Incubator environment adjusting method and system

Abstract

The invention relates to the technical field of incubator environment adjustment, in particular to an incubator environment adjustment method and system. The method comprises the following steps: arranging an electronic monitoring device, a temperature and humidity sensor and an infrared temperature sensor in an incubator, carrying out omnibearing form image acquisition and eggshell temperature monitoring on an incubated egg, and obtaining surface temperature time sequence fluctuation data. Real-time temperature and humidity data in the incubator are collected through a temperature and humidity sensor, and the abnormal degree of embryonic development is predicted by combining temperature centralized recognition of a heat conduction area; based on the prediction data, performing stage demand interval matching on the eggshell surface temperature, and optimizing a hatching temperature interval; and then, balancing the temperature and humidity environment in the incubator through interactive adjustment of the space temperature and humidity data, then constructing a heating / space temperature and humidity control model, sending the heating / space temperature and humidity control model to the terminal, and executing incubator environment adjustment. The incubator environment adjusting technology is optimized, so that the incubator environment adjusting technology is more perfect.

Description

Technical Field

[0001] The present invention relates to the technical field of incubator environment regulation, and particularly to an incubator environment regulation method and system. Background Art

[0002] During the hatching process, environmental factors such as temperature, humidity, and ventilation are crucial for the development of embryos. The environment of the incubator must be maintained within an optimal range to ensure the healthy development of the embryos in the eggs and a high hatching success rate. Conventional incubators usually rely on simple temperature and humidity control systems, but these systems have problems such as insufficiently timely feedback responses to environmental data and inaccurate temperature and humidity regulation, which affect the hatching effect. The application of intelligent control technology and sensors has become an important direction for the optimization management of incubators. By using a variety of sensors to monitor the internal environment of the incubator in real time and combining big data analysis and machine learning algorithms, factors such as temperature and humidity fluctuations, heat conduction areas, and embryo development status can be more accurately identified, thereby realizing fine regulation of the incubator environment. This data-driven regulation method can improve the hatching rate. However, a traditional incubator environment regulation method has a low accuracy in predicting abnormal embryo development, resulting in a large error in the regulation of the incubator environment (wherein, the regulation of the incubator environment includes heating the hatching eggs and regulating the temperature and humidity in the incubator space). Summary of the Invention

[0003] Based on this, it is necessary to provide an incubator environment regulation method and system to solve at least one of the above technical problems.

[0004] To achieve the above object, an incubator environment regulation method, the method includes the following steps: Step S1: Deploy electronic monitoring devices, temperature and humidity sensors, and infrared temperature sensors in multiple directions inside the incubator. Use the electronic monitoring devices to collect omnidirectional morphological images of the hatching eggs to obtain omnidirectional morphological images of the hatching eggs; according to the omnidirectional morphological images of the hatching eggs, and use the infrared temperature sensors to monitor the heating temperature state on the outer surface of the hatching eggs in multiple directions to obtain surface temperature time-series fluctuation characteristic data; Step S2: Use the temperature and humidity sensors to collect real-time data on the temperature and humidity in the internal space of the incubator to obtain real-time data on the temperature and humidity in the space; according to the real-time data on the temperature and humidity in the space, identify the temperature concentration in the heat conduction area for the surface temperature time-series fluctuation characteristic data to obtain temperature concentration data in the heat conduction area; according to the temperature concentration data in the heat conduction area, simulate and predict the degree of abnormal development of the embryos in the eggs to obtain prediction data on the degree of abnormal embryo development; Step S3: Based on the prediction data on the degree of abnormal embryo development, match the stage temperature demand intervals on the outer surface of the eggshell in multiple directions for the surface temperature time-series fluctuation characteristic data to obtain an optimized stage temperature matching interval; Step S4: Perform spatial temperature and humidity interactive phased balance adjustment on the real-time spatial temperature and humidity data according to the optimized interval matching the phased temperature, and obtain the spatially and temporally phased balanced adjustment data of temperature and humidity; construct an incubator heating / spatial temperature and humidity control model based on the random forest algorithm for the spatially and temporally balanced adjustment learning data of temperature and humidity, and obtain the heating / spatial temperature and humidity control model; send the heating / spatial temperature and humidity control model to the terminal to execute the incubator environment adjustment method.

