A method and system for coordinated temperature and humidity control in multiple temperature zones of cold chain transportation

Abstract

This invention relates to the field of intelligent control technology in cold chain logistics, specifically a method and system for coordinated temperature and humidity control in multiple temperature zones during cold chain transportation. The method includes: acquiring temperature and humidity data, evaporator status data, actuator status data, and environmental data for each temperature zone; constructing a digital twin model of the multi-temperature zone transportation environment; using a spatiotemporal graph convolutional network to extract the heat transfer coupling characteristics and temperature and humidity correlation characteristics of each temperature zone to form a global state representation; generating a coordinated control sequence for compressor speed, electronic expansion valve opening, evaporator fan speed, and electric heating power in each temperature zone through rolling optimization using a decision core; generating a feedforward decoupling compensation signal based on the temperature and humidity coupling dynamic model and superimposing it with the coordinated control sequence to obtain the final control command for each temperature zone; and driving each actuator to perform coordinated control actions to achieve closed-loop coordinated regulation. This invention achieves global coordinated optimization control of temperature and humidity in multiple temperature zones, improving temperature control accuracy and energy efficiency.

Description

Technical Field

[0001] This invention relates to the field of intelligent control technology for cold chain logistics, specifically to a method and system for coordinated temperature and humidity control in multiple temperature zones during cold chain transportation. Background Technology

[0002] With the explosive growth of fresh food e-commerce, pharmaceutical cold chain, and high-end food logistics industries, the market demand for flexible and high-precision cold chain transportation with "multiple temperatures per vehicle" and "multiple products per container" is becoming increasingly urgent. For example, a refrigerated truck for urban delivery needs to transport frozen meat at -18°C, fresh milk at 4°C, and tropical fruits at 10°C simultaneously. To meet this demand, modern cold chain equipment generally adopts a multi-temperature zone design, dividing the compartment into multiple independent areas by setting up intermediate partitions or independent evaporators.

[0003] However, high-precision and high-stability control in multi-temperature environments faces three core challenges:

[0004] Strong coupling of temperature and humidity with passive dehumidification issues: The surface temperature of the evaporator providing cooling to the freezer compartment is typically well below 0°C, continuously condensing water vapor in the air into frost during the refrigeration process. This causes a sharp drop in the relative humidity of the cold storage compartment (as low as below 20%), severely impacting the freshness of fruits, vegetables, flowers, and other goods. Traditional solutions include adding electric defrosting or humidifiers, but these are not only energy-intensive but also introduce drastic temperature fluctuations.

[0005] Heat load disturbance and airflow temperature crossing: Frequent door opening and closing operations, heat conduction and trace airflow infiltration between different temperature zones, and the "preemptive" heat absorption of newly placed goods can cause sudden changes in the heat load of each temperature zone. Traditional independent PID controllers can only make a lagging response based on the temperature deviation of their own zone, which can easily lead to coupled oscillations such as "one zone cooling drastically, its cold energy leaking to adjacent zones, causing overcompensation in adjacent zones".

[0006] Multi-actuator coordination conflict: The system has multiple actuators (compressors, electronic expansion valves, evaporator fans, and electric heaters) that share the same refrigerant circulation loop and limited power resources. For example, increasing the compressor speed to improve the cooling rate of the freezer compartment will simultaneously increase the refrigerant flow in the refrigerator compartment, leading to overcooling in the refrigerator compartment. Existing control strategies lack global optimization capabilities, often resulting in the energy-efficient dilemma of "compressors running at high speed, but multiple temperature zones failing to meet standards."

[0007] The closest existing solution is a "multi-zone refrigerated truck temperature control system based on independent PID control." Its implementation is as follows: a temperature sensor and controller are independently installed in each temperature zone. Each controller independently calculates the opening degree of the electronic expansion valve or the evaporator fan speed for its zone based on the deviation between the set temperature and the actual measured temperature. The compressor then starts, stops, or adjusts its speed in stages based on the signal from the zone with the highest demand.

[0008] Compared to manual adjustment or mechanical temperature control, this solution achieves independent automatic adjustment of each temperature zone, has a simple structure, lower cost, and can meet basic insulation requirements. However, this solution has the following technical defects and limitations:

[0009] Lack of coordinated control actions: Each temperature zone controller operates independently, leading to an unreasonable distribution of overall cooling capacity. For example, the freezer zone requires a large amount of cooling, causing the compressor to run at high speed, resulting in excessive refrigerant flowing into the refrigerator zone and causing overshoot (temperature far below the set point) in the refrigerator zone. To correct the overshoot, the refrigerator zone controller is forced to shut down its electronic expansion valve or even activate the electric heater, resulting in significant energy waste.

[0010] Unable to resolve the temperature and humidity coupling issue: only temperature is controlled, completely ignoring humidity. The severe dehumidification effect accompanying the refrigeration process cannot be compensated for, resulting in extremely dry air in the refrigerated area and significant dehydration of goods. Even if an independent humidifier is added later, its linkage with temperature control is very crude, easily causing condensation on the cabinet.

