Self-adaptive environment control system and method based on multi-mode biological feedback

Abstract

The invention discloses a self-adaptive environment control system and method based on multi-mode biological feedback. The system integrates physical sensing and two types of biological sensing, on one hand, an environment-biological coupling model is calibrated in real time through'two-channel self-adaption ', on the other hand, a set point is dynamically adjusted according to a welfare index, and an optimal control sequence is solved through MPC. According to the adaptive environment control method based on multi-modal biological feedback, continuous self-calibration can be achieved under data drift and working condition changes, biological requirements are accurately recognized, temperature, humidity, wind, light, ventilation and other factors are dynamically regulated and controlled, the control precision and energy efficiency are remarkably improved, fluctuation and stress are reduced, growth and output are improved, and the adaptive environment control method based on multi-modal biological feedback is suitable for large-scale popularization and application. The method is suitable for livestock and greenhouses.

Description

Technical Field

[0001] This invention relates to the field of intelligent control technology, specifically to an adaptive environmental control system and method based on multimodal biological feedback. Background Technology

[0002] Currently, in intensive livestock farming, greenhouse cultivation, and other bio-production fields, environmental control systems are crucial for ensuring production efficiency and biowelfare. Existing technologies mainly include: 1. Threshold-based on-off control / PID control: This method compares readings from environmental sensors (such as thermometers) with a fixed setpoint to control the start and stop of heating, ventilation, and other equipment. While simple, this method suffers from poor control accuracy and is prone to causing significant fluctuations in environmental parameters and energy waste.

[0003] 2. Preset-programmed control: Environmental parameters are set in stages according to the growth stage of the organism (such as age in days, growth cycle). This method is an improvement over simple on / off control, but its preset "growth curve" or "environment curve" is a static model based on historical statistical data, which cannot respond to the actual health status, vitality and genetic differences of the current batch of organisms, resulting in poor adaptability.

[0004] 3. Advanced control based on model predictive control (MPC): This method uses mathematical models to predict future environmental changes and optimizes control strategies to achieve energy saving and precision.

[0005] However, even the most advanced MPC systems generally suffer from the following fundamental flaws in existing technologies: (1) Staticization of model parameters: The key biological parameters (e.g., total heat production of animal populations and transpiration rate of plants) in the system model (such as thermodynamic model) on which MPC depends are usually fixed values ​​obtained by looking up tables in literature or estimating offline. These parameters cannot reflect the real-time dynamic changes of organisms caused by factors such as feeding, stress, and health status, resulting in distorted model predictions and deviation of control effects from the optimal.

[0006] (2) Humanization of control targets: The targets of the control system (such as the temperature setpoint) are set by humans based on experience or standards. This "optimal value" does not necessarily represent the organism's true "subjective feeling" or "welfare state". For example, the set temperature of 26°C may be too cold or too hot for the pigs due to changes in air humidity, airflow or the pigs' own health condition, and the existing system cannot detect this "perception deviation".

[0007] In summary, existing technologies lack a closed-loop mechanism that allows control systems to obtain feedback directly from the organism itself and dynamically correct its internal model and target setpoint accordingly. Summary of the Invention

[0008] The purpose of this invention is to overcome the shortcomings of existing technologies and provide an adaptive environmental control system and method based on multimodal biofeedback. This invention senses the macroscopic physiological state and macroscopic behavioral and welfare state of biological populations online and in real time, and then integrates these two biofeedback mechanisms into a unified predictive control framework to achieve dual adaptation of the model and the objective, thereby achieving higher control accuracy, resource efficiency, and biological welfare levels.

