Method for regulating livestock house environment based on non-contact livestock respiration rate measurement

Abstract

The present application relates to the field of intelligent breeding and livestock house environment control, and particularly relates to a livestock house environment regulation method based on non-contact livestock and poultry respiration rate measurement. The method acquires synchronous visible light image sequences, synchronous thermal imaging image sequences, environment parameter data and regulation device operation state data of each environment regulation partition in the livestock house; identifies livestock and poultry individuals and determines nose measurement areas and chest and abdomen measurement areas; determines individual respiration rates based on nose thermal plume periodic changes and chest and abdomen periodic displacements; determines target subgroups and reference subgroups according to individual respiration rate deviation degrees; executes a trial regulation sequence on candidate regulation devices, acquires respiration rate change amplitudes, effective time delays and recovery processes, and constructs a respiration response matrix; and then determines formal regulation combinations and their execution sequences according to the respiration response matrix, environment parameters, operation boundaries and resource consumption constraints, and updates the respiration response matrix in a verification period. The method is suitable for livestock house scenes such as dairy cow houses, pig houses and poultry houses.

Description

Technical Field

[0001] This invention relates to the field of smart farming and livestock housing environmental control technology, and in particular to a livestock housing environmental control method based on non-contact livestock and poultry respiratory rate measurement. Background Technology

[0002] Environmental factors such as temperature, humidity, airflow, and harmful gases in livestock and poultry houses directly affect the growth, health, and production performance of livestock and poultry. Therefore, monitoring and regulating the livestock house environment has become an important technical direction in modern livestock farming environmental control. A common existing approach involves installing temperature and humidity sensors or monitoring nodes in the livestock house to collect environmental parameters. Then, based on preset thresholds, environmental indices, or control rules, the operation of control equipment such as fans, sprinklers, and wet curtains is adjusted to maintain the indoor environment within a relatively suitable range.

[0003] For example, patent CN109407624A discloses a cattle shed environmental control system and method based on the temperature and humidity index (THI). This patent acquires environmental data such as temperature and humidity by setting up data collection nodes at different locations in the cattle shed, statistically analyzes and visualizes the data from each node, calculates the THI based on the environmental data, and then regulates the cattle shed control equipment to maintain stable temperature and humidity and alleviate heat stress in dairy cows. The key technical point of this patent is: based on the collected environmental parameters inside the shed, the THI characterizes the risk of heat stress and implements environmental control accordingly.

[0004] However, the aforementioned existing technologies primarily rely on environmental parameters such as temperature and humidity, along with their calculated temperature and humidity indices, for regulation. Essentially, their control methods are still driven by environmental parameters. In actual livestock farming scenarios, the immediate physiological responses of livestock populations within different environmental control zones to the same control device may vary. Relying solely on environmental parameters or indices cannot directly reflect the actual respiratory stress state of individual livestock within a given zone, nor can it distinguish the differences in the actual effects of different candidate control devices on respiratory recovery in the current zone and at the current moment. In other words, while existing technologies can achieve livestock housing regulation based on environmental parameters, they lack specific technical solutions for further identifying the actual responses of different candidate control devices to respiratory recovery based on the real-time physiological state of the livestock population within a given zone, and thus determining a more suitable formal control combination.

[0005] Therefore, the main technical problem to be solved by this application is: how to provide a livestock house environment control method based on non-contact livestock and poultry respiratory rate measurement, so that when the respiratory state of livestock and poultry is abnormal, the actual respiratory response of the livestock and poultry group in the current environmental control zone to different candidate control devices can be identified, and the formal control combination can be determined and updated accordingly. Summary of the Invention

[0006] To overcome the aforementioned technical deficiencies, the present invention aims to provide a method for regulating livestock housing environment based on non-contact livestock respiratory rate measurement. This invention first obtains the real-time respiratory status of the livestock population within the current environmental regulation zone based on non-contact livestock respiratory rate measurement. Then, it uses a trial regulation sequence to identify the actual response of different candidate regulation devices to the respiratory recovery of the livestock population. Based on this, a respiratory response matrix is ​​constructed to determine and update the formal regulation combination, thereby solving the technical problem in the prior art of being unable to implement adaptive environmental regulation based on the actual respiratory response of the livestock population within the current zone.

[0007] This invention discloses a method for controlling the environment of livestock houses based on non-contact livestock and poultry respiratory rate measurement, comprising the following steps: S1. Acquire synchronous visible light image sequences, synchronous thermal imaging image sequences, environmental parameter data, and operating status data of control devices for each environmental control zone in the livestock house. The environmental parameter data shall include at least air temperature, relative humidity, air velocity, carbon dioxide concentration, and ammonia concentration. S2. Identify individual livestock and poultry based on synchronous visible light image sequences, and determine the spatial location, body surface orientation, nose measurement area, and chest and abdomen measurement area of ​​each individual livestock and poultry. S3. For each individual livestock and poultry, the individual respiratory rate is determined based on the periodic changes of the exhaled thermal plume in the nasal measurement area and the periodic displacement of the chest and abdomen measurement areas, and the individual respiratory rate is assigned to the corresponding environmental control zone. S4. Based on the degree of deviation between the individual respiratory rate and the corresponding historical stable period respiratory rate, select the target subgroup in the partition to be regulated, and select the reference subgroup in the partition adjacent to the partition to be regulated that has not received the current trial regulation action. S5. When the respiratory status of the target subgroup exceeds the preset target range, a trial control sequence is executed on the control area. The trial control sequence includes applying short-term adjustment actions to multiple candidate control devices individually, and simultaneously collecting respiratory rate changes of the target subgroup and the reference subgroup during each short-term adjustment action. S6. Based on the respiratory rate change, onset delay and recovery process of the target subgroup relative to the reference subgroup, determine the respiratory response parameters of each candidate control device in the current environmental control zone, and construct the respiratory response matrix of the current environmental control zone by combining the correspondence between individual livestock and poultry and airflow path. S7. Based on the respiratory response matrix, current environmental parameters, operating boundaries of the control device, and resource consumption constraints, determine the formal control combination and its execution sequence; S8. Perform environmental regulation on the control zone according to the formal regulation combination, compare the measured respiratory rate recovery results with the predicted results within the preset verification period, update the respiratory response matrix when the update conditions are met, and redetermine the formal regulation combination.

[0008] Preferably, in step S4, the target subgroup consists of livestock individuals located within the control zone, of the same breed and growth stage, and whose respiratory rate deviation is greater than a preset deviation threshold; the reference subgroup consists of livestock individuals located within adjacent zones, of the same breed and growth stage, and who have not received the action of the candidate control device that has been applied to them in the current trial control sub-stage.

[0009] Preferably, the method further includes the step of determining the airflow exposure coefficient of individual livestock and poultry; the airflow exposure coefficient is determined based on the spatial location of the individual livestock and poultry, the orientation of its body surface, the location of the air intake component, the location of the air exhaust component, the location of the local directional air supply device, and the interpolation result of the local airflow velocity, and is used to characterize the relative intensity of the airflow effect received by the corresponding individual livestock and poultry from each candidate control device.

[0010] Preferably, in step S3, the individual respiratory rate is output only for valid measurement individuals, and the valid measurement individuals meet the following conditions: The exhaled thermal plume periodic signal in the nasal measurement area and the periodic displacement signal in the chest and abdomen measurement areas both meet the preset periodic conditions. The difference between the first respiratory rate obtained based on the exhaled hot plume period signal and the second respiratory rate obtained based on the period displacement signal is not greater than a preset consistency threshold. Furthermore, continuous tracking and identification of the same livestock and poultry individual were maintained during the trial control sequence and verification period.

[0011] Preferably, in step S5, the trial regulation sequence is implemented in a single device, single variable, and segmented execution manner. In each trial regulation sub-stage, only one regulation variable of a candidate regulation device is changed, while the other candidate regulation devices remain unchanged. Furthermore, a recovery interval is set between two adjacent trial regulation sub-stages until the difference in respiratory rate between the target subgroup and the reference subgroup returns to a preset stable range before the next trial regulation sub-stage is executed.

[0012] Preferably, the candidate control device includes at least two of the following: a variable frequency fan, an air intake component, a wet curtain device, a spray device, a roof sprinkler device, and a local directional air supply device; and, a first test intensity and a second test intensity are sequentially applied to at least one candidate control device, wherein the second test intensity is applied when the target subgroup produces a differential change in respiration rate relative to the reference subgroup under the first test intensity.