[0011] The system suffers from delayed and unpredictable response to disturbances: after strong disturbances such as opening or closing doors, the system waits for the temperature sensor to detect a significant deviation before it begins to respond. This "reactive" adjustment mode results in large temperature fluctuation peaks (up to ±5℃ or more) and long recovery times, which is fatal for temperature-sensitive goods such as vaccines.

[0012] Lack of awareness of the overall system status: It is impossible to perceive the degree of evaporator frost formation, or predict the future changes in the combined heat load of multiple temperature zones. All decisions are based on the local temperature deviation at the current moment, which is short-sighted.

[0013] In summary, there is an urgent need for a multi-temperature zone temperature and humidity collaborative control method and system for cold chain transportation that can achieve global collaborative optimization, decoupled temperature and humidity control, disturbance feedforward prediction, and adaptive learning, so as to promote the development of intelligent control technology for cold chain logistics. Summary of the Invention

[0014] To address the aforementioned technical problems, this invention provides a method and system for coordinated temperature and humidity control in multiple temperature zones of cold chain transportation, aiming to achieve global coordinated optimization control and adaptive adjustment of temperature and humidity in multiple temperature zones, thereby significantly improving temperature control accuracy and energy efficiency.

[0015] To achieve the above objectives, the present invention provides the following technical solution:

[0016] On one hand, embodiments of the present invention provide a method for coordinated temperature and humidity control in multiple temperature zones during cold chain transportation, the method comprising the following steps:

[0017] S100: Acquire temperature and humidity data, evaporator status data, actuator status feedback data and external environment data for each temperature zone, and construct a digital twin model of the multi-temperature zone transportation environment based on the temperature and humidity data, evaporator status data, actuator status feedback data and external environment data for each temperature zone.

[0018] S200, based on the digital twin model, a spatiotemporal graph convolutional network is used to extract the heat transfer coupling features and temperature and humidity correlation features of each temperature range to form a global state representation of the current system;

[0019] S300, the global state representation is input into the decision core based on the hybrid architecture of model predictive control and deep reinforcement learning. By solving the optimization function with multiple objectives of minimizing temperature and humidity deviation, minimizing energy consumption and suppressing disturbances, the coordinated control sequence of compressor speed, electronic expansion valve opening, evaporator fan speed and electric heating power in each temperature zone is generated through rolling optimization.

[0020] S400 generates a feedforward decoupling compensation signal based on a temperature and humidity coupling dynamic model, and superimposes the feedforward decoupling compensation signal with the cooperative control sequence to obtain the final control command for each temperature zone.

[0021] S500, according to the final control command of each temperature zone, drives the compressor, electronic expansion valve, evaporator fan and electric heater to perform coordinated control actions, so as to realize closed-loop coordinated regulation of temperature and humidity in each temperature zone.

[0022] Optionally, in S100, acquiring temperature and humidity data for each temperature zone, evaporator status data, actuator status feedback data, and external environmental data includes:

[0023] S110 collects temperature and humidity data for each temperature zone by using high-precision digital temperature and humidity sensors arranged at the return air inlet, air outlet and cargo center area of ​​each temperature zone.

[0024] S120: By using temperature sensors and pressure sensors arranged on the air inlet and air outlet sides of each evaporator, the evaporator inlet air temperature, evaporator outlet air temperature, evaporator inlet air pressure and evaporator outlet air pressure are collected, and the superheat and subcooling are calculated based on the evaporator inlet air temperature, evaporator outlet air temperature, evaporator inlet air pressure and evaporator outlet air pressure.

[0025] S130 can read the compressor speed, the opening degree of each electronic expansion valve, the speed of each evaporator fan, the power of each electric heater, and the status of the door magnetic switch in real time through the controller local area network bus;

[0026] S140 collects data on external ambient temperature and vehicle driving status through an ambient temperature sensor and a global positioning system module located outside the vehicle compartment.

[0027] S150, the temperature and humidity data of each temperature zone, the superheat and subcooling, the compressor speed and the opening degree of each electronic expansion valve, the speed of each evaporator fan and the power of each electric heater and the status of the door magnetic switch, as well as the external ambient temperature and vehicle driving status data, are aggregated to the central computing unit at a sampling frequency not lower than a preset frequency.

[0028] Optionally, in S100, the construction of a digital twin model of the multi-temperature zone transportation environment based on the temperature and humidity data of each temperature zone, evaporator status data, actuator status feedback data, and external environment data includes:

[0029] S160, based on the first law of thermodynamics, establishes a lumped parameter heat balance equation and a moisture balance equation for each temperature zone. The heat balance equation characterizes the relationship between the temperature change rate of the temperature zone and the cooling capacity, the heat transfer of the building envelope, the heat intrusion of the door opening, the heat of the cargo breathing, and the heat transfer of adjacent temperature zones. The moisture balance equation characterizes the relationship between the humidity change rate of the temperature zone and the dehumidification capacity of the evaporator, the evaporation of the cargo, and the moisture exchange of adjacent temperature zones.

[0030] S170, Based on the system identification experiment, the model parameters of the heat balance equation and the moisture balance equation are obtained offline. The model parameters include the heat capacity and thermal resistance of each temperature zone.