[0005] Preferably, step S1 includes the following steps: Step S11: Deploy electronic monitoring devices, temperature and humidity sensors, and infrared temperature sensors in multiple directions inside the incubator, and collect the omnidirectional morphological images of the hatching eggs through the electronic monitoring devices to obtain the omnidirectional morphological images of the hatching eggs; Step S12: According to the omnidirectional morphological images of the hatching eggs, and monitor the heating temperature state on the outer surface of the eggshell in multiple directions through the infrared temperature sensor to obtain the monitoring data of the temperature state on the outer surface of the eggshell; Step S13: Perform time-series fluctuation analysis on the monitoring data of the temperature state on the outer surface of the eggshell to obtain the time-series fluctuation data of the surface temperature; Step S14: Perform feature analysis on the time-series fluctuation data of the surface temperature to obtain the time-series fluctuation feature data of the surface temperature.

[0006] Preferably, step S2 includes the following steps: Step S21: Collect the real-time spatial temperature and humidity data inside the incubator through the temperature and humidity sensors to obtain the real-time spatial temperature and humidity data; Step S22: Perform heat conduction region temperature concentration identification on the time-series fluctuation feature data of the surface temperature according to the real-time spatial temperature and humidity data and the omnidirectional morphological images of the hatching eggs to obtain the heat conduction region temperature concentration data; Step S23: Perform distribution enhancement gradient numerical quantization on the heat conduction region temperature concentration data to obtain the temperature distribution enhancement numerical gradient data; Step S24: Perform simulation prediction on the abnormal degree of embryo development inside the egg according to the temperature distribution enhancement numerical gradient data and the omnidirectional morphological images of the hatching eggs to obtain the prediction data of the abnormal degree of embryo development.

[0007] Preferably, step S22 includes the following steps: Step S221: Perform hatching egg curvature structure calculation on the omnidirectional morphological images of the hatching eggs to obtain the hatching egg curvature structure data; perform end meandering angle calculation on the omnidirectional morphological images of the hatching eggs to obtain the end meandering angle of the hatching eggs; Step S222: Perform texture rough surface distribution difference identification on the omnidirectional morphological images of the hatching eggs based on the hatching egg curvature structure data and the end meandering angle of the hatching eggs to obtain the texture rough surface distribution difference data; Step S223: Perform axial ratio analysis based on the curvature structure data of the hatching egg and the meandering angle at the end of the hatching egg to obtain the axial ratio data of the hatching egg; Step S224: Extract the humidity increment interval from the real-time data of space temperature and humidity to obtain the space humidity increment interval; perform analysis on the heat capacity rising interval of the real-time data of space temperature and humidity based on the space humidity increment interval to obtain the heat capacity rising interval; Step S225: Evaluate the effective conversion increment of medium heat conduction according to the heat capacity rising interval to obtain the medium heat conduction increment data; Step S226: Perform clustering on the thermal conduction region diffusion fluctuation differences of the surface temperature time series fluctuation characteristic data according to the texture rough surface distribution difference data, the axial ratio data of the hatching egg, and the medium heat conduction increment data to obtain the thermal conduction diffusion fluctuation difference clustering data; Step S227: Identify the temperature concentration in the thermal conduction region for the thermal conduction diffusion fluctuation difference clustering data to obtain the temperature concentration data in the thermal conduction region.