[0013] Preferably, in step S6, the respiratory response parameters include at least the respiratory rate decrease magnitude, onset delay, recovery half-life, and respiratory rate decrease efficiency per unit resource; the respiratory response parameters are determined by the differential changes between the target subgroup and the reference subgroup in the corresponding trial control sub-stage; when the differential change direction of the same candidate control device under the first trial intensity and the second trial intensity is inconsistent, or the onset delay exceeds the preset maximum allowable duration, the candidate control device is marked as a currently invalid device and removed from the determination process of the formal control combination.

[0014] Preferably, when synchronous respiratory rate changes occur in adjacent zones during the execution of the trial control sequence, a zone coupling correction step is also included. The zone coupling correction step determines the zone coupling coefficient based on the state of the shared air intake component, the state of the shared air exhaust component, and the state of the shared local directional air supply device, and then constructs the respiratory response matrix after deducting the coupling component in the differential change.

[0015] Preferably, before executing step S5, an initialization step for the respiratory response matrix is ​​also included. The initialization step selects historical records from the historically verified control records that are consistent with the current external enthalpy range, livestock and poultry species, growth stage, and stocking density range to form an initial respiratory response matrix, and uses the respiratory response parameters obtained from the current trial control sequence to correct the initial respiratory response matrix.

[0016] Preferably, step S7 further includes a slow recovery subgroup determination step; the slow recovery subgroup is determined based on historically verified control records, and the corresponding livestock and poultry individuals meet the condition that the recovery half-life is in the preset high quantile range and is in the same airflow exposure coefficient range as the current partition; the formal control combination is used to restore the respiratory rate of the slow recovery subgroup to the preset target range as the control target.

[0017] Preferably, in step S7, the formal control combination is determined in the following manner: First, based on the efficiency of respiratory rate reduction per unit resource, the candidate control devices that are not currently marked as invalid devices are sorted among the current candidate control devices, and the candidate control device with the best ranking is selected as the first formal control action. Then, based on the respiratory response matrix, predict the remaining respiratory rate deviation after the first formal control action is executed; Subsequently, based on the remaining respiratory rate deviation, the second formal control action and subsequent formal control actions are determined sequentially until the respiratory rate of the predicted target subgroup recovers to the preset target range or reaches the operating boundary of the control device.

[0018] Preferably, when the formal control combination includes a humidification control device, the humidification control device is activated only when the current humidity margin and the predicted ammonia dilution demand simultaneously meet the preset activation conditions; when the preset activation conditions are not met, a candidate control device that does not increase humidity and is not marked as a currently invalid device is used to replace the humidification control device.

[0019] Preferably, in step S8, during the verification period, based on the continuously tracked livestock individuals constituting the target subgroup, the predicted onset time delay, the predicted decrease in respiratory rate, and the predicted recovery half-life are compared with the corresponding measured values; when at least two of them exceed the corresponding error threshold, the trial regulation sequence is re-executed.

[0020] Preferably, when the number of valid individuals measured during the verification period is lower than the preset minimum number, the update of the respiratory response matrix is ​​paused, and the most recently verified respiratory response matrix is ​​used, combined with the current changes in environmental parameters, to maintain or limit the adjustment of the formal control combination.

[0021] Preferably, after each formal regulation cycle is completed, the trial regulation sequence, respiratory response parameters, respiratory response matrix, formal regulation combination, and verification results are recorded; only when the verification results meet the preset accuracy conditions are the corresponding records written into the historically verified regulation record library, and the records in the historically verified regulation record library are called when the respiratory response matrix is ​​initialized subsequently.

[0022] Compared with existing technologies, the above technical solution has the following advantages: 1. This invention first obtains the real-time respiratory status of livestock and poultry groups within the current environmental control zone based on non-contact livestock and poultry respiratory rate measurement, then identifies the actual effect of different candidate control devices on the respiratory recovery of the livestock and poultry groups through trial control sequences, and determines the formal control combination and its execution order accordingly. This allows livestock housing environmental control to no longer rely solely on fixed mode switching of environmental parameters or indices such as temperature and humidity, but to implement adaptive control based on the actual physiological response of livestock and poultry groups within the current environmental control zone. Therefore, it can significantly improve the matching degree between environmental control and the actual needs of livestock and poultry.

[0023] 2. By setting up a target subgroup and a reference subgroup, and identifying the trial regulation results based on the differential respiratory changes between the two, this invention can effectively reduce the interference of natural environmental fluctuations, synchronous changes in adjacent zones, and individual differences on the regulation judgment results, thereby improving the accuracy of identifying the effects of different candidate regulation devices and making the determination of formal regulation combinations more targeted and reliable.

[0024] 3. This invention constructs and updates a respiratory response matrix to structurally characterize response parameters such as the respiratory rate decrease magnitude, onset delay, recovery half-life, and respiratory rate decrease efficiency per unit resource corresponding to candidate control devices, and directly uses them in the decision-making process of formal control combinations. Therefore, it can dynamically adjust the control strategy according to the current partition, the current population, and the current environmental boundary conditions, and has strong adaptive and continuous optimization capabilities.

[0025] 4. By setting a single device, a single variable, and a segmented trial control sequence, combined with a mechanism to eliminate currently ineffective devices, this invention can distinguish the actual effectiveness of different candidate control devices in the current environmental control zone. This avoids including control devices that do not significantly promote the recovery of the current respiratory state of livestock and poultry or have a delayed response in the formal control combination. Therefore, it can reduce ineffective control, reduce the frequency of ineffective switching, and improve the efficiency of control resource utilization.

[0026] 5. By introducing airflow exposure coefficient, zonal coupling correction, and slow recovery subgroup identification mechanism, this invention can further consider factors such as the spatial location of individual livestock and poultry, body surface orientation, airflow path, and group recovery differences. This not only takes into account the overall recovery effect within the environmental control zone, but also increases the attention paid to slower-recovering groups, thereby improving the precision of livestock housing environmental control.

[0027] 6. In the formal control process, the present invention sets dual constraints on the humidification control device, namely humidity margin and ammonia dilution requirement, and adopts a maintenance or amplitude limit correction strategy when the number of effective measurement individuals is insufficient. Therefore, it can reduce unnecessary humidity increase, water consumption and energy consumption while ensuring the recovery effect of livestock and poultry respiratory state, and improve the operational stability of the system in scenarios with insufficient measurement coverage, changes in boundary conditions or local disturbances.

[0028] 7. This invention adopts a non-contact method for measuring the respiratory rate of livestock and poultry, which eliminates the need to install contact sensors on the surface of livestock and poultry. This reduces interference with the normal activities and feeding management of livestock and poultry, and makes it easy to deploy and implement in various livestock housing scenarios such as dairy cow sheds, pig sheds, and poultry sheds. Therefore, it has good engineering application value and promotion prospects. Attached Figure Description

[0029] Figure 1 This is a schematic diagram of the livestock house environment control system based on non-contact livestock and poultry respiratory rate measurement according to the present invention.

[0030] Figure 2 This is a flowchart of the method for the livestock housing environment control system based on non-contact livestock and poultry respiratory rate measurement according to the present invention.

[0031] Figure 3 A schematic diagram for constructing a regulatory sequence and respiratory response matrix.

[0032] Figure 4 This is a schematic diagram for formal adjustment, verification, and updating.

[0033] Figure 5 This is a schematic diagram showing the spatial distribution of environmental control zones, airflow paths, and target / reference subgroups.

[0034] Figure 6 This is a schematic diagram illustrating the changes in respiratory rate between the target subgroup and the reference subgroup before and after regulation. Detailed Implementation

[0035] The following is combined Figures 1 to 6 The present invention will be further described in detail below. It should be understood that the following embodiments are used to clearly and completely illustrate the technical solution of the present invention, and are not intended to limit the scope of protection of the present invention. Equivalent adjustments made by those skilled in the art to the arrangement of the acquisition equipment, the deployment of the controller, the details of the algorithm implementation, the parameter range, and the selection of the control device without departing from the concept of the present invention should all be understood as falling within the technical concept of the present invention.