[0031] S180: Using the current temperature zone state quantity as the initial condition, input the predicted disturbance quantity in the future short time domain, perform rolling simulation using the heat balance equation and the moisture balance equation, and output the predicted heat load sequence in the future time domain for each temperature zone.

[0032] Optionally, in S200, the step of extracting heat transfer coupling features and temperature-humidity correlation features for each temperature range based on the digital twin model using a spatiotemporal graph convolutional network to form a global state representation of the current system includes:

[0033] S210 treats each temperature zone as a node in a graph structure, constructs a spatial adjacency matrix based on the adjacency relationship between each temperature zone and the airflow infiltration path, and constructs a node feature matrix based on the temperature, humidity, heat load prediction value and evaporator status of each temperature zone.

[0034] S220, the spatial adjacency matrix and the node feature matrix are input into the first graph convolutional layer of the spatiotemporal graph convolutional network. The graph convolutional layer aggregates the state information of adjacent temperature zones and outputs heat transfer coupling features.

[0035] S230, the heat transfer coupling feature is input into the second gated recurrent unit layer of the spatiotemporal graph convolutional network. The gated recurrent unit layer extracts the temporal evolution law of the state of each node and outputs the temperature and humidity correlation feature.

[0036] S240, the high-dimensional feature vectors of all nodes output by the spatiotemporal graph convolutional network are concatenated to form a low-dimensional and compact global state representation vector.

[0037] Optionally, in S300, the step of inputting the global state representation into the decision core based on a hybrid architecture of model predictive control and deep reinforcement learning, and generating a coordinated control sequence of compressor speed, electronic expansion valve opening, evaporator fan speed, and electric heating power for each temperature zone through rolling optimization by solving an optimization function with multiple objectives of minimizing temperature and humidity deviation, minimizing energy consumption, and suppressing disturbances, includes:

[0038] S310, the global state representation vector and the predicted heat load sequence are input into the offline trained policy network, which is trained based on the soft actor-commentator algorithm. The policy network outputs a candidate action sequence, which includes the compressor target speed, the electronic expansion valve target opening degree of each temperature zone, the evaporator fan target speed of each temperature zone, and the electric heater target power of each temperature zone for each control cycle in the predicted time domain.

[0039] S320, in each control cycle, the first action in the candidate action sequence is executed. In the next control cycle, the candidate action sequence is recalculated based on the updated actual state and the first action in the updated candidate action sequence is executed to achieve rolling time-domain closed-loop optimization control.

[0040] Optionally, in S400, the generation of the feedforward decoupling compensation signal based on the temperature and humidity coupling dynamic model includes:

[0041] S410, Establish a temperature and humidity coupled transfer function model, wherein the temperature and humidity coupled transfer function model characterizes the quantitative mapping relationship between temperature control quantity and humidity change quantity;

[0042] S420, input the temperature control command in the cooperative control sequence into the temperature and humidity coupling transfer function model, and calculate the predicted value of humidity change caused by the temperature control command;

[0043] S430, a feedforward decoupling compensation signal is generated based on the humidity change prediction value. The feedforward decoupling compensation signal includes a humidifier control signal and an evaporator fan speed adjustment signal, so that the humidity change path is decoupled from the temperature control quantity.

[0044] Optionally, after S500, the method further includes:

[0045] S610 packages and stores sensor data, control action sequences, and cargo quality assessment results from each transportation mission into a cloud database.

[0046] S620: Periodically extract new data from the cloud database, perform incremental training on the policy network, and update the network model parameters;

[0047] The S630 sends the updated network model parameters to the vehicle edge computing unit to achieve continuous optimization of the control strategy.

[0048] On the other hand, embodiments of the present invention provide a multi-temperature zone temperature and humidity coordinated control system for cold chain transportation, comprising:

[0049] At least one processor;

[0050] At least one memory for storing at least one program;

[0051] When the at least one program is executed by the at least one processor, the at least one processor performs the method described above.

[0052] On the other hand, embodiments of the present invention provide a computer-readable storage medium storing a processor-executable program, which, when executed by a processor, is used to perform the above-described method.

[0053] On the other hand, embodiments of the present invention provide a computer program product, including a computer program or computer instructions, the computer program or computer instructions being stored in a memory, a processor of a computer device reading the computer program or computer instructions from the memory, and the processor executing the computer program or computer instructions to cause the computer device to perform the above-described method.

[0054] The embodiments of the present invention have the following beneficial effects:

[0055] The multi-temperature zone state representation method proposed in this invention, based on digital twins and spatiotemporal graph convolution, constructs a multi-temperature zone thermodynamic digital twin model and uses a spatiotemporal graph convolutional network to automatically extract the dynamic thermal coupling features of each temperature zone to form a global state representation. This effectively solves the problem of lacking global state perception of the system in the prior art and realizes global collaborative optimization control.

