight coefficient, is the temperature change rate function, and For temperature limit, is the maximum microwave power, is the maximum heat exchange rate.

5. The intelligent microwave-heat exchange composite milk sterilization system according to claim 1 is characterized in that: The digital twin module uses the following deep reinforcement learning method to optimize the sterilization process: Among them, θ is the policy network parameter, α is the learning rate, is the policy gradient, and: In the formula, is the parameterized policy function, γ is the discount factor, is the reward at time t.

6. The control method of the intelligent microwave-heat exchange composite milk sterilization system according to any one of claims 1 to 5, characterized in that: The method comprises the following steps: a) regulating the flow rate and temperature of the fresh milk through the mixing module; b) heating the milk using a multi-frequency magnetic field of a microwave sterilization module; c) cooperatively controlling the temperature of the milk through the microwave power regulator and the heat exchanger of the temperature control module; d) Optimize sterilization process parameters using deep reinforcement learning methods of the digital twin module; Wherein, step d) adopts the following double Q network algorithm: In the formula, Q is the main network, Q is the target Q network, and is the network parameter, r is the immediate reward, γ is the discount factor, s and a are the current state and action, s' and a' are the next state and action, and the target network is updated every fixed number of steps: ← .

7. The control method according to claim 6, characterized in that: The method further comprises the following steps: e) Use neural network model to analyze the nutritional components in fresh milk and provide reference for microwave sterilization time; f) Based on the Kalman filter algorithm, predict the nutritional components of milk at each moment; g) Use the predicted nutrients to match the optimal sterilization time and achieve precise sterilization; Among them, the neural network model adopts the following attention mechanism: Where Q is the query matrix, K is the key matrix, and V is the value matrix. is the dimension of the key vector.

8. The control method according to claim 6, characterized in that: The method further comprises the following steps: h) Control the mixing module, microwave sterilization module and temperature control module through the edge computing unit; i) Use Kalman filter algorithm to optimize control signals and improve system accuracy and efficiency; The edge computing unit adopts the following distributed optimization algorithm: s.t. In the formula, is the global objective function, is the local objective function of the ith edge node, is the decision variable of the ith edge node, and c is the global constraint.

9. The control method according to claim 6, characterized in that: The method further comprises the following steps: j) Collect data generated by the mixing module, microwave sterilization module, and temperature control module; k) Analyze and store the collected data through the processor of the digital twin module; l) Training sterilization process parameters in a virtual environment; m) passing the optimized sterilization process parameters to the digital twin model for further optimization; The digital twin model adopts the following generative adversarial network (GAN) structure: In the formula, D is the discriminator, G is the generator, is the real data distribution, is the noise distribution, x is the real data sample, and z is the noise sample.

10. The control method according to claim 6, characterized in that: The method further comprises the following steps: n) Use deep reinforcement learning methods to combine and control the mixing module, microwave sterilization module, and temperature control module; o) Continuously optimize the sterilization process in online production to achieve intelligent sterilization; The deep reinforcement learning method adopts the following hierarchical reinforcement learning algorithm: In the formula, is the total Q value function, is the high-level Q-value function used to select the sub-goal g, It is a low-level Q-value function used to select a specific action a, s is the current state, g is the sub-goal, and a is the specific action.

Images