[0036] The core problem addressed by this implementation method is that it does not directly switch between fixed modes based on environmental parameters or indices, nor does it simply replace temperature and humidity indices with livestock respiratory rate as a new fixed threshold control variable. Instead, it first obtains the real-time respiratory status of livestock within the current environmental control zone based on non-contact livestock respiratory rate measurement. Then, it identifies the actual effect of different candidate control devices on the respiratory recovery of livestock within the current zone through trial control sequences, and determines the formal control combination and its execution order accordingly, thereby improving the matching degree between environmental control and the actual physiological needs of the current zone. The following detailed description uses a lactating dairy cow shed as the main implementation scenario, followed by supplementary implementation methods for pig sheds and poultry sheds to illustrate the universal adaptation path of this invention for "livestock and poultry".

[0037] In this embodiment, a lactating dairy cow shed is used as an example application scenario. The lactating dairy cow shed is divided along its length into a first environmental control zone, a second environmental control zone, a third environmental control zone, and a fourth environmental control zone. Each environmental control zone is equipped with a visible light acquisition unit, a thermal imaging acquisition unit, an environmental parameter acquisition unit, and a control device. The control device includes a variable frequency fan, an air intake component, a wet curtain device, a spray device, a roof sprinkler system, and a localized directional air supply device. All control devices in each environmental control zone are connected to the same controller, which is communicatively connected to an image processing server, an edge computing unit, and a historical data storage unit, thereby forming a system as follows: Figure 1 The system architecture shown. Figure 5The spatial distribution relationships of environmental control zones, main airflow paths, local directional airflow paths, and target subgroups and reference subgroups are further illustrated, thus providing a geometric basis for subsequent calculation of airflow exposure coefficients and zonal coupling correction.

[0038] In a preferred embodiment, the visible light acquisition unit is installed below the top beam of the cattle shed at a height of 4.2m, with a viewing angle covering the feeding passage and lying area of ​​the corresponding environmental control zone. The image sampling frequency of the visible light acquisition unit is set to 25 frames / s. The thermal imaging acquisition unit is installed adjacent to the visible light acquisition unit, with a sampling frequency of 10 frames / s. The environmental parameter acquisition unit includes a temperature sensor, a humidity sensor, an air velocity sensor, a carbon dioxide concentration sensor, and an ammonia concentration sensor, with the environmental parameter sampling period for each sensor set to 5s. The operating status data of the control device is fed back in real time by the controller and the field actuators, including the variable frequency fan speed, the opening angle of the air inlet component, the operating status of the circulating water pump of the wet curtain device, the duty cycle of the spray device, the valve opening of the roof sprinkler device, and the air outlet speed of the local directional air supply device. To ensure data comparability between different environmental control zones, sensors of corresponding types in each zone are calibrated using a unified calibration standard. Specifically, the calibration error of temperature sensors is controlled within ±0.2℃, the calibration error of humidity sensors is controlled within ±2%RH, the calibration error of air velocity sensors is controlled within ±0.05m / s, and the carbon dioxide concentration and ammonia concentration sensors are calibrated for zero point and span using standard gases, respectively.

[0039] After the acquisition system is deployed, the visible light acquisition unit and the thermal imaging acquisition unit are first synchronized with time stamps, and spatial calibration is completed using fixed calibration boards set up in each environmental control zone. After calibration, a mapping relationship is established between the visible light image coordinates and the thermal imaging image coordinates, ensuring that the subsequent nasal measurement area can be accurately mapped to the corresponding position in the thermal imaging image. At the same time, the timestamps of each environmental parameter acquisition unit are aligned with the image timestamps to facilitate the subsequent creation of a corresponding dataset of "image-physiology-environment-control action" at the same moment. Preferably, the time synchronization error between the visible light image and the thermal imaging image is controlled within 100ms, and the synchronization error between the timestamps of each environmental parameter and the image timestamp is controlled within 500ms, in order to reduce the adverse effects of asynchronous sampling on respiratory rate measurement and response parameter identification.

[0040] like Figure 2As shown, the method of the present invention first performs a data acquisition step. The controller synchronously acquires visible light image sequences, thermal imaging image sequences, environmental parameter data, and control device operating status data for each environmental control zone according to a unified time reference. To suppress the impact of on-site dust, backlight, and partial occlusion on recognition accuracy, the visible light images undergo brightness equalization, noise filtering, and motion blur correction before entering the recognition process; the thermal imaging images undergo bad pixel repair and temperature drift correction; and the environmental parameter data undergoes outlier removal through a sliding time window. Preferably, the length of the environmental parameter sliding time window can be set to 30s to 120s; in this embodiment, 60s is used. When an environmental parameter deviates from the median by more than three times the median absolute deviation within the sliding time window, it is determined to be an outlier and is removed.

[0041] The subsequent steps involve individual livestock identification and measurement area determination. The image processing server identifies the contour, center point, trunk axis, and key head points of each cow based on a continuous visible light image sequence, and assigns a unique continuous tracking identifier to each cow using a continuous tracking algorithm. After identifying the key head points, the nose measurement area is determined in the visible light image based on the geometric relationship between the nasal region and the facial contour, and this nose measurement area is mapped to the thermal imaging image using the aforementioned spatial calibration relationship. Simultaneously, the chest and abdomen measurement areas are determined in the visible light image based on the image boundaries of the cow's anterior chest edge, rib arch line, and lower abdominal edge. To facilitate subsequent calculation of airflow intensity, the spatial position and body orientation of each cow within the environmental control zone are also recorded simultaneously. The spatial position is represented by a two-dimensional position in the zone coordinate system, and the body orientation is determined by the trunk's main axis direction. For continuous tracking identifiers, this implementation further stipulates that: if the same individual loses no more than 8 consecutive frames, it will be re-identified by proximity location, torso size and body surface texture features and the original continuous tracking identifier will be maintained; if it loses more than 8 consecutive frames, it will be determined as tracking interruption, and the individual will no longer be included in the target subgroup, reference subgroup or verification individual range in the current trial control cycle, thereby ensuring that the comparison between the previous and next time periods is based on the same batch of comparable objects.

[0042] In this embodiment, to uniformly illustrate the hierarchical relationship between the various signals, the raw temperature data acquired within the nasal measurement area is first used to form a nasal average temperature time series. Then, a thermal plume periodic signal with respiratory cycle characteristics is extracted from this nasal average temperature time series. Finally, the exhaled thermal plume periodic change is characterized based on the thermal plume periodic signal. Therefore, the nasal average temperature time series is the raw input, the thermal plume periodic signal is the periodic signal extracted from the raw input, and the exhaled thermal plume periodic change is a respiratory-related change feature further characterized by the thermal plume periodic signal.

[0043] In a preferred embodiment, the individual respiratory rate is output only to dairy cows that meet the measurement reliability conditions. The measurement reliability conditions refer to the fact that the periodic changes in the exhaled thermal plume in the nasal measurement area and the periodic displacement in the chest and abdomen measurement area both meet preset periodicity conditions, and the difference between the first respiratory rate and the second respiratory rate calculated based on the two types of signals is not greater than a preset consistency threshold. Furthermore, during the trial control sequence and verification period, the same dairy cow is continuously tracked without interruption. If continuous tracking is interrupted, the nasal measurement area is obscured, the thermal plume period disappears, there is severe misclassification of the chest and abdomen boundaries, or the deviation between the two types of respiratory rates is too large, the corresponding individual will temporarily not participate in the physiological response identification of the current control cycle. Preferably, the preset consistency threshold can be set to 2 to 6 times / min, and in this embodiment, it is set to 4 times / min; the preset periodicity conditions can be determined jointly by the frequency domain main peak energy ratio and the time domain peak spacing stability, for example, the frequency domain main peak energy ratio is not less than 0.55, and the peak spacing variation coefficient for four consecutive respiratory cycles is not higher than 0.20.