[0056] The proposed collaborative optimization method based on a hybrid architecture of model predictive control and deep reinforcement learning models the multi-temperature zone collaborative control problem as a constrained multi-objective optimization problem. It uses the soft actor-commentator algorithm to train a policy network to approximate the rolling time-domain optimization problem of model predictive control, thereby achieving global optimal allocation of refrigeration resources and avoiding resource conflicts and energy waste caused by independent control.

[0057] The temperature and humidity coordinated control method based on feedforward decoupling proposed in this invention establishes a temperature and humidity coupling transfer function model and designs a feedforward decoupling compensator. When a cooling or heating command is issued, a humidity compensation signal is output synchronously, realizing independent and precise coordinated control of temperature and humidity. While maintaining stable temperature, the relative humidity is precisely controlled within the optimal storage range for goods.

[0058] The disturbance suppression method based on heat load prediction proposed in this invention uses a digital twin model to predict the future heat load of typical disturbances such as door opening and closing and the insertion of new goods. The predicted value is then introduced as a feedforward quantity into the model predictive control optimization function, realizing predictive compensation control after the disturbance occurs and significantly reducing the maximum dynamic deviation.

[0059] The multi-actuator global energy collaborative allocation method proposed in this invention aims to minimize the total energy consumption of the system. By using an intelligent decision-making module to uniformly calculate and allocate the actions of each actuator, it avoids resource conflicts and energy waste caused by independent control and significantly improves the system's energy efficiency ratio.

[0060] The proposed control model continuous evolution architecture based on incremental learning uses the cargo quality assessment results after the actual transportation task is completed as a supervision signal to periodically perform incremental training on the deep reinforcement learning decision model deployed in the cloud, and then distributes the optimized model to the edge, thereby realizing the continuous optimization and self-evolution capability of the control strategy. Attached Figure Description

[0061] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0062] Figure 1 This is a flowchart illustrating the multi-temperature zone temperature and humidity coordinated control method for cold chain transportation in an embodiment of the present invention.

[0063] Figure 2 This is a schematic diagram of the overall architecture of the multi-temperature zone temperature and humidity coordinated control system for cold chain transportation in this embodiment of the invention. Detailed Implementation

[0064] The following will provide a clear and complete description of the concept, specific structure, and technical effects of the present invention in conjunction with embodiments and accompanying drawings, so as to fully understand the purpose, solution, and effects of the present invention. It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other.

[0065] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention. In the following description, when referring to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the embodiments of this invention; they are merely examples of apparatuses and methods consistent with some aspects of the embodiments of this invention as detailed in the appended claims.

[0066] Unless otherwise defined, all technical and scientific terms used in this invention have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used in this invention is for descriptive purposes only and is not intended to limit the invention.

[0067] refer to Figure 1 ,like Figure 1 The image shows a method for coordinated temperature and humidity control in multiple temperature zones during cold chain transportation, provided by an embodiment of the present invention. The method includes the following steps:

[0068] S100 acquires temperature and humidity data, evaporator status data, actuator status feedback data, and external environmental data for each temperature zone, and constructs a digital twin model of the multi-temperature zone transportation environment.

[0069] S200, based on the digital twin model, a spatiotemporal graph convolutional network is used to extract the heat transfer coupling features and temperature and humidity correlation features of each temperature range to form a global state representation of the current system;

[0070] S300, the global state representation is input into the decision core based on the hybrid architecture of model predictive control and deep reinforcement learning. By solving the optimization function with multiple objectives of minimizing temperature and humidity deviation, minimizing energy consumption and suppressing disturbances, the coordinated control sequence of compressor speed, electronic expansion valve opening, evaporator fan speed and electric heating power in each temperature zone is generated through rolling optimization.

[0071] S400 generates a feedforward decoupling compensation signal based on a temperature and humidity coupling dynamic model, and superimposes the feedforward decoupling compensation signal with the cooperative control sequence to obtain the final control command for each temperature zone.

[0072] S500, according to the final control command of each temperature zone, drives the compressor, electronic expansion valve, evaporator fan and electric heater to perform coordinated control actions, so as to realize closed-loop coordinated regulation of temperature and humidity in each temperature zone.

[0073] This invention provides a method and system for coordinated temperature and humidity control in multi-temperature zones of cold chain transportation. By constructing a digital twin model of the multi-temperature zone transportation environment, and using lumped-parameter heat balance equations and humidity balance equations for rolling simulation, it outputs the predicted heat load sequence for each temperature zone in the future time domain. This effectively solves the problem of lacking the ability to predict future heat load changes in existing technologies, achieving feedforward prediction and suppression of disturbances. Based on the digital twin model, a spatiotemporal graph convolutional network is used to extract the heat transfer coupling characteristics and temperature-humidity correlation characteristics of each temperature zone, forming a global state representation. This effectively solves the problem of lacking global state awareness in existing technologies, achieving global coordinated optimization control. A coordinated control sequence is generated through rolling optimization of the decision core based on a hybrid architecture of model predictive control and deep reinforcement learning. Temperature and humidity decoupling compensation is performed through a feedforward decoupling compensator, effectively solving the problems of temperature and humidity coupling runaway and multi-actuator coordination conflicts in existing technologies, achieving independent and precise coordinated control of temperature and humidity.