[0008] Preferably, step S24 includes the following steps: Step S241: Perform analysis on the entropy change of protein denaturation in embryonic tissues based on the temperature distribution enhanced numerical gradient data to obtain the entropy change data of embryonic protein denaturation; Step S242: Simulate the cascade amplification effect of embryonic heat damage on the entropy change data of embryonic protein denaturation to obtain the cascade amplification effect of embryonic heat damage; Step S243: Estimate the exponential growth of the oxygen consumption demand of the embryo according to the temperature distribution enhanced numerical gradient data to obtain the exponentially growing data of the oxygen consumption demand; Step S244: Identify the density of eggshell pores for the omnidirectional morphological image of the hatching egg to obtain the eggshell pore density data; Step S245: Simulate and predict the degree of embryonic hypoxic metabolic disorder for the exponentially growing data of the oxygen consumption demand according to the eggshell pore density data to obtain the predicted data of the degree of hypoxic metabolic disorder; Step S246: Simulate and predict the degree of abnormal embryonic development based on the cascade amplification effect of embryonic heat damage and the predicted data of the degree of hypoxic metabolic disorder to obtain the predicted data of the degree of abnormal embryonic development.

[0009] Preferably, step S242 includes the following steps: Perform reconstruction processing on the entropy change data of embryonic protein denaturation to obtain the entropy flow space distribution tensor; Analyze the equivalent change of enzyme inactivation for the entropy change data of embryonic protein denaturation according to the entropy flow space distribution tensor to obtain the equivalent change data of enzyme inactivation; Derive the multi-phase weakening rate of enzyme inactivation from the data of equivalent changes in enzyme inactivation to obtain the multi-phase weakening rate of enzyme inactivation; Perform numerical integration of the cell division arrest time series based on the multi-phase weakening rate of enzyme inactivation to obtain the numerical value of the cell division arrest time series; Simulate the embryo heat damage cascade amplification effect on the embryo protein denaturation entropy change data according to the multi-phase weakening rate of enzyme inactivation and the numerical value of the cell division arrest time series to obtain the embryo heat damage cascade amplification effect.

[0010] Preferably, step S245 includes the following steps: Perform calculation and estimation of the eggshell gas exchange capacity per unit time based on the eggshell pore density data to obtain the estimated eggshell gas exchange capacity per unit time; Quantitatively evaluate the lack of embryo oxygen content for the exponentially growing data of oxygen consumption demand based on the estimated eggshell gas exchange capacity to obtain the quantitative data of the lack of embryo oxygen content; Evaluate the compensatory enhancement of glycolysis under the embryo stress state based on the quantitative data of the lack of embryo oxygen content to obtain the data of compensatory enhancement of glycolysis; Analyze the ratio of the lactic acid accumulation increment rate for the data of compensatory enhancement of glycolysis to generate the ratio of the lactic acid accumulation increment rate; Estimate the probability of embryo myocardial contractility loss according to the ratio of the lactic acid accumulation increment rate to obtain the probability of embryo myocardial contractility loss; Simulate and predict the degree of embryo hypoxic metabolic disorder based on the ratio of the lactic acid accumulation increment rate and the probability of embryo myocardial contractility loss to obtain the predicted data of the degree of hypoxic metabolic disorder.

[0011] Preferably, step S3 includes the following steps: Step S31: Normalize the predicted data of embryo developmental abnormality degree to obtain the normalized data of developmental abnormality degree; Step S32: Match the phased temperature demand intervals on the outer surface of the eggshell in multiple directions for the data of the phased fluctuation characteristics of the surface temperature based on the normalized data of the developmental abnormality degree to obtain the phased temperature demand intervals; Step S33: Iteratively simulate and optimize the phased temperature demand intervals to obtain the optimized phased temperature matching intervals.

[0012] Preferably, step S4 includes the following steps: Step S41: Perform spatial temperature and humidity interactive phased balance adjustment on the real-time data of spatial temperature and humidity according to the optimized phased temperature matching intervals to obtain the spatial temperature and humidity phased balance adjustment data; Step S42: Perform logical learning on the spatial temperature and humidity phased balance adjustment data to obtain the spatial temperature and humidity balance adjustment learning data. Step S43: Based on the random forest algorithm, construct an incubator heating / space temperature and humidity control model for the phased temperature matching optimization interval and the space temperature and humidity balance adjustment learning data, and obtain the heating / space temperature and humidity control model; send the heating / space temperature and humidity control model to the terminal to execute the incubator environment adjustment method.