[0044] In the specific calculation process, the first respiratory rate is extracted based on the thermal imaging temperature sequence of the nasal measurement area. Specifically, the pixel temperature within the nasal measurement area is averaged to obtain the nasal average temperature time series; bandpass filtering and peak-valley identification are performed on the nasal average temperature time series to extract the thermal plume period signal, thereby obtaining the first respiratory rate, which is denoted as . Simultaneously, the second respiratory rate is extracted based on the optical flow displacement and boundary normal displacement of the chest and abdomen measurement regions. Specifically, sub-pixel boundary tracking is performed on the chest and abdomen measurement regions to obtain the displacement time series of the chest and abdomen contours along the normal direction; the displacement time series is smoothed, periodic detected, and the spacing between adjacent respiratory peaks is analyzed to obtain the second respiratory rate, which is denoted as [missing information]. In a preferred embodiment, the individual respiratory rate can be calculated using the following formula, denoted as: :

[0045] in, As the reliability weight of the thermal plume signal, The reliability weights for the chest and abdominal displacement signals are determined based on the thermal imaging temperature peak amplitude, period continuity, and the occlusion ratio of the nasal measurement area; the reliability weights for the chest and abdominal displacement signals are determined based on optical flow stability, contour boundary integrity, and consistency between adjacent periods. When the difference between the first respiratory rate and the second respiratory rate exceeds a preset consistency threshold, the controller determines that the individual is in a current unreliable measurement state and does not output the individual's respiratory rate. This method utilizes both the sensitivity of thermal imaging to exhaled thermal plumes and the direct characterization of respiratory mechanical motion by chest and abdominal displacement, thereby improving the stability of individual respiratory rate measurement. Preferably, both the reliability weights for the thermal plume signal and the chest and abdominal displacement signal are normalized to the range of 0 to 1; for example, when the thermal plume peak amplitude is high, the period is continuous, and the occlusion ratio is low, Approximately 1; when the chest and abdomen contours are complete, the displacement trajectory is stable, and the period intervals are consistent. The values ​​are close to 1. These weights can be determined by empirical rules or calibrated using historically validated datasets.

[0046] After measuring the respiratory rate of multiple dairy cows within each environmental control zone, the controller assigns the respiratory rate of each individual to the corresponding environmental control zone and determines the target subgroup and reference subgroup based on the historical stable respiratory rate of that zone. The historical stable respiratory rate can be the statistical result of the stable respiratory rate within the same time period, similar external enthalpy range, and similar stocking density within the past 7 days. If the current respiratory rate of a dairy cow deviates from the corresponding historical stable respiratory rate by more than a preset deviation threshold, then that cow enters the candidate target subgroup of the current environmental control zone. The controller further screens the dairy cows in the candidate target subgroup for consistency in livestock breed and growth stage, which in this embodiment is manifested as consistency screening of lactation stage, body size grade, and lactation period range, thereby obtaining the target subgroup. Correspondingly, the controller selects dairy cows in adjacent environmental control zones that are also lactating, have similar lactation stages, and have not received the action of the candidate control device that has been applied to them in the current trial control sub-stage, forming the reference subgroup. The purpose of this setup is to ensure that the target subgroup reflects the population in the current control zone that needs priority in restoring respiratory function, while the reference subgroup reflects the natural fluctuation background without the current exploratory control action. Preferably, the preset deviation threshold can be set to 5% to 20%, and in this embodiment, it is set to 10%; the number of reference subgroups is preferably not less than 0.8 times the number of target subgroups to ensure that the differential background is sufficiently stable.

[0047] To correlate the strength of airflow effects on different individual dairy cows with those of various candidate control devices, this embodiment also introduces an airflow exposure coefficient. The airflow exposure coefficient is denoted as... The airflow exposure coefficient is calculated based on the individual cow's spatial location, body orientation, location of air intake components, location of air exhaust components, location of local directional air supply devices, and interpolation results of local airflow velocity. In a preferred embodiment, the airflow exposure coefficient can be calculated using the following formula:

[0048] in, This is the distance normalization factor, used to characterize the distance from an individual dairy cow to the centerline of the main airflow of the corresponding candidate control device; The orientation normalization factor is used to characterize the angular relationship between the direction of the cow's head and chest and the direction of the main airflow. It is a local air velocity normalization factor used to characterize the local wind speed intensity at the location of an individual dairy cow. , and These are empirical weighting coefficients. In this embodiment, , and The values ​​can be 0.35, 0.25, and 0.40, respectively. A higher calculated airflow exposure coefficient indicates that the individual cow is more likely to be directly affected by the airflow from the corresponding candidate control device. By establishing the airflow exposure coefficient, the respiratory response obtained from subsequent trial control can be linked to the spatial airflow path, providing a basis for constructing a respiratory response matrix. Preferably, the airflow exposure coefficient can be divided into low-exposure, medium-exposure, and high-exposure intervals based on historical data for subsequent slow-recovery subgroup screening and priority determination of formal control combinations. For example, the intervals can be divided according to the 33% and 67% quantiles of all samples.

[0049] After determining the target subgroup, reference subgroup, and airflow exposure coefficient, the controller enters the trial control phase. This trial control phase does not directly employ the predetermined cooling mode, but rather... Figure 3As shown, short-term adjustment actions are sequentially applied to candidate control devices in a "single device, single variable, segmented execution" manner. For example, in the first environmental control zone, the controller first only increases the frequency of the variable frequency fan, while keeping the opening of the air intake component, the circulating water volume of the wet curtain device, the duty cycle of the spray device, the valve opening of the roof sprinkler device, and the air outlet speed of the local directional air supply device unchanged. After completing this trial control sub-stage, the controller does not immediately proceed to the next device trial, but sets a recovery interval until the difference in respiration rate between the target subgroup and the reference subgroup returns to the preset stable range before proceeding to the next trial control sub-stage. The setting of the recovery interval can effectively reduce the influence of the residual effect of the previous sub-stage on the identification results of the subsequent sub-stage, making the independent role of different candidate control devices clearer. Preferably, the duration of a single trial control sub-stage can be set to 2 min to 6 min, and in this embodiment, the variable frequency fan and the local directional air supply device are set to 3 min to 4 min; the recovery interval can be set to 3 min to 10 min, and in this embodiment, it is set to 5 min. When the difference in respiratory rate between the target subgroup and the reference subgroup falls within ±1.0 breaths / min for 60 consecutive seconds, it is determined that the target subgroup has returned to the preset stable range.

[0050] In one specific embodiment, a variable frequency fan is used as the first candidate control device. Its first test intensity is set to increase from 35Hz to 40Hz for 3 minutes. If, under this first test intensity, the target subgroup shows a differential change in respiration rate relative to the reference subgroup, a second test intensity is applied to the variable frequency fan, increasing the frequency from 40Hz to 45Hz for another 3 minutes. A local directional air supply device is used as the second candidate control device. Its first test intensity is an increase in outlet air velocity from 1.2m / s to 1.8m / s for 4 minutes. If, under this first test intensity, a differential change in respiration rate is also observed, a second test intensity is applied, increasing the outlet air velocity to 2.2m / s. Air inlet components, evaporative cooling pads, spray devices, and roof sprinkler systems can be configured with similar test intensities and durations. Applying a first test intensity and a second test intensity sequentially to at least one candidate control device helps identify its response direction, onset speed, and resource utilization efficiency at different intensities, thereby filtering out devices that are incompatible with the current environmental control zone or have unstable responses. Preferably, for the evaporative cooling pad device, the first test intensity can be 50% of the circulating water pump flow rate, and the second test intensity can be 70% of the circulating water pump flow rate; for the spray device, the first test intensity can be 20% duty cycle, and the second test intensity can be 35% duty cycle; for the roof sprinkler device, the first test intensity can be 30% valve opening, and the second test intensity can be 50% valve opening. Each test intensity can be adjusted according to the actual livestock shed volume, the rated capacity of the device, and seasonal operating conditions.

[0051] In each trial control sub-phase, the controller continuously calculates the average respiratory rate of the target subgroup and the average respiratory rate of the reference subgroup. The average respiratory rate of the target subgroup is denoted as... The average respiratory rate of the reference subgroup is denoted as To eliminate the influence of natural environmental fluctuations and synchronous changes in adjacent zones on the judgment results, this implementation method uses differential respiratory change values ​​for response identification. These differential respiratory change values ​​are denoted as... The calculation formula is as follows:

[0052] in, To correspond to the baseline value of the average respiratory rate of the target subgroup at the beginning of the trial regulation sub-phase, This corresponds to the baseline value of the average respiratory rate of the reference subgroup at the start of the trial regulation phase. If the differential respiratory change value is negative and its absolute value continues to increase, it indicates that the current trial regulation action, relative to the natural background fluctuations, can effectively reduce the respiratory rate of the target subgroup. Figure 6 An example is shown, illustrating the changes in respiratory rate of the target subgroup and the reference subgroup before and after trial regulation, where the temporal relationship between the trial regulation sub-phase, onset time delay, and recovery phase can be visually observed.