[0074] The core of this embodiment lies in the closed-loop intelligent control architecture of "perception-prediction-decision-execution", which achieves global optimization control of temperature and humidity in multiple temperature zones through digital twin modeling, spatiotemporal feature extraction, hybrid intelligent decision-making and multi-actuator collaboration.

[0075] like Figure 2 As shown, the overall system architecture provided in this embodiment of the invention includes: a multi-source data acquisition and preprocessing module, a digital twin and heat load prediction module, a spatiotemporal feature extraction and state characterization module, a hybrid intelligent decision-making and control module, a multi-actuator driving and decoupling module, and an incremental learning and model update module. These modules are connected sequentially to form a closed loop of data flow and control flow.

[0076] In some embodiments, S100, acquiring temperature and humidity data for each temperature zone, evaporator status data, actuator status feedback data, and external environment data includes:

[0077] S110 collects temperature and humidity data for each temperature zone by using high-precision digital temperature and humidity sensors arranged at the return air inlet, air outlet and cargo center area of ​​each temperature zone.

[0078] S120: By using temperature sensors and pressure sensors arranged on the air inlet and air outlet sides of each evaporator, the evaporator inlet air temperature, evaporator outlet air temperature, evaporator inlet air pressure and evaporator outlet air pressure are collected, and the superheat and subcooling are calculated based on the evaporator inlet air temperature, evaporator outlet air temperature, evaporator inlet air pressure and evaporator outlet air pressure.

[0079] S130 can read the compressor speed, the opening degree of each electronic expansion valve, the speed of each evaporator fan, the power of each electric heater, and the status of the door magnetic switch in real time through the controller local area network bus;

[0080] S140 collects data on external ambient temperature and vehicle driving status through an ambient temperature sensor and a global positioning system module located outside the vehicle compartment.

[0081] S150, the temperature and humidity data of each temperature zone, the superheat and subcooling, the compressor speed and the opening degree of each electronic expansion valve, the speed of each evaporator fan and the power of each electric heater and the status of the door magnetic switch, as well as the external ambient temperature and vehicle driving status data, are aggregated to the central computing unit at a sampling frequency not lower than a preset frequency.

[0082] This embodiment utilizes high-precision digital temperature and humidity sensors deployed at the return air inlets, supply air inlets, and cargo center areas of each temperature zone to comprehensively and accurately collect temperature and humidity data for each zone. Temperature and pressure sensors located on the air inlet and outlet sides of each evaporator acquire the evaporator's operating status and calculate superheat and subcooling based on the collected data, thereby determining the degree of frost formation on the evaporator. Real-time actuator status data is read via the controller area network bus, ensuring data timeliness and reliability. Environmental temperature sensors and a GPS module located outside the vehicle compartment acquire external ambient temperature and vehicle driving status data, providing a basis for heat load prediction. All data is aggregated to the central computing unit at a sampling frequency no less than a preset frequency, ensuring data synchronization and integrity.

[0083] In some embodiments, S100, constructing a digital twin model of a multi-temperature zone transportation environment includes:

[0084] S160, based on the first law of thermodynamics, establishes a lumped parameter heat balance equation and a moisture balance equation for each temperature zone. The heat balance equation characterizes the relationship between the temperature change rate of the temperature zone and the cooling capacity, the heat transfer of the building envelope, the heat intrusion of the door opening, the heat of the cargo breathing, and the heat transfer of adjacent temperature zones. The moisture balance equation characterizes the relationship between the humidity change rate of the temperature zone and the dehumidification capacity of the evaporator, the evaporation of the cargo, and the moisture exchange of adjacent temperature zones.

[0085] S170, Based on the system identification experiment, the model parameters of the heat balance equation and the moisture balance equation are obtained offline. The model parameters include the heat capacity and thermal resistance of each temperature zone.

[0086] S180: Using the current temperature zone state quantity as the initial condition, input the predicted disturbance quantity in the future short time domain, perform rolling simulation using the heat balance equation and the moisture balance equation, and output the predicted heat load sequence in the future time domain for each temperature zone.

[0087] This embodiment establishes lumped-parameter heat balance and humidity balance equations for each temperature zone based on the first law of thermodynamics, enabling precise description of the thermodynamic dynamic characteristics of each zone. The heat balance equation characterizes the relationship between the temperature change rate of each zone and the cooling capacity, heat transfer from the building envelope, heat intrusion from door openings, heat from cargo respiration, and heat transfer from adjacent temperature zones. The humidity balance equation characterizes the relationship between the humidity change rate of each zone and the evaporator dehumidification capacity, cargo evaporation, and moisture exchange between adjacent temperature zones. Model parameters are obtained offline through system identification experiments, ensuring model accuracy. Using the current temperature zone state variables as initial conditions, and inputting predicted disturbances in the short-term future time domain, rolling simulations are performed using the heat balance and humidity balance equations to output the predicted heat load sequence for each temperature zone in the future time domain. This achieves prediction of future heat load changes, providing a foundation for subsequent feedforward control.