[0013] Preferably, the present invention also provides an incubator environment adjustment system for executing the above-mentioned incubator environment adjustment method. The incubator environment adjustment system includes: Temperature time-series fluctuation monitoring module, which is used to deploy electronic monitoring devices, temperature and humidity sensors, and infrared temperature sensors in multiple directions inside the incubator, collect omnidirectional morphological images of the hatching eggs through the electronic monitoring devices to obtain omnidirectional morphological images of the hatching eggs; according to the omnidirectional morphological images of the hatching eggs, and monitor the heating temperature state on the outer surface of the eggshell in multiple directions through the infrared temperature sensor to obtain surface temperature time-series fluctuation characteristic data; Embryonic development abnormality degree prediction module, which is used to collect real-time data of the space temperature and humidity inside the incubator through the temperature and humidity sensors to obtain real-time space temperature and humidity data; identify the temperature concentration in the heat conduction area of the surface temperature time-series fluctuation characteristic data according to the real-time space temperature and humidity data to obtain heat conduction area temperature concentration data; simulate and predict the degree of embryonic development abnormality in the egg according to the heat conduction area temperature concentration data to obtain embryonic development abnormality degree prediction data; Stage temperature demand matching module for the outer surface of the eggshell, which is used to perform multi-directional matching of the stage temperature demand interval on the outer surface of the eggshell for the surface temperature time-series fluctuation characteristic data based on the embryonic development abnormality degree prediction data to obtain a phased temperature matching optimization interval; Heating / space temperature and humidity control model construction module, which is used to perform space temperature and humidity interactive phased balance adjustment on the real-time space temperature and humidity data according to the phased temperature matching optimization interval to obtain space temperature and humidity phased balance adjustment data; construct an incubator heating / space temperature and humidity control model based on the random forest algorithm for the space temperature and humidity balance adjustment learning data to obtain the heating / space temperature and humidity control model; send the heating / space temperature and humidity control model to the terminal to execute the incubator environment adjustment method.

[0014] The beneficial effects of the present invention are as follows. By deploying electronic monitoring devices, temperature and humidity sensors, and infrared temperature sensors in multiple directions inside the incubator, the morphological images of the hatching eggs and the data on the surface temperature changes can be comprehensively collected. The omnidirectional morphological images of the hatching eggs obtained through the electronic monitoring devices can accurately understand the external conditions of each hatching egg and promptly detect any morphological abnormalities. In addition, the data on the characteristics of the surface temperature fluctuations of the eggshell provided by the infrared temperature sensors helps to monitor the heating state of the hatching eggs in real time, ensuring that the temperature is maintained within an appropriate range during the hatching process, thereby improving the hatching success rate. By collecting the spatial temperature and humidity data inside the incubator in real time through the temperature and humidity sensors, the changes in the environment can be accurately recorded, providing reliable data support for subsequent adjustments. By combining the spatial temperature and humidity data to identify the temperature concentration in the heat conduction region based on the temporal fluctuation characteristics of the surface temperature, the regions with large temperature fluctuations can be efficiently located, avoiding problems with embryo development caused by local temperature differences during the hatching process. By simulating and predicting the degree of embryo development abnormalities, potential hatching problems can be detected early, which helps to take adjustment measures in the early stage. Based on the prediction data of the degree of embryo development abnormalities, the surface temperature requirements of the hatching eggs can be optimized stage by stage, effectively adjusting the temperature control strategy to make the hatching process more precise. The generation of this stage-by-stage temperature matching optimization interval can dynamically adjust the temperature requirements of the hatching eggs according to different stages of embryo development, ensuring that the temperature conditions in each stage are the most suitable, thereby reducing the risks brought by uneven or fluctuating temperatures and contributing to improving the hatching quality and the healthy development of the embryo. Thus, the hatching success rate is increased. By performing interactive stage-by-stage balance adjustment on the real-time spatial temperature and humidity data according to the stage-by-stage temperature matching optimization interval, the temperature and humidity changes inside the incubator can be accurately controlled, keeping them always within an ideal range. This balance adjustment can eliminate the negative impacts of temperature and humidity fluctuations on the hatching process and ensure that the embryo obtains stable environmental conditions during the hatching process. By using the random forest algorithm to learn the spatial temperature and humidity balance adjustment data, the incubator can intelligently construct a heating and spatial temperature and humidity control model. This control model can not only automatically adjust the heating system of the incubator to maintain an appropriate eggshell temperature but also dynamically adjust the spatial humidity of the incubator to ensure that the hatching requirements in different stages are met. This precise adjustment improves the efficiency and stability of the hatching process, reduces the need for manual intervention, and ensures that each hatching stage is carried out under optimal conditions, thereby significantly increasing the hatching rate and the quality of embryo development. Therefore, the present invention makes an improved treatment for a traditional method of incubator environment adjustment, solves the problem that the traditional method of incubator environment adjustment has a low accuracy in predicting embryo development abnormalities, resulting in a large error in incubator environment adjustment, improves the accuracy of predicting embryo development abnormalities, and reduces the error in incubator environment adjustment. Among them, the incubator environment adjustment includes the heating of the hatching eggs and the temperature and humidity adjustment of the incubator space. Description of the Drawings