[0053] Based on differential respiratory rate changes, this implementation method determines respiratory response parameters for each candidate control device. These parameters include at least the magnitude of respiratory rate decrease, onset delay, recovery half-life, and respiratory rate decrease efficiency per unit resource. The magnitude of the respiratory rate decrease is denoted as... It is the minimum absolute value of the differential respiratory change within the corresponding trial regulation sub-stage; the onset time delay is denoted as... It is the time required from the start of the trial control action until the differential respiratory change value first reaches the preset decrease threshold; the recovery half-life is denoted as It is the time required for the differential respiratory rate change to recover from its maximum decrease to half its previous value after the exploratory control action has ceased; the efficiency of respiratory rate decrease per unit resource is denoted as... The calculation formula is as follows:

[0054] in, This refers to the amount of resources consumed by the corresponding candidate control device during the trial control sub-stage. The resources can be electricity consumption, water consumption, or converted comprehensive energy consumption. In actual control, if the differential change direction of the same candidate control device under the first and second trial intensities is inconsistent, or if the effective delay exceeds the preset maximum allowable duration, the candidate control device is marked as an invalid device and removed during the subsequent determination of the formal control combination. This method avoids including devices that are ineffective for the current zone or have a lag risk in the formal control. Preferably, the preset decrease threshold can be set to 1.5 times / min to 4.0 times / min, and in this embodiment, it is 2.0 times / min; the preset maximum allowable duration can be set to 4 min to 10 min, and in this embodiment, it is 7 min. For converted comprehensive energy consumption, it can be obtained by weighted calculation of electricity and water consumption; for example, 1 kWh of electricity and 20 L of water consumption are considered equivalent resource quantities. By removing currently ineffective devices from the formal control combination, we can avoid including control devices that do not significantly promote the recovery of livestock and poultry respiratory status within the current environmental control zone or whose response is lagging, thereby reducing ineffective energy consumption and ineffective action switching.

[0055] To ensure that the respiratory response matrix is ​​not an abstract concept, but rather a data structure that can be directly constructed by those skilled in the art, this embodiment provides further explanation. The respiratory response matrix is ​​denoted as... In a preferred embodiment, the rows of the respiratory response matrix correspond to the currently effective candidate control devices, and the columns correspond to response parameters such as the rate of respiratory rate decrease, onset delay, recovery half-life, respiratory rate decrease efficiency per unit resource, and coupling correction weight, which can be specifically represented as follows:

[0056] in, This represents the number of currently effective candidate control devices. For the first The coupling correction weights correspond to the candidate control devices. If different airflow exposure coefficient ranges or different target subgroup types are further considered, sub-matrices can be constructed for different ranges, or hierarchical label indexes can be added to the matrix. Each element in the respiratory response matrix comes from the measured statistical results of the trial control sub-stage. Therefore, those skilled in the art can directly fill the matrix based on the collected differential respiratory change sequence, energy consumption sequence, and environmental parameter sequence, rather than relying on abstract prior assumptions.

[0057] In complex cattle shed environments, adjacent zones may share air intake components, exhaust components, or local directional air supply devices. Therefore, when a trial control action is performed in a certain environmental control zone, adjacent zones may experience synchronous changes in respiration rate. To address this, this embodiment further implements a zone coupling correction step. The controller determines the zone coupling coefficient based on the status of the shared air intake components, the shared exhaust components, and the shared local directional air supply devices. After subtracting the coupling component from the differential respiration change value, a respiration response matrix is ​​constructed. After coupling correction, the actual independent effects of different candidate control devices within the current environmental control zone can be more accurately identified. Preferably, the zone coupling coefficient can be jointly determined based on the opening degree of the shared device, the distance from the shared device to the boundary of the adjacent zone, and the real-time airflow velocity at the boundary, and normalized to the range of 0 to 1. When the zone coupling coefficient is higher than 0.6, the controller applies a stronger correction to the differential respiration change value of the trial control sub-stage. When the zone coupling coefficient is lower than 0.2, the crosstalk between adjacent zones is considered weak, and only a slight correction is performed.

[0058] In a preferred embodiment, the controller initializes the respiratory response matrix before entering the trial control step. The initial respiratory response matrix is ​​not an empty matrix, but rather constructed from historically validated control records that match the current external enthalpy range, livestock breed, growth stage, and stocking density range. Subsequently, the initial respiratory response matrix is ​​corrected using the respiratory response parameters obtained from the current trial control sequence. This avoids completely refining the identification from zero in each control cycle while ensuring that the current matrix reflects the true response of the current partition, the current population, and the current environment. Preferably, a weighted update method can be used to correct the matrix, for example:

[0059] in, The initial respiratory response matrix is ​​formed by historical verification through records. This is the current matrix obtained from the current trial control sequence. This refers to the historical weighting coefficient. Preferably, It can be set to 0.3 to 0.7, and in this embodiment, it is set to 0.5. This way, it can both inherit historical stable experience and reflect the real-time identification results of the current working conditions.

[0060] To facilitate the standardization of terminology in the following embodiments, in this embodiment, "formal control combination" refers to a set of multiple formal control actions determined for the current environmental control zone within a control cycle; "formal control action" refers to a single control action in the formal control combination, such as increasing the frequency of a variable frequency fan or activating a local directional air supply device; "formal control cycle" refers to the complete closed-loop process from the determination of the formal control combination, through execution, verification, updating or maintenance, until the end of the current control cycle.

[0061] In the formal control combination determination phase, the controller first excludes candidate control devices that have been marked as currently invalid from the current candidate control devices. Then, it performs an initial ranking of the remaining candidate control devices based on the efficiency of respiratory rate reduction per unit resource. The initial ranking result is used to reflect the priority response capability under unit resource consumption. Subsequently, for the candidate control devices with higher initial ranking, the controller further compares their onset delay and recovery half-life. When the difference in respiratory rate reduction efficiency per unit resource between two candidate control devices is less than a preset efficiency difference, the one with the shorter onset delay is selected as the preceding formal control action. When the difference in onset delay is also less than the preset delay difference, the one with the shorter recovery half-life is selected as the preceding formal control action. Preferably, the preset efficiency difference can be set to 0.10 to 0.25, and in this embodiment, it is 0.15; the preset delay difference can be set to 0.5 min to 1.5 min, and in this embodiment, it is 1.0 min. After determining the first formal control action, the controller predicts the remaining respiratory rate deviation after the execution of the first formal control action based on the respiratory response matrix, and continues to determine the second formal control action and subsequent formal control actions. Thus, the respiratory response matrix is ​​not only constructed, but also directly participates in the decision-making process of the formal regulatory combination.

[0062] If a dairy herd with a historically slow recovery exists within the current environmental control zone, the respiratory recovery of this slow-recovery subgroup is prioritized as the control target. The slow-recovery subgroup is not simply represented by a single historical individual, but rather obtained through group feature extraction. Specifically, the controller first selects dairy individuals from historically validated control records whose recovery half-life falls within a preset high quantile interval and is within the same airflow exposure coefficient interval as the current zone. Then, based on the spatial distribution, airflow exposure coefficient interval, respiratory rate recovery trajectory, and body surface orientation of these dairy individuals, clustering is performed to form a slow-recovery group feature template. Subsequently, within the current environmental control zone, a set of individuals from the target subgroup whose matching degree with the slow-recovery group feature template is higher than a preset template matching threshold is found, and this set of individuals is used as the slow-recovery subgroup for the current control cycle. Preferably, the preset high quantile interval can be the top 20%–30% of the recovery half-life sample distribution; in this embodiment, it is the top 25%. The preset template matching threshold can be set to 0.6–0.8; in this embodiment, it is 0.7. This prevents the formal control combination from only meeting the average recovery requirement while ignoring the slowest-recovering group, thereby improving the adaptability of the scheme to individual differences in actual aquaculture scenarios.