[0088] In some embodiments, S200, the step of extracting heat transfer coupling features and temperature-humidity correlation features for each temperature range based on the digital twin model using a spatiotemporal graph convolutional network to form a global state representation of the current system includes:

[0089] S210 treats each temperature zone as a node in a graph structure, constructs a spatial adjacency matrix based on the adjacency relationship between each temperature zone and the airflow infiltration path, and constructs a node feature matrix based on the temperature, humidity, heat load prediction value and evaporator status of each temperature zone.

[0090] S220, the spatial adjacency matrix and the node feature matrix are input into the first graph convolutional layer of the spatiotemporal graph convolutional network. The graph convolutional layer aggregates the state information of adjacent temperature zones and outputs heat transfer coupling features.

[0091] S230, the heat transfer coupling feature is input into the second gated recurrent unit layer of the spatiotemporal graph convolutional network. The gated recurrent unit layer extracts the temporal evolution law of the state of each node and outputs the temperature and humidity correlation feature.

[0092] S240, the high-dimensional feature vectors of all nodes output by the spatiotemporal graph convolutional network are concatenated to form a low-dimensional and compact global state representation vector.

[0093] This embodiment treats each temperature zone as a node in a graph structure. A spatial adjacency matrix is ​​constructed based on the adjacency relationships between temperature zones and the airflow infiltration path, accurately representing the spatial connectivity between them. The predicted temperature, humidity, heat load, and evaporator status of each temperature zone are used to construct a node feature matrix, comprehensively reflecting the current state of each zone. The spatial adjacency matrix and node feature matrix are input into the first graph convolutional layer of the spatiotemporal graph convolutional network. This layer aggregates the state information of adjacent temperature zones and outputs heat transfer coupling features, capturing the heat transfer relationships between temperature zones. The heat transfer coupling features are then input into the second gated recurrent unit layer. This layer extracts the temporal evolution of each node's state and outputs temperature and humidity correlation features, capturing the temporal trend of each temperature zone's state. Finally, the high-dimensional feature vectors of all nodes output by the spatiotemporal graph convolutional network are concatenated to form a low-dimensional, compact global state representation vector, providing comprehensive state information for subsequent decision-making.

[0094] In some embodiments, S300, the step of inputting the global state representation into the decision core based on a hybrid architecture of model predictive control and deep reinforcement learning, and generating a coordinated control sequence of compressor speed, electronic expansion valve opening, evaporator fan speed, and electric heating power for each temperature zone through rolling optimization by solving an optimization function with multiple objectives of minimizing temperature and humidity deviation, minimizing energy consumption, and suppressing disturbances, includes:

[0095] S310, the global state representation vector and the predicted heat load sequence are input into the offline trained policy network, which is trained based on the soft actor-commentator algorithm. The policy network outputs a candidate action sequence, which includes the compressor target speed, the electronic expansion valve target opening degree of each temperature zone, the evaporator fan target speed of each temperature zone, and the electric heater target power of each temperature zone for each control cycle in the predicted time domain.

[0096] S320, in each control cycle, the first action in the candidate action sequence is executed. In the next control cycle, the candidate action sequence is recalculated based on the updated actual state and the first action in the updated candidate action sequence is executed to achieve rolling time-domain closed-loop optimization control.

[0097] This embodiment inputs the global state representation vector and the predicted heat load sequence into an offline-trained policy network. The policy network, trained using a soft actor-commentator algorithm, efficiently handles the continuous action space and encourages exploration by maximizing expected reward and policy entropy, thus avoiding local optima. The policy network outputs a candidate action sequence, which includes the compressor target speed, the electronic expansion valve target opening degree for each temperature zone, the evaporator fan target speed for each temperature zone, and the electric heater target power for each temperature zone in each control cycle within the predicted time domain. In each control cycle, the first action in the candidate action sequence is executed. In the next control cycle, the candidate action sequence is recalculated based on the updated actual state, and the first action in the updated candidate action sequence is executed, achieving rolling time-domain closed-loop optimization control with good real-time performance and adaptability.

[0098] In some embodiments, S400, the generation of the feedforward decoupling compensation signal based on the temperature and humidity coupling dynamic model includes:

[0099] S410, Establish a temperature and humidity coupled transfer function model, wherein the temperature and humidity coupled transfer function model characterizes the quantitative mapping relationship between temperature control quantity and humidity change quantity;

[0100] S420, input the temperature control command in the cooperative control sequence into the temperature and humidity coupling transfer function model, and calculate the predicted value of humidity change caused by the temperature control command;

[0101] S430, a feedforward decoupling compensation signal is generated based on the humidity change prediction value. The feedforward decoupling compensation signal includes a humidifier control signal and an evaporator fan speed adjustment signal, so that the humidity change path is decoupled from the temperature control quantity.

[0102] This embodiment establishes a temperature-humidity coupling transfer function model to characterize the quantitative mapping relationship between temperature control and humidity changes, accurately describing the coupling characteristics between temperature and humidity. The temperature control command from the coordinated control sequence is input into the temperature-humidity coupling transfer function model to calculate the predicted humidity change caused by the temperature control command, thus achieving humidity prediction. Based on the predicted humidity change, a feedforward decoupling compensation signal is generated, including a humidifier control signal and an evaporator fan speed adjustment signal. This decouples the humidity change path from the temperature control, achieving independent and precise coordinated control of temperature and humidity.