[0009] To achieve the above objectives, the technical solution designed by the present invention is as follows: This invention provides an adaptive environment control method based on multimodal biological feedback, comprising the following steps: S1 collects physical environmental parameters within the target environment in real time, as well as biofeedback data reflecting the state of the biological population collected by biosensors of different modalities; the biofeedback data includes at least a first type of biological data reflecting the macroscopic physiological and metabolic state and a second type of biological data reflecting the macroscopic behavioral and welfare state. S2 Based on the first type of biological data, the current value of at least one key biological source parameter in the physical-biological coupling model is calculated online using a physiological metabolism conversion algorithm, and the physical-biological coupling model is updated in real time using the current value. S3 Based on the second type of biological data, a quantitative biological welfare index is calculated through a behavior pattern recognition algorithm, and the environmental target setpoint of the environmental control system is dynamically adjusted according to the deviation between the biological welfare index and the preset target welfare index. S4 Input the updated physical-biological coupling model from step S2 and the adjusted environmental target setpoint from step S3 into the predictive control module, and calculate the optimal control command for the environmental control equipment in the future period through the optimization algorithm. S5 The optimal control command is sent to the environmental control device for execution to adjust the target environment; Steps S1 to S5 are executed cyclically to form a continuous closed-loop control.

[0010] Furthermore, the specific steps in step S2 for calculating the current values ​​of key biological source parameters based on the first type of biological data include: S21: Nighttime stable window screening. Within a fixed period of the day, the time window in which the minimum ventilation rate is continuously and stably operated, the indoor temperature fluctuation is less than the set threshold, and the CO2 concentration sequence has no negative slope anomalies is selected as the nighttime stable window. S22: Calculate the CO2 production rate based on the data within the nighttime stable window; S23: Calculate the total heat production of the biological population based on the CO2 production rate; S24: Update the physical-biological coupling model using the total heat production as a time-varying biological source parameter.

[0011] Furthermore, the physical-biological coupling model is a pigsty thermodynamic model, whose state-space equation is described as: x{k+1}=F x k +G u k +Gd Q{pigs,k}; where x k This is the system state vector, containing the indoor temperature; u k The control input vector includes the airflow of the ventilation equipment and the power of the heater; Q{pigs, k} is the total heat production, which serves as a measurable feedforward perturbation input; F, G, Gd are model parameter matrices obtained through system identification or physical derivation.

[0012] Furthermore, in step S22, the CO2 production rate is calculated according to the following formula: Where V is the volume of the biological environment, t is time, and C is... in,t Let C be the indoor CO2 concentration at time t. out,t Let Q be the external CO2 concentration at time t, Δt be the sampling time interval, and Q be the external CO2 concentration at time t. vent For minimum ventilation rate, G CO2 This represents the CO2 production rate. This represents the indoor CO2 concentration at the previous sampling time.

[0013] Furthermore, in step S23, the total heat production is calculated according to the following formula. : Where RQ is the respiratory entropy, G CO2 Based on the calculated CO2 production rate, The total heat production of the biological community.

[0014] Furthermore, the target optimization in step S3 specifically includes the following sub-steps: S31: By analyzing the top view images in the second type of biological data, the behavioral characteristics of the biological population are extracted, including calculating the clustering index based on the number of connected components in the image and calculating the dispersion index based on the average nearest neighbor distance between individuals; S32: Calculate the quantified thermal comfort index based on the clustering index and the dispersion index; S33: Dynamically generate a temperature setpoint based on the deviation between the thermal comfort index and the target comfort level.

[0015] Furthermore, the method for calculating the clustering index in step S31 is as follows: ,in, The number of connected components. The total number of organisms; the dispersion index DI is obtained by normalizing the average nearest neighbor distance of individual organisms; In step S32, the formula for calculating the Thermal Comfort Index (TCI) is as follows: in, It is the expected dispersion under comfortable conditions. It is the slope of the sigmoid function. It is the penalty weight for group behavior; the TCI value range is [0,1], and the higher the value, the more comfortable it is.

[0016] Furthermore, in step S4, the physical-biological coupling model updated in step S2 is compared with the dynamic setpoint obtained in step S3. The environmental target setpoint sequence is formed by superposition, and model predictive control is used to solve the control sequence of the execution equipment in each control cycle; the model predictive control solves the following optimization problem in each control cycle: To optimize the solution variables, k is the discrete time step index, and N is the prediction time domain length. For temperature tracking weighting coefficients, For the predicted future indoor temperature, For dynamic target temperature setpoint, Energy consumption cost weighting coefficient This is a power consumption function.

[0017] Furthermore, the constraints of the model predictive control are as follows: in, The system dynamics constraints defined for the physical-biological coupling model updated in step S2. The total heat production is calculated in real time; Control variables were defined to define the operational constraints of environmental control equipment. The upper and lower limits of feasibility.