[0063] When dealing with humidification control devices, this embodiment further introduces a dual constraint of humidity margin and ammonia dilution requirement. If the formal control combination includes a wet curtain device or a spray device, the controller will only activate the humidification control device when both the current humidity margin and the predicted ammonia dilution requirement meet the preset activation conditions. The humidity margin can be defined as the difference between the current allowable humidity limit of the environmental control zone and the current relative humidity; the ammonia dilution requirement can be determined based on the deviation between the current ammonia concentration and the preset ammonia target concentration. If the humidity margin is insufficient or the predicted ammonia dilution requirement does not meet the conditions, a candidate control device that does not increase humidity and is not currently marked as invalid will replace the humidification control device, such as a variable frequency fan, air inlet component, or local directional air supply device. In this way, the adverse effects of unnecessary humidity increases on the thermal comfort of the cattle shed and the bedding environment can be reduced. Preferably, the humidity margin can be set to allow the humidification control device to be activated only when the RH is not less than 5%, and the ammonia dilution requirement can be set to the current ammonia concentration being higher than the target concentration by 2%. The above indicates that enhanced dilution is required.

[0064] After the formal regulatory package is implemented, such as Figure 4As shown, the controller enters the verification period. During the verification period, the controller compares the predicted onset delay, predicted respiratory rate decrease, and predicted recovery half-life with the corresponding measured values ​​based on the continuously tracked cows constituting the target subgroup. When at least two of the above three parameters exceed the corresponding error threshold, the controller determines that the current respiratory response matrix does not match the actual state of the current partition, and thus re-executes the trial control sequence to update the response identification results of the current partition. If the number of effective measured individuals is lower than the preset minimum number during the verification period, for example, due to cow clustering and obstruction, local strong backlight, or temporary thermal imaging frame loss, the controller suspends the update of the respiratory response matrix and uses the most recently verified respiratory response matrix, combined with the current environmental parameter changes, to maintain or limit the formal control combination, thereby ensuring the stability of the field control and preventing unreasonable and frequent switching due to insufficient short-term measurement coverage. Preferably, the verification period can be set to 8 min to 20 min, and in this embodiment, it is 12 min; the preset minimum number of individuals to be effectively measured can be set to 50% to 70% of the theoretical number of the target subgroup, and in this embodiment, it is 60%; the error thresholds between prediction and actual measurement can be set as follows: onset time delay error threshold 2 min, respiratory rate decrease error threshold 1.5 breaths / min, and recovery half-life error threshold 3 min.

[0065] Regarding the maintenance or limiting correction, this embodiment further quantifies it as follows: When the number of effective measured individuals is less than 60% of the theoretical number of the target subgroup, the update of the respiratory response matrix is ​​paused; when the number of effective measured individuals is further less than 40% of the theoretical number of the target subgroup, not only is the update paused, but a stricter limiting is also applied to each executed variable in the current formal control combination. The limiting means that, relative to the final executed value of the previous formal control cycle, the change in each executed variable does not exceed a preset limiting ratio. Preferably, the preset limiting ratio can be set to 5% to 15%, and in this embodiment, it is 10%. For example, if the operating frequency of the variable frequency fan at the end of the previous formal control cycle is 42Hz, then when the number of effective measured individuals is significantly insufficient, the frequency adjustment range of the variable frequency fan will not exceed ±4.2Hz; if the outlet air velocity of the local directional air supply device is 1.8m / s, then the adjustment range will not exceed ±0.18m / s. In this way, drastic changes in the control quantity can be suppressed when the measurement coverage is insufficient, avoiding unreasonable control jumps caused by temporary measurement gaps.

[0066] After each formal control cycle is completed, the controller records the trial control sequence, respiratory response parameters, respiratory response matrix, formal control combination, and verification results. Only when the verification results meet the preset accuracy conditions is the corresponding record written into the historically verified control record library, and the record in the historically verified control record library is called when the respiratory response matrix is ​​initialized subsequently. This forms a closed-loop evolution mechanism of "historical record initialization - current trial identification - formal control execution - verification correction - effective record storage", enabling the system to gradually improve its adaptability to different seasons, different herd states, and different environmental control zones during long-term operation. Preferably, the preset accuracy conditions can be set as follows: at least two of the three indicators of predicted onset time delay, predicted respiratory rate decline, and predicted recovery half-life meet the corresponding error threshold requirements, and there are no abnormalities in environmental parameter acquisition, image synchronization, or device failure or abnormal shutdown during the execution of the trial control sequence. Furthermore, if the environmental parameter packet loss rate exceeds 5%, the thermal imaging frame loss exceeds 2 seconds, or the feedback status of the control device is inconsistent with the control command during a formal control cycle, the record will not be written into the historically verified control record library, even if the other error indicators meet the threshold, thus ensuring the reliability of the data used for subsequent initialization.

[0067] The following example experiment further illustrates the implementation process of the present invention and its effect compared to the aforementioned prior art. The comparative example uses a scheme that controls mode switching solely based on indoor temperature, humidity, and temperature-humidity index, without performing non-contact respiratory rate measurement, target and reference subgroup construction, trial control sequences, respiratory response matrix construction, and verification updates. To ensure comparability, both the example group and the comparative example group selected two lactating dairy cow environmental control zones with essentially identical structures, orientations, feeding scales, and control equipment configurations. These zones were run continuously for 10 days, with the statistical period selected being the period from 14:00 to 17:00, when the external high temperature and humidity load was relatively heavy. Furthermore, the average weight of dairy cows, lactation stage intervals, milk production period intervals, and herd density in both groups were kept similar; the sensor models, installation heights, sampling cycles, and rated capacities of the control devices in both groups were kept consistent; in the comparative example, when the temperature-humidity index reached the preset first-level threshold, the variable frequency fan was activated; when it reached the second-level threshold, a wet curtain device was added; and when it reached the third-level threshold, a spray device was added, with a fixed hysteresis threshold used for mode rollback. The experimental conditions for both groups are shown in Table 1.

[0068] Test conditions table 1

[0069] In the example group, the controller first obtains the respiratory rate of each individual dairy cow according to the above method, constructs a target subgroup and a reference subgroup, and then conducts trial adjustments to the variable frequency fan, local directional air supply device, wet curtain device, and roof sprinkler device in sequence. Taking the first environmental control zone on the 4th day of the experiment as an example, the response parameters of the candidate control devices obtained for the target subgroup of this zone are shown in Table 2.

[0070] Table 2 of Exemplary Respiratory Response Parameters for Candidate Control Devices in the First Environmental Control Zone

[0071] As shown in Table 2, in this exemplary scenario, the localized directional air supply device exhibits a higher efficiency in reducing the respiration rate per unit resource under the first trial intensity. While the variable frequency fan brings a greater reduction in respiration rate under the second trial intensity, its efficiency in reducing the respiration rate per unit resource is slightly lower than that of the localized directional air supply device. Therefore, the controller prioritizes the localized directional air supply device as the first formal control action, and then superimposes the variable frequency fan as the second formal control action. The evaporative cooling pad device is not prioritized in this formal control combination due to its longer activation delay and constraints when the current humidity margin is insufficient.

[0072] The main results for the example group and the comparative example group within the same statistical period are shown in Table 3.

[0073] Table 3: Exemplary Comparison Results Between the Example Group and the Comparative Example Group

[0074] As shown in Table 3, the example group outperformed the comparative group in terms of average onset time of the target subgroup, time required for respiratory rate to recover to the target range, and the magnitude of the decrease in average respiratory rate of the target subgroup within 20 minutes. Furthermore, the example group had lower single-cycle power consumption, single-cycle water consumption, and single-cycle relative humidity increase than the comparative group. This indicates that by employing non-contact respiratory rate measurement, trial control sequences, respiratory response matrices, and verification and update mechanisms, it is possible to more effectively select truly effective control devices and their execution order within the current environmental control zone. This improves respiratory recovery speed while reducing unnecessary energy and water consumption and suppressing the side effects of excessive humidification.

[0075] To further illustrate the role of the verification update mechanism, on day 7 of the aforementioned experiment, due to changes in external wind direction, the local wind speed distribution near the intake side of the first environmental control zone changed. After executing formal control, the controller found that both the predicted onset delay and the predicted recovery half-life exceeded the error threshold. Therefore, it re-executed the trial control sequence to update the respiratory response matrix. Some response parameters before and after the update are shown in Table 4.

[0076] Table 4 shows exemplary parameter changes before and after the response matrix update.

[0077] As shown in Table 4, when the external boundary conditions changed, the response relationship that was previously more favorable to the local directional ventilation device changed, while the relative efficiency of the variable frequency fan increased. After the controller corrected the respiratory response matrix through the verification update mechanism, the subsequent formal control combination was adjusted from "local directional ventilation device priority" to "variable frequency fan priority plus local directional ventilation device", thus maintaining a better respiratory recovery effect. This shows that the present invention does not simply rely on fixed thresholds or fixed equipment priorities for environmental control, but can dynamically adjust the control strategy according to the current environmental control zone, the current herd status, and the current environmental boundary conditions.