[0103] In some embodiments, after S500, the method further includes:

[0104] S610 packages and stores sensor data, control action sequences, and cargo quality assessment results from each transportation mission into a cloud database.

[0105] S620: Periodically extract new data from the cloud database, perform incremental training on the policy network, and update the network model parameters;

[0106] The S630 sends the updated network model parameters to the vehicle edge computing unit to achieve continuous optimization of the control strategy.

[0107] This embodiment achieves data accumulation by packaging and storing sensor data, control action sequences, and cargo quality assessment results from each transportation mission into a cloud database. New data is periodically extracted from the cloud database to incrementally train the strategy network, updating the network model parameters and enabling continuous optimization of the control strategy. The updated network model parameters are then distributed to the onboard edge computing unit, achieving online updates of the control strategy. This allows the system to adapt to changes in different cargo and operating conditions, achieving a self-evolving capability that becomes smarter with use.

[0108] This invention also provides a multi-temperature zone temperature and humidity coordinated control system for cold chain transportation, comprising:

[0109] At least one processor;

[0110] At least one memory for storing at least one program;

[0111] When the at least one program is executed by the at least one processor, the at least one processor performs the method described in any of the preceding statements.

[0112] The content of the above method embodiments is applicable to this embodiment. The specific functions implemented in this embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments. Therefore, they will not be repeated here.

[0113] This invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method described in any of the above embodiments.

[0114] It is understood that the content of the above method embodiments is applicable to this storage medium embodiment. The specific functions implemented in this storage medium embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments.

[0115] This invention also provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the method described in any of the above embodiments.

[0116] It is understood that the content of the above method embodiments is applicable to this device embodiment. The specific functions implemented by this device embodiment are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0117] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may optionally include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0118] It will be understood by those skilled in the art that all or some of the steps and systems in the methods disclosed above can be implemented as software, firmware, hardware, and suitable combinations thereof. Some or all of the physical components can be implemented as software executed by a processor, such as a central processing unit, digital signal processor, or microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit. Such software can be distributed on a computer-readable medium, which can include computer storage media (or non-transitory media) and communication media (or transient media). As is known to those skilled in the art, the term computer storage media includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data). Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, digital versatile disc (DVD) or other optical disc storage, magnetic cartridges, magnetic tape, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and is accessible to a computer. Furthermore, as is known to those skilled in the art, communication media typically include computer-readable instructions, data structures, program modules, or other data in modulated data signals such as carrier waves or other transmission mechanisms, and may include any information delivery medium.

[0119] The embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

Claims

1. A method for coordinated temperature and humidity control in multiple temperature zones during cold chain transportation, characterized in that, The method includes the following steps: S100 acquires temperature and humidity data, evaporator status data, actuator status feedback data, and external environmental data for each temperature zone, and constructs a digital twin model of the multi-temperature zone transportation environment. S200, based on the digital twin model, a spatiotemporal graph convolutional network is used to extract the heat transfer coupling features and temperature and humidity correlation features of each temperature range to form a global state representation of the current system; S300, the global state representation is input into the decision core based on the hybrid architecture of model predictive control and deep reinforcement learning. By solving the optimization function with multiple objectives of minimizing temperature and humidity deviation, minimizing energy consumption and suppressing disturbances, the coordinated control sequence of compressor speed, electronic expansion valve opening, evaporator fan speed and electric heating power in each temperature zone is generated through rolling optimization. S400 generates a feedforward decoupling compensation signal based on a temperature and humidity coupling dynamic model, and superimposes the feedforward decoupling compensation signal with the cooperative control sequence to obtain the final control command for each temperature zone. S500, according to the final control command of each temperature zone, drives the compressor, electronic expansion valve, evaporator fan and electric heater to perform coordinated control actions, so as to realize closed-loop coordinated regulation of temperature and humidity in each temperature zone.

2. The method according to claim 1, characterized in that, In S100, acquiring temperature and humidity data for each temperature zone, evaporator status data, actuator status feedback data, and external environmental data includes: S110 collects temperature and humidity data for each temperature zone by using high-precision digital temperature and humidity sensors arranged at the return air inlet, air outlet and cargo center area of ​​each temperature zone. S120: By using temperature sensors and pressure sensors arranged on the air inlet and air outlet sides of each evaporator, the evaporator inlet air temperature, evaporator outlet air temperature, evaporator inlet air pressure and evaporator outlet air pressure are collected, and the superheat and subcooling are calculated based on the evaporator inlet air temperature, evaporator outlet air temperature, evaporator inlet air pressure and evaporator outlet air pressure. S130 can read the compressor speed, the opening degree of each electronic expansion valve, the speed of each evaporator fan, the power of each electric heater, and the status of the door magnetic switch in real time through the controller local area network bus; S140 collects data on external ambient temperature and vehicle driving status through an ambient temperature sensor and a global positioning system module located outside the vehicle compartment. S150, the temperature and humidity data of each temperature zone, the superheat and subcooling, the compressor speed and the opening degree of each electronic expansion valve, the speed of each evaporator fan and the power of each electric heater and the status of the door magnetic switch, as well as the external ambient temperature and vehicle driving status data, are aggregated to the central computing unit at a sampling frequency not lower than a preset frequency.