[0018] This invention also provides an adaptive environmental control system based on multimodal biofeedback. The system includes: sensors, a processor, and an execution device; the sensors are deployed in a biological production environment to collect environmental and biological data; the processor is communicatively connected to the sensors; the execution device is communicatively connected to the processor to receive control commands and execute operations to regulate the biological production environment; wherein... The sensor includes: Physical environment sensor for measuring at least one physical environment parameter among temperature, humidity, and gas concentration; The first biosensor is used to collect data reflecting the macroscopic physiological and metabolic state of biological populations in a non-contact manner; The second biosensor is used to collect data reflecting the macroscopic behavior and welfare status of biological groups; The processor includes: A physical-biological coupling model is used to describe the interaction between the environment and organisms, and includes at least one biological source parameter that varies over time; The model calibration channel is used to receive data from the first biosensor, calculate the current value of the biological source parameter, and update the physical-biological coupling model using the current value. The target optimization channel is used to receive data from the second biosensor, calculate a quantified biological welfare index, and dynamically adjust the environmental target setpoint based on the difference between the biological welfare index and the target value. The predictive controller uses the updated physical-biological coupling model and the adjusted environmental target setpoint to generate control commands for the actuator through optimized calculations. The execution equipment includes at least one of a fan, a heater, a humidifier, a cooling device, and a lighting device.

[0019] The beneficial effects of this invention are: Compared with the prior art, the present invention has the following significant advantages and positive effects: 1. Excellent adaptability: The system of this invention can perceive the real state of organisms in real time, freeing the control model and target from static and rigidity, and greatly improving the system's adaptability and robustness to various disturbances (such as sudden weather changes and biological stress).

[0020] 2. Higher control precision and resource efficiency: Due to the more accurate model and more reasonable objectives, the system of this invention can maintain the environment in the most favorable state for organisms with lower control costs (energy consumption, water consumption), avoiding excessive or insufficient regulation.

[0021] 3. Significantly improve biological welfare and production performance: This invention directly links control decisions to quantified biological welfare indicators, which can minimize environmental stress and improve animal health, weight gain rate, or plant yield and quality.

[0022] 4. Strong framework versatility: The system architecture and method of this invention is a general platform. By replacing different Class A and Class B sensors, it can be easily applied to the breeding of different animals such as pigs, chickens, and cattle, and even aquaculture and greenhouse cultivation, with great scalability and application prospects. Attached Figure Description

[0023] Figure 1 This is a block diagram of the architecture of an adaptive environmental control system based on multimodal biofeedback. Figure 2 The flowchart shows an adaptive environment control method based on multimodal biological feedback. Figure 3 This is a schematic diagram of a specific implementation method in a precision pigsty environment. Detailed Implementation

[0024] The present invention will now be described in further detail with reference to specific embodiments, so that those skilled in the art can understand it.

[0025] Example 1 An Adaptive Environmental Control System Based on Multimodal Biofeedback The system in this embodiment includes a sensing layer, a processor, and an execution device; the sensors are deployed in the biological production environment to collect environmental and biological data; the processor is communicatively connected to the sensors; the execution device is communicatively connected to the processor to receive control commands and execute operations to regulate the biological production environment; wherein, 1. The sensor includes: (1) A physical environment sensor for measuring at least one physical environment parameter among temperature, humidity and gas concentration; (2) A first biosensor, used to collect data reflecting the macroscopic physiological and metabolic state of a biological population in a non-contact manner; (3) A second biosensor is used to collect data reflecting the macroscopic behavior and welfare status of biological groups.

[0026] 2. The processor includes: (1) A physical-biological coupling model is used to describe the interaction between the environment and organisms, and includes at least one biological source parameter that varies over time; (2) Model calibration channel, used to receive data from the first biosensor, calculate the current value of the biological source parameter, and use the current value to update the physical-biological coupling model; (3) Target optimization channel, used to receive data from the second biosensor, calculate a quantitative biological welfare index, and dynamically adjust the environmental target set point based on the difference between the biological welfare index and the target value; (4) Predictive controller, using the updated physical-biological coupling model and the adjusted environmental target setpoint, generates control commands for the execution device through optimization calculation; 3. The equipment includes at least one of the following: fan, heater, humidifier, cooling equipment, and lighting equipment.