[0078] The following describes the implementation method for pigsty adaptation. In the pigsty application scenario, the method of dividing the environmental control zones, the execution logic of the trial control sequence, the determination logic of the target subgroup and the reference subgroup, and the construction logic of the respiratory response matrix are the same as those in the aforementioned dairy cow shed implementation method. The difference is that the measurement area for the pig's nose is preferably determined in the area around the tip of the nose and the nostrils, and the measurement area for the chest and abdomen is preferably determined in the area with obvious undulations on the lateral edge of the thorax and the ventral side. Since the differences in body surface texture among fattening pigs are relatively small, continuous tracking markers can rely more on the joint determination of individual outline, body size, key ear points, and positional continuity. For fattening pigs, the respiratory rate range is usually higher than that of adult dairy cows, so the bandpass filter range and the preset consistency threshold can be adjusted accordingly. Preferably, the consistency threshold of the first respiratory rate and the second respiratory rate of fattening pigs can be set to 5 breaths / min, the trial control duration can be set to 2 min to 4 min, and the recovery interval can be set to 4 min to 8 min; the air outlet velocity of the local directional air supply device is preferably lower than that of the dairy cow shed to reduce stress disturbance when pigs gather. Through the above adaptation, individual respiratory rate can still be measured by utilizing the periodic changes in nasal thermal plume and the periodic displacement of the chest and abdomen. Furthermore, the actual effects of variable frequency fans, spray devices, and air intake components on the recovery of respiratory status in pig herds can be identified by exploring and controlling the sequence.

[0079] In a more specific pigsty embodiment, a single environmental control zone within a fattening pigsty is used as the object, containing 24 fattening pigs. The visible light acquisition unit is installed at a height of 3.5m, and the thermal imaging acquisition unit is installed at a height of 3.3m. The controller first identifies the nasal measurement area based on the pig's snout contour and the front edge of its head, and then identifies the chest and abdomen measurement area based on the lateral edge of the pig's chest and the ventral contour. When the respiratory rate of an individual fattening pig deviates by more than 12% from its historical stable respiratory rate, that individual is included in the candidate target subgroup. The controller selects fattening pigs with the same weight range and age range in adjacent environmental control zones, and which have not received the action of the adjustment device in the current trial control sub-phase, to form a reference subgroup. Subsequently, a first trial intensity of 38Hz and a second trial intensity of 42Hz are applied to the variable frequency fan, a first trial intensity of 18% duty cycle and a second trial intensity of 28% duty cycle are applied to the spray device, and a first trial intensity of 25% opening increment and a second trial intensity of 40% opening increment are applied to the air intake component. The trial control results show that, under hot and humid summer conditions, the first trial intensity of the variable frequency fan can reduce the average respiration rate of the target subgroup by 4.3 breaths / min, with an onset time of 2.8 minutes. In contrast, the spray device, under the same conditions, has an onset time of 5.6 minutes and a significant increase in humidity. Therefore, the controller prioritizes the variable frequency fan as the initial control action, and then considers the remaining respiration rate deviation and adds it to the air intake components. This demonstrates that the present invention can also achieve environmental control in pigsty scenarios through a process of identification followed by decision-making.

[0080] The following describes an implementation method adapted for poultry houses. In layer or broiler houses, due to the larger number of poultry individuals, their smaller size, and higher group density, a combination of group measurement and local individual tracking can be used. Specifically, poultry individuals within the same perch area or the same ground grid area can be grouped into local measurement groups. Within these groups, thermal plume detection is performed on several individuals with relatively stable postures and sufficient head and neck exposure, focusing on the beak tip and nostrils. Simultaneously, the second respiratory rate is extracted by analyzing the micro-motion displacement of the chest feather boundary. For poultry, the nasal measurement area can be adapted to the area near the beak tip and nostrils, while the chest and abdomen measurement area can be adapted to the area with significant chest feather undulation. Because poultry have a high respiratory rate, the upper limit of the bandpass filter should be increased accordingly, and the duration of the trial-and-control sub-stage should be shortened to capture faster physiological responses. Preferably, the trial-and-control duration in the poultry house scenario can be set to 1-3 minutes, the recovery interval to 2-5 minutes, and the consistency threshold to approximately 6 times / min. In poultry houses, the localized directional air supply device can correspond to an adjustable air guide nozzle or a small-area air belt control device. Through the above adaptation, the present invention can still complete the construction of the target subgroup, reference subgroup, differential respiratory changes, respiratory response parameters, and respiratory response matrix, thereby realizing environmental regulation in response to the real-time respiratory status of poultry flocks.

[0081] In a more specific poultry house embodiment, a single environmental control zone of a broiler house is taken as the object. The zone is divided into 6 local measurement grids according to the ground grid. In each local measurement grid, 6 broilers with sufficient head and neck exposure and continuous stability for more than 20 seconds are selected as local measurement individuals. The controller extracts the first respiratory rate based on the change in thermal plume flow near the beak of the local measurement individuals, and extracts the second respiratory rate based on the slight movement of the chest feather boundary, and judges the consistency between the two. When the local average respiratory rate measured in the local measurement grid deviates from the historical stable period by more than 15%, the broiler group corresponding to the local measurement grid is included in the candidate target subgroup. Subsequently, a first test intensity of 1.4 m / s and a second test intensity of 1.8 m / s are applied to the local wind belt control device, and a first test intensity of 32 Hz and a second test intensity of 36 Hz are applied to the ventilation fan. The results of the trial showed that in a high-density broiler environment, the local wind belt control device had a shorter onset time for broiler groups in high-exposure areas, while the ventilation fan was more beneficial to the reduction of the overall average respiration rate. Therefore, the controller prioritized the local wind belt control device as the first formal control action, and then superimposed the ventilation fan as the second formal control action.

[0082] To further illustrate the principles for setting parameter ranges for different livestock and poultry species, Table 5 is provided as an example of preferred parameter ranges.

[0083] Table 5 Examples of Optimal Parameter Ranges for Different Livestock and Poultry Scenarios

[0084] To provide more intuitive examples of the implementation of pig and poultry house parameters, Tables 6 and 7 are provided.

[0085] Table 6 Example parameters of pigsty adaptation implementation methods

[0086] Table 7 Example parameters of poultry house adaptation implementation methods

[0087] The parameter ranges and example values ​​in Tables 5 to 7 are merely preferred examples. Those skilled in the art can adaptively adjust the parameter ranges based on the livestock species, age, growth stage, environmental control device capacity, and site structure. The aforementioned parameter ranges, together with the specific implementation process described above, demonstrate that this invention is not limited to dairy cow shed scenarios, but can be implemented using the same technical concept in different livestock shed scenarios.

[0088] In a further preferred embodiment, after initializing the respiratory response matrix, the controller also performs an environmental adaptability analysis step based on an artificial intelligence model. Specifically, the controller inputs a feature set consisting of the average respiratory rate of the target subgroup, the average respiratory rate of the reference subgroup, the deviation of the target subgroup's respiratory rate, air temperature, relative humidity, air velocity, carbon dioxide concentration, ammonia concentration, airflow exposure coefficient, operating status of candidate control devices, historically validated control records, and respiratory response parameters obtained from the current trial control sequence. This feature set is then input into the environmental adaptability analysis model. The environmental adaptability analysis model outputs the environmental adaptability score, respiratory stress level, priority of candidate control devices, target control intensity, suggested formal control actions, and suggested execution order for the current environmental control zone. The controller integrates the output of the environmental adaptability analysis model with the response parameters represented by the respiratory response matrix. When the priority of candidate control devices output by the environmental adaptability analysis model matches the priority reflected in the respiratory response matrix, the corresponding result is directly used to determine the formal control combination. When the priority of candidate control devices output by the environmental adaptability analysis model does not match the priority reflected in the respiratory response matrix, a comprehensive comparison is made based on the efficiency of respiratory rate decline per unit resource, onset delay, recovery half-life, and historical verification error. The result with the smaller historical verification error is prioritized to determine the formal control combination. Through this method, the artificial intelligence model does not make black-box decisions independently of the respiratory response matrix. Instead, it further analyzes the environmental adaptability of the current environmental control zone based on the existing trial control identification results, thereby improving the adaptability of the formal control combination to the current zone, the current population, and the current boundary conditions.