[0015] Figure 1 It is a schematic diagram of the step process of an incubator environment adjustment method; Figure 2 It is Figure 1 a schematic diagram of the detailed implementation steps of step S2 in Figure 3 It is Figure 1 a schematic diagram of the detailed implementation steps of step S3 in Specific implementation manner

[0016] Please refer to Figures 1 to 3 , an incubator environment adjustment method, the method includes the following steps: Step S1: Deploy electronic monitoring devices, temperature and humidity sensors, and infrared temperature sensors in multiple directions inside the incubator. Collect omnidirectional morphological images of the hatching eggs through the electronic monitoring devices to obtain omnidirectional morphological images of the hatching eggs; According to the omnidirectional morphological images of the hatching eggs, and monitor the heating temperature state on the outer surface of the eggshell in multiple directions through the infrared temperature sensor to obtain surface temperature time series fluctuation characteristic data; Step S2: Collect real-time temperature and humidity data of the internal space of the incubator through the temperature and humidity sensors to obtain real-time temperature and humidity data of the space; Identify the temperature concentration in the heat conduction area for the surface temperature time series fluctuation characteristic data according to the real-time temperature and humidity data of the space to obtain heat conduction area temperature concentration data; Simulate and predict the abnormal degree of embryo development in the egg according to the heat conduction area temperature concentration data to obtain embryo development abnormal degree prediction data; Step S3: Match the phased temperature demand intervals on the outer surface of the eggshell in multiple directions for the surface temperature time series fluctuation characteristic data based on the embryo development abnormal degree prediction data to obtain a phased temperature matching optimization interval; Step S4: Perform phased balance adjustment of the space temperature and humidity for the real-time temperature and humidity data of the space according to the phased temperature matching optimization interval to obtain phased balance adjustment data of the space temperature and humidity; Build an incubator heating / space temperature and humidity control model for the space temperature and humidity balance adjustment learning data based on the random forest algorithm to obtain a heating / space temperature and humidity control model; Send the heating / space temperature and humidity control model to the terminal to execute the incubator environment adjustment method.