[0027] Example 2 An Adaptive Environmental Control Method Based on Multimodal Biological Feedback This embodiment utilizes the system from Example 1 to implement an adaptive environmental control method based on multimodal biofeedback. In this example, the system is applied to a precise pigsty environment, combined with... Figures 1-3 As shown, the specific steps include: I. Real-time acquisition of physical environmental parameters within the target environment, and biofeedback data reflecting the state of the biological population collected through biosensors of different modalities; the biofeedback data includes at least type I biological data reflecting macroscopic physiological and metabolic states and type II biological data reflecting macroscopic behavioral and welfare states; specifically, 1. The target environment in this embodiment is a nursery pig house with dimensions of 10m (length) × 5m (width) × 2.5m (height), and a total volume V = 125 m³. 3 .

[0028] 2. The biological group consists of 50 nursery pigs, with an initial average weight of 100 piglets. Total number These parameters serve as the initial inputs to the system.

[0029] 3. The first biosensor is a SenseAir K30 non-dispersive infrared (NDIR) sensor with a range of 0-5000ppm and an accuracy of ±30ppm. It is installed in the central area of ​​the pigsty at a height of 1.0m (the height at which the pigs move).

[0030] 4. The second biosensor is a Hikvision DS-2CD3T45-I5 4-megapixel network camera, which is installed on the top and vertically overhead to capture images of the entire pigsty.

[0031] 5. Physical environment sensor: Standard industrial-grade temperature and humidity sensor, accuracy ±0.5℃, ±2%RH.

[0032] All sensor data is collected, aligned, and stored in 1-minute intervals.

[0033] 2. Based on the first type of biological data collected by the first biosensor, the current value of at least one key biological source parameter in the physical-biological coupling model is calculated online through a physiological metabolism conversion algorithm, and the physical-biological coupling model is updated in real time using the current value. The specific steps include: 1. Nighttime stable window screening: Within a fixed period of the day, select the time window where the minimum ventilation rate is continuously and stably operated, the indoor temperature fluctuation is less than the set threshold, and there are no negative slope anomalies in the CO2 concentration sequence as the nighttime stable window. Specifically, in this embodiment, the fixed time period is from 01:00 to 04:00 every day, the threshold is set at ±0.5℃, the minimum ventilation rate is selected and it has been running stably for more than 60 minutes. To ensure data validity, the first time window that meets the above conditions and has a length of not less than 30 minutes is selected as the nighttime stable window.

[0034] 2. Calculate the CO2 production rate based on data within the nighttime stable window; Specifically, the change in CO2 concentration in the pigsty satisfies the following mass balance equation: After discretization, the CO2 production rate can be calculated: Among them, G CO2 V is the CO2 production rate; t is the volume of the biological environment; C is the time. in,t Let C be the indoor CO2 concentration at time t. out,t Let Q be the external CO2 concentration at time t, Δt be the sampling time interval, and Q be the external CO2 concentration at time t. vent For minimum ventilation rate, This represents the indoor CO2 concentration at the previous sampling time.

[0035] 3. Calculate the total heat production of the biological community based on the CO2 production rate; Specifically, according to the Brouwer equation or the simplified Weir energy metabolism equation, metabolic heat production in animals is directly related to gas exchange. For pigs on a specific diet, their respiratory entropy RQ ( The total heat production is approximately constant (RQ = 0.98 in this embodiment). The total heat production is calculated using the following formula. : Where RQ is the respiratory entropy, G CO2 Based on the calculated CO2 production rate, The total heat production of the biological community.

[0036] 4. Update the physical-biological coupling model using total heat production as a time-varying biological source parameter; Specifically, the physical-biological coupling model is a pigsty thermodynamic model, whose state-space equation is described as: x{k+1}=F x k +G u k +Gd Q{pigs,k}; where x k This is the system state vector, containing the indoor temperature; u k The control input vector includes the airflow of the ventilation equipment and the power of the heater; Q{pigs, k} is the total heat production, which serves as a measurable feedforward perturbation input; F, G, Gd are model parameter matrices obtained through system identification or physical derivation.