[0089] In a preferred embodiment, the historically validated control record library serves not only as the initial respiratory response matrix but also as a sample database for the environmental adaptability analysis model. The sample database includes at least pre-control respiratory rate data, environmental parameter data, airflow exposure characteristics, candidate control device operating status data, trial control sequence parameters, formal control combinations, measured respiratory rate recovery results, and validation results. After each formal control cycle, the controller writes new samples into the sample database. When the number of new samples reaches a preset update quantity, or when the validation error continuously exceeds a preset error threshold, the controller triggers parameter updates, incremental training, or feature weight corrections for the environmental adaptability analysis model to achieve continuous optimization of the model and database. Preferably, the preset update quantity can be set to 20 to 100 sets of new samples; in this embodiment, 50 sets are used. The number of times the preset error threshold is continuously exceeded can be set to 3 to 10 times; in this embodiment, 5 times are used. Through this method, the environmental adaptability analysis model can be continuously optimized according to changes in different seasons, different livestock and poultry population states, and different environmental boundary conditions, thereby further improving the accuracy and stability of environmental control.

[0090] It should be noted that the embodiments of the present invention have better implementability and are not intended to limit the present invention in any way. Any person skilled in the art may use the above-disclosed technical content to change or modify it into equivalent effective embodiments. However, any modifications or equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall still fall within the scope of the technical solution of the present invention.

Claims

1. A method for controlling the environment of livestock houses based on non-contact livestock and poultry respiratory rate measurement, characterized in that, Includes the following steps: S1. Acquire synchronous visible light image sequences, synchronous thermal imaging image sequences, environmental parameter data, and operating status data of the control devices for each environmental control zone in the livestock house. The environmental parameter data includes at least air temperature, relative humidity, air velocity, carbon dioxide concentration, and ammonia concentration. S2. Identify individual livestock and poultry based on the synchronous visible light image sequence, and determine the spatial location, body surface orientation, nose measurement area, and chest and abdomen measurement area of ​​each individual livestock and poultry. S3. For each of the livestock and poultry individuals, the individual respiratory rate is determined based on the periodic change of the exhaled thermal plume in the nasal measurement area and the periodic displacement of the chest and abdomen measurement area, and the individual respiratory rate is assigned to the corresponding environmental control zone. S4. Based on the degree of deviation between the individual respiratory rate and the corresponding historical stable period respiratory rate, select a target subgroup in the partition to be regulated, and select a reference subgroup in the partition adjacent to the partition to be regulated that has not received the current trial regulation action. S5. When the respiratory state of the target subgroup exceeds the preset target range, a trial control sequence is executed on the partition to be controlled. The trial control sequence includes applying short-term adjustment actions to multiple candidate control devices individually, and synchronously collecting the respiratory rate changes of the target subgroup and the reference subgroup during each short-term adjustment action. S6. Based on the respiratory rate change amplitude, onset delay and recovery process of the target subgroup relative to the reference subgroup, determine the respiratory response parameters of each candidate control device in the current environmental control zone, and construct the respiratory response matrix of the current environmental control zone by combining the correspondence between individual livestock and poultry and airflow path. S7. Based on the respiratory response matrix, current environmental parameters, operating boundaries of the control device, and resource consumption constraints, determine the formal control combination and its execution order; S8. Perform environmental regulation on the partition to be regulated according to the formal regulation combination, compare the measured respiratory rate recovery result with the predicted result within the preset verification period, update the respiratory response matrix when the update condition is met, and redetermine the formal regulation combination.

2. The method for controlling the livestock housing environment based on non-contact livestock and poultry respiratory rate measurement according to claim 1, characterized in that, In step S4, the target subgroup consists of livestock and poultry individuals located within the control zone, of the same breed and growth stage, and whose respiratory rate deviation is greater than a preset deviation threshold; the reference subgroup consists of livestock and poultry individuals located within adjacent zones, of the same breed and growth stage, and who have not received the action of the candidate control device that has been applied to them in the current trial control sub-stage.

3. The method for controlling the livestock housing environment based on non-contact livestock and poultry respiratory rate measurement according to claim 1, characterized in that, It also includes the step of determining the airflow exposure coefficient of individual livestock and poultry; the airflow exposure coefficient is determined based on the spatial position of the individual livestock and poultry, the orientation of the body surface, the position of the air intake component, the position of the air exhaust component, the position of the local directional air supply device, and the interpolation result of the local airflow velocity, and is used to characterize the relative intensity of the airflow effect received by the corresponding individual livestock and poultry from each of the candidate control devices.

4. The method for controlling the livestock housing environment based on non-contact livestock and poultry respiratory rate measurement according to claim 1, characterized in that, In step S3, the individual respiratory rate is output only for valid measurement individuals, and the valid measurement individuals meet the following conditions: The exhaled thermal plume periodic signal in the nasal measurement area and the periodic displacement signal in the chest and abdomen measurement area both meet the preset periodic conditions. The difference between the first respiratory rate obtained based on the exhaled hot plume periodic signal and the second respiratory rate obtained based on the periodic displacement signal is not greater than a preset consistency threshold. Furthermore, the same individual livestock and poultry is continuously tracked during the trial control sequence and the verification period.

5. The method for controlling the livestock housing environment based on non-contact livestock and poultry respiratory rate measurement according to claim 1, characterized in that, In step S5, the trial regulation sequence is implemented in a single device, single variable, segmented execution manner. In each trial regulation sub-stage, only one regulation variable of one candidate regulation device is changed, while the other candidate regulation devices remain unchanged. Furthermore, a recovery interval is set between two adjacent trial regulation sub-stages until the respiratory rate difference between the target subgroup and the reference subgroup returns to a preset stable range before the next trial regulation sub-stage is executed.

6. The livestock housing environment control method based on non-contact livestock and poultry respiratory rate measurement according to claim 5, characterized in that, The candidate control device includes at least two of the following: variable frequency fan, air intake component, wet curtain device, spray device, roof sprinkler device, and local directional air supply device; and, a first test intensity and a second test intensity are sequentially applied to at least one of the candidate control devices, wherein the second test intensity is applied when the target subgroup produces a differential change in respiration rate relative to the reference subgroup under the first test intensity.

7. The livestock housing environment control method based on non-contact livestock and poultry respiratory rate measurement according to claim 6, characterized in that, In step S6, the respiratory response parameters include at least the respiratory rate decrease magnitude, onset delay, recovery half-life, and respiratory rate decrease efficiency per unit resource. The respiratory response parameters are determined by the differential changes between the target subgroup and the reference subgroup in the corresponding trial control sub-stages. When the differential change direction of the same candidate control device is inconsistent under the first trial intensity and the second trial intensity, or when the onset delay exceeds the preset maximum allowable duration, the candidate control device is marked as a currently invalid device and removed from the determination process of the formal control combination.

8. The livestock housing environment control method based on non-contact livestock and poultry respiratory rate measurement according to claim 7, characterized in that, When synchronous respiratory rate changes occur in adjacent partitions during the execution of the trial regulation sequence, a partition coupling correction step is also included. The partition coupling correction step determines the partition coupling coefficient based on the shared air intake component status, the shared air exhaust component status, and the shared local directional air supply device status, and then deducts the coupling component in the differential change before constructing the breathing response matrix.

9. The method for controlling the livestock housing environment based on non-contact livestock and poultry respiratory rate measurement according to claim 8, characterized in that, Before executing step S5, the initialization step of the respiratory response matrix is ​​also included. The initialization step selects historical records from the historically verified control records that are consistent with the current external enthalpy range, livestock and poultry species, growth stage and stocking density range to form an initial respiratory response matrix, and uses the respiratory response parameters obtained from the current trial control sequence to correct the initial respiratory response matrix.

10. The livestock housing environment control method based on non-contact livestock and poultry respiratory rate measurement according to claim 4, characterized in that, In step S8, during the verification period, based on the continuously tracked livestock individuals constituting the target subgroup, the predicted onset time delay, the predicted decrease in respiratory rate, and the predicted recovery half-life are compared with the corresponding measured values; when at least two of them exceed the corresponding error threshold, the trial regulation sequence is re-executed.

Images