3. The method according to claim 1, characterized in that, In S100, the construction of a digital twin model of a multi-temperature zone transportation environment includes: S160, based on the first law of thermodynamics, establishes a lumped parameter heat balance equation and a moisture balance equation for each temperature zone. The heat balance equation characterizes the relationship between the temperature change rate of the temperature zone and the cooling capacity, the heat transfer of the building envelope, the heat intrusion of the door opening, the heat of the cargo breathing, and the heat transfer of adjacent temperature zones. The moisture balance equation characterizes the relationship between the humidity change rate of the temperature zone and the dehumidification capacity of the evaporator, the evaporation of the cargo, and the moisture exchange of adjacent temperature zones. S170, Based on the system identification experiment, the model parameters of the heat balance equation and the moisture balance equation are obtained offline. The model parameters include the heat capacity and thermal resistance of each temperature zone. S180: Using the current temperature zone state quantity as the initial condition, input the predicted disturbance quantity in the future short time domain, perform rolling simulation using the heat balance equation and the moisture balance equation, and output the predicted heat load sequence in the future time domain for each temperature zone.

4. The method according to claim 1, characterized in that, In S200, based on the digital twin model, a spatiotemporal graph convolutional network is used to extract the heat transfer coupling features and temperature-humidity correlation features of each temperature range to form a global state representation of the current system, including: S210 treats each temperature zone as a node in a graph structure, constructs a spatial adjacency matrix based on the adjacency relationship between each temperature zone and the airflow infiltration path, and constructs a node feature matrix based on the temperature, humidity, heat load prediction value and evaporator status of each temperature zone. S220, the spatial adjacency matrix and the node feature matrix are input into the first graph convolutional layer of the spatiotemporal graph convolutional network. The graph convolutional layer aggregates the state information of adjacent temperature zones and outputs heat transfer coupling features. S230, the heat transfer coupling feature is input into the second gated recurrent unit layer of the spatiotemporal graph convolutional network. The gated recurrent unit layer extracts the temporal evolution law of the state of each node and outputs the temperature and humidity correlation feature. S240, the high-dimensional feature vectors of all nodes output by the spatiotemporal graph convolutional network are concatenated to form a low-dimensional and compact global state representation vector.

5. The method according to claim 1, characterized in that, In S300, the global state representation is input into the decision core based on a hybrid architecture of model predictive control and deep reinforcement learning. By solving an optimization function with multiple objectives—minimizing temperature and humidity deviation, minimizing energy consumption, and suppressing disturbances—a coordinated control sequence for compressor speed, electronic expansion valve opening, evaporator fan speed, and electric heating power in each temperature zone is generated through rolling optimization. This includes: S310, the global state representation vector and the predicted heat load sequence are input into the offline trained policy network, which is trained based on the soft actor-commentator algorithm. The policy network outputs a candidate action sequence, which includes the compressor target speed, the electronic expansion valve target opening degree of each temperature zone, the evaporator fan target speed of each temperature zone, and the electric heater target power of each temperature zone for each control cycle in the predicted time domain. S320, in each control cycle, the first action in the candidate action sequence is executed. In the next control cycle, the candidate action sequence is recalculated based on the updated actual state and the first action in the updated candidate action sequence is executed to achieve rolling time-domain closed-loop optimization control.

6. The method according to claim 1, characterized in that, In S400, the generation of the feedforward decoupling compensation signal based on the temperature and humidity coupling dynamic model includes: S410, Establish a temperature and humidity coupled transfer function model, wherein the temperature and humidity coupled transfer function model characterizes the quantitative mapping relationship between temperature control quantity and humidity change quantity; S420, input the temperature control command in the cooperative control sequence into the temperature and humidity coupling transfer function model, and calculate the predicted value of humidity change caused by the temperature control command; S430, a feedforward decoupling compensation signal is generated based on the humidity change prediction value. The feedforward decoupling compensation signal includes a humidifier control signal and an evaporator fan speed adjustment signal, so that the humidity change path is decoupled from the temperature control quantity.

7. The method according to claim 5, characterized in that, Following S500, the method further includes: S610 packages and stores sensor data, control action sequences, and cargo quality assessment results from each transportation mission into a cloud database. S620: Periodically extract new data from the cloud database, perform incremental training on the policy network, and update the network model parameters; The S630 sends the updated network model parameters to the vehicle edge computing unit to achieve continuous optimization of the control strategy.

8. A multi-temperature zone temperature and humidity coordinated control system for cold chain transportation, characterized in that, include: At least one processor; At least one memory for storing at least one program; When the at least one program is executed by the at least one processor, the at least one processor performs the method as described in any one of claims 1 to 7.

9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 7.

Images