[0037] The specific technical solutions for heat production estimation implemented in this module, including but not limited to an automatic filtering algorithm for stable nighttime data windows, a calculation method for animal respiratory gas production rate based on CO2 mass balance, and an energy conversion model from gas production rate to metabolic heat production, have been described in detail in the Chinese invention patent application filed by the applicant on September 8, 2025, with application number [2025112738081] "A Real-time Measurement Method for Total Heat Production Rate of Animal Population".

[0038] In short, this module utilizes the principle that, under minimal ventilation conditions, the rate of CO2 accumulation in the barn is directly proportional to the total respiratory metabolic intensity of the pig herd, thereby controlling total heat production. Non-contact online estimation. This invention directly calls the output of this module. This is used for real-time calibration of subsequent physical-biological coupling models.

[0039] Third, based on the second type of biological data collected by the second biosensor, a quantitative biological welfare index is calculated through a behavior pattern recognition algorithm, and the environmental target setpoint of the environmental control system is dynamically adjusted according to the deviation between the biological welfare index and the preset target welfare index. Specifically, the following steps are included: 1. By analyzing top-view images in the second type of biological data, behavioral characteristics of biological groups are extracted, including calculating the clustering index based on the number of connected components in the image and calculating the dispersion index based on the average nearest neighbor distance between individuals; Specifically, this embodiment employs a lightweight computer vision algorithm to ensure low cost and high real-time performance, eliminating the need for complex deep learning training. Inter-frame difference or Gaussian mixture models are used for dynamic background modeling and foreground (pig) segmentation, generating a binarized mask image. Contour detection is then performed on the binarized mask image to calculate the number of connected components ("pig clusters"). .

[0040] The clustering index is calculated as follows: ,in, The number of connected components. Total number of organisms; The dispersion index DI is obtained by normalizing the average nearest neighbor distance of an individual organism, that is, by calculating the average nearest neighbor distance D between all the centroids of the outline. mean The more dispersed the pigs are, the more D mean The larger D is, the more mean After normalization, the dispersion index DI is obtained; 2. Based on the clustering index and the dispersion index, the quantified thermal comfort index (TCI) is calculated. Thermal Comfort Index (TCI) is a comprehensive quantitative indicator reflecting the thermal comfort of pigs. The formula for calculating the TCI is: in, It is the expected dispersion (calibrated value) under comfortable conditions. It is the slope of the sigmoid function. It is the penalty weight for group behavior; the TCI value range is [0,1], and the higher the value, the more comfortable it is.

[0041] Traditional thermal indices such as THI (Temperature-Humidity Index) and PMV (Predicted Mean Vote) are calculated based on physical environmental parameters (temperature, humidity, and wind speed), predicting theoretical comfort levels. The TCI of this invention, however, is based on actual biological behavioral feedback, directly measuring the subjective feelings of the pig herd. When theory and reality differ (e.g., pigs feel cold due to illness), TCI can detect this discrepancy and correct the control target, representing a fundamental leap forward from existing technologies.

[0042] 3. Based on the deviation between the thermal comfort index and the target comfort level, dynamically generate the temperature setpoint; Specifically, a limiting proportional (P) controller is used to generate dynamic setpoints. Among them, error , The target comfort level is 0.8 in this embodiment. As a proportionality coefficient, and simultaneously set The rate of change is limited to ≤0.2℃ / 15min and the absolute range is limited to [24℃, 30℃].

[0043] Fourth, input the updated physical-biological coupling model from step two and the adjusted environmental target setpoints from step three into the predictive control module (MPC), and use the optimization algorithm to calculate the optimal control command for the environmental control equipment in the future period of time. Specifically, the physical-biological coupling model updated in step two is combined with the environmental target setpoints obtained in step three. The environmental target setpoint sequence is formed by superimposing the data. The model updated in S2 and the dynamic setpoints generated in S3 are used together as the model and reference trajectory input for MPC. The optimal control sequence is solved in each control cycle. In each control cycle (15 minutes in this embodiment), Model Predictive Control is used to solve the control sequence of the execution equipment. Model Predictive Control solves the following optimization problem in each control cycle: The meanings of each character are shown in Table 1.

[0044] Table 1 Character Meaning The constraints for model predictive control are as follows: in, The system dynamics constraints defined by the physical-biological coupling model updated in step two. The total heat production is calculated in real time; Control variables were defined to define the operational constraints of environmental control equipment. The upper and lower limits of feasibility.

[0045] 5. Issue optimal control commands to environmental control equipment for execution to adjust the target environment; Steps one through five are executed cyclically to form a continuous closed-loop control.

[0046] Experimental Results: To demonstrate the technical effectiveness of this invention, a 24-hour simulation comparison experiment was conducted. Scenario: Simulating a nighttime cooling process. Control Group: PID control with a fixed temperature setpoint (28℃). Experimental Group: Using the adaptive environmental control method based on multimodal biofeedback of this invention. The results are shown in Table 2.

[0047] Table 2 Experimental Comparison Results Conclusion: The results clearly show that the present invention can significantly reduce system energy consumption while ensuring the thermal comfort of pigs (higher and more stable TCI), achieving a dual improvement in welfare and efficiency, and the technical effect is remarkable.

[0048] It should be noted that those skilled in the art will understand that the above embodiments are merely one of the preferred embodiments of the present invention. Other technical means can be used to implement the various modules of the present invention without departing from the core idea of ​​the invention. For example, the first biosensor can also be a device for measuring animal oxygen consumption, ammonia evaporation rate, or average body temperature of a group; the second biosensor can also be a microphone array that determines stress levels by analyzing sound, or a device that analyzes body surface temperature distribution through infrared thermal imaging. All these variations should fall within the protection scope of the present invention.

[0049] All other parts not described in detail are existing technologies. Although the above embodiments have provided a detailed description of the present invention, they are only some embodiments of the present invention, not all embodiments. People can obtain other embodiments based on these embodiments without creative effort, and these embodiments all fall within the protection scope of the present invention.

Claims

1. An adaptive environmental control method based on multimodal biological feedback, characterized in that, Includes the following steps: S1 collects physical environmental parameters within the target environment in real time, as well as biofeedback data reflecting the state of the biological population collected by biosensors of different modalities; the biofeedback data includes at least a first type of biological data reflecting the macroscopic physiological and metabolic state and a second type of biological data reflecting the macroscopic behavioral and welfare state. S2 Based on the first type of biological data, the current value of at least one key biological source parameter in the physical-biological coupling model is calculated online using a physiological metabolism conversion algorithm, and the physical-biological coupling model is updated in real time using the current value. S3 Based on the second type of biological data, a quantitative biological welfare index is calculated through a behavior pattern recognition algorithm, and the environmental target setpoint of the environmental control system is dynamically adjusted according to the deviation between the biological welfare index and the preset target welfare index. S4 Input the updated physical-biological coupling model from step S2 and the adjusted environmental target setpoint from step S3 into the predictive control module, and calculate the optimal control command for the environmental control equipment in the future period through the optimization algorithm. S5 The optimal control command is sent to the environmental control device for execution to adjust the target environment; Steps S1 to S5 are executed cyclically to form a continuous closed-loop control.

2. The adaptive environment control method based on multimodal biological feedback according to claim 1, characterized in that, The specific steps in step S2 for calculating the current values ​​of key biological source parameters based on the first type of biological data include: S21: Nighttime stable window screening. Within a fixed period of the day, the time window in which the minimum ventilation rate is continuously and stably operated, the indoor temperature fluctuation is less than the set threshold, and the CO2 concentration sequence has no negative slope anomalies is selected as the nighttime stable window. S22: Calculate the CO2 production rate based on the data within the nighttime stable window; S23: Calculate the total heat production of the biological population based on the CO2 production rate; S24: Update the physical-biological coupling model using the total heat production as a time-varying biological source parameter.

3. The adaptive environment control method based on multimodal biological feedback according to claim 2, characterized in that, The physical-biological coupling model is a pigsty thermodynamic model, and its state-space equation is described as: x{k+1}=F x k +G u k +Gd Q{pigs,k}; where x k This is the system state vector, containing the indoor temperature; u k The control input vector includes the airflow of the ventilation equipment and the power of the heater; Q{pigs, k} is the total heat production, which serves as a measurable feedforward perturbation input; F, G, Gd are model parameter matrices obtained through system identification or physical derivation.

4. The adaptive environment control method based on multimodal biological feedback according to claim 3, characterized in that, In step S22, the CO2 production rate is calculated according to the following formula: ； Where V is the volume of the biological environment, t is time, and C is... in,t Let C be the indoor CO2 concentration at time t. out,t Let Q be the external CO2 concentration at time t, Δt be the sampling time interval, and Q be the external CO2 concentration at time t. vent For minimum ventilation rate, G CO2 This represents the CO2 production rate. This represents the indoor CO2 concentration at the previous sampling time.

5. The adaptive environment control method based on multimodal biological feedback according to claim 4, characterized in that, In step S23, the total heat production is calculated according to the following formula. : Where RQ is the respiratory entropy, G CO2 The CO2 production rate calculated according to claim 4, The total heat production of the biological community.

6. The adaptive environment control method based on multimodal biological feedback according to claim 1, characterized in that, The objective optimization in step S3 specifically includes the following sub-steps: S31: By analyzing the top view images in the second type of biological data, the behavioral characteristics of the biological population are extracted, including calculating the clustering index based on the number of connected components in the image and calculating the dispersion index based on the average nearest neighbor distance between individuals; S32: Calculate the quantified thermal comfort index based on the clustering index and the dispersion index; S33: Dynamically generate a temperature setpoint based on the deviation between the thermal comfort index and the target comfort level.

7. The adaptive environment control method based on multimodal biological feedback according to claim 6, characterized in that, The calculation method for the clustering index in step S31 is as follows: ,in, The number of connected components. The total number of organisms; the dispersion index DI is obtained by normalizing the average nearest neighbor distance of individual organisms; In step S32, the formula for calculating the Thermal Comfort Index (TCI) is as follows: in, It is the expected dispersion under comfortable conditions. It is the slope of the sigmoid function. It is the penalty weight for group behavior; the TCI value range is [0,1], and the higher the value, the more comfortable it is.

8. The adaptive environment control method based on multimodal biological feedback according to claim 6, characterized in that, In step S4, the physical-biological coupling model updated in step S2 is compared with the dynamic setpoint obtained in step S3. The environmental target setpoint sequence is formed by superposition, and model predictive control is used to solve the control sequence of the execution equipment in each control cycle; the model predictive control solves the following optimization problem in each control cycle: To optimize the solution variables, k is the discrete time step index, and N is the prediction time domain length. For temperature tracking weighting coefficients, For the predicted future indoor temperature, For dynamic target temperature setpoint, Energy consumption cost weighting coefficient This is a power consumption function.

9. The adaptive environment control method based on multimodal biological feedback according to claim 8, characterized in that, The constraints of the model predictive control are as follows: in, The system dynamics constraints defined for the physical-biological coupling model updated in step S2. The total heat production is calculated in real time; Control variables were defined to define the operational constraints of environmental control equipment. The upper and lower limits of feasibility.

10. An adaptive environmental control system based on multimodal biological feedback, characterized in that, The system includes: sensors, a processor, and an execution device; the sensors are deployed in the biological production environment for collecting environmental and biological data; the processor is communicatively connected to the sensors; the execution device is communicatively connected to the processor for receiving control commands and executing operations to regulate the biological production environment; wherein, The sensor includes: Physical environment sensor for measuring at least one physical environment parameter among temperature, humidity, and gas concentration; The first biosensor is used to collect data reflecting the macroscopic physiological and metabolic state of biological populations in a non-contact manner; The second biosensor is used to collect data reflecting the macroscopic behavior and welfare status of biological groups; The processor includes: A physical-biological coupling model is used to describe the interaction between the environment and organisms, and includes at least one biological source parameter that varies over time; The model calibration channel is used to receive data from the first biosensor, calculate the current value of the biological source parameter, and update the physical-biological coupling model using the current value. The target optimization channel is used to receive data from the second biosensor, calculate a quantified biological welfare index, and dynamically adjust the environmental target setpoint based on the difference between the biological welfare index and the target value. The predictive controller uses the updated physical-biological coupling model and the adjusted environmental target setpoint to generate control commands for the actuator through optimized calculations. The execution equipment includes at least one of a fan, a heater, a humidifier, a cooling device, and a lighting device.

Images