[0017] In the embodiment of the present invention, referring to Figure 1 as described, it is a schematic diagram of the step process of an incubator environment adjustment method of the present invention. In this example, the incubator environment adjustment method includes the following steps: Step S1: Deploy electronic monitoring devices, temperature and humidity sensors, and infrared temperature sensors in multiple directions inside the incubator. Use the electronic monitoring devices to collect omnidirectional morphological images of the hatching eggs to obtain omnidirectional morphological images of the hatching eggs. According to the omnidirectional morphological images of the hatching eggs, and use the infrared temperature sensors to monitor the heating temperature state on the outer surface of the eggshell in multiple directions to obtain surface temperature time-series fluctuation characteristic data; In the embodiment of the present invention, 8 miniature high-definition electronic monitoring devices are evenly installed on the surrounding walls inside the incubator, located in the four directions of east, south, west, and north, as well as the four corner positions of northeast, southeast, southwest, and northwest. The resolution of each electronic monitoring device is 4096×2160 pixels, the frame rate is 30 frames per second, and the field of view angle is 120°; at the same time, 16 miniature temperature and humidity sensors are installed inside the incubator. The temperature measurement range is -40°C to 80°C, the accuracy is ±0.5°C, the humidity measurement range is 0%RH to 100%RH, and the accuracy is ±2%RH. The sensors are evenly distributed in the 4 upper corners, 4 lower corners, and 8 evenly distributed positions in the middle of the incubator; 8 infrared temperature sensors are installed in a circular distribution around each hatching egg. The temperature measurement range is -70°C to 380°C, and the accuracy is ±0.02°C. The 8 sensors are located above, below, left, right, upper left, upper right, lower left, and lower right of the hatching egg respectively, and the distance from the surface of the hatching egg is 2 cm. The electronic monitoring device collects omnidirectional morphological images of the hatching eggs every 5 seconds. The collected images are processed by the SIFT feature extraction algorithm to extract 2048 feature points such as texture, color, and morphology on the surface of the hatching egg, and a feature vector of the hatching egg is constructed; the infrared temperature sensor samples the surface temperature of the hatching egg every 2 seconds. The sampled data obtains the temperature fluctuation law through time series analysis, and the temperature fluctuation signal is decomposed at multiple scales by wavelet transform to extract the characteristic fluctuation with a fluctuation frequency in the range of 0.05Hz to 0.5Hz. The fluctuation amplitude threshold is set to 0.5°C, and the duration threshold is set to 10 seconds. When the temperature fluctuation amplitude exceeds the threshold and the duration exceeds the threshold, it is marked as a significant temperature fluctuation point, and the time, temperature, fluctuation amplitude, etc. of the significant temperature fluctuation point are recorded to construct a surface temperature time-series fluctuation characteristic data set.

[0018] Step S2: Use the temperature and humidity sensors to collect real-time temperature and humidity data of the internal space of the incubator to obtain real-time temperature and humidity data of the space; identify the temperature concentration in the heat conduction area for the surface temperature time-series fluctuation characteristic data according to the real-time temperature and humidity data of the space to obtain heat conduction area temperature concentration data; simulate and predict the abnormal degree of embryo development in the egg according to the heat conduction area temperature concentration data to obtain embryo development abnormal degree prediction data; In the embodiment of the present invention, four groups of high-precision temperature and humidity sensors are evenly deployed in the internal space of the incubator. Each group of sensors is distributed at the four corners of the upper, lower, left, and right of the incubator, and the unified sampling frequency is set to sample once every 10 seconds. The collected data includes the current space temperature value (unit: degree Celsius) and the relative humidity value (unit: %RH). The real-time data is uniformly transmitted to the main control module through the CAN bus. The main control module uses the linear interpolation method to perform spatial interpolation on the temperature and humidity data collected in different regions to form a three-dimensional temperature and humidity field distribution model inside the incubator at each moment. Taking the above temperature and humidity distribution model as the input, the heat conduction region identification process is performed on the surface temperature time series fluctuation characteristic data of the hatching eggs. The specific method is as follows: First, a heat flux density estimation model based on Fourier derivative filtering is constructed according to the three-dimensional surface structure of the hatching eggs. At each surface point, according to the change in the included angle between the temperature fluctuation trend direction of its adjacent grid and the spatial temperature gradient direction, the heat conduction direction and aggregation degree are judged, the temperature high-density aggregation region is identified, the region label and boundary space range where heat conduction is concentrated are extracted to form the temperature concentration data in the heat conduction region. For the identified heat conduction concentrated region, the average trajectory of the temperature change with time and the first derivative trend change in this region are further extracted, and a standard deviation difference matrix is formed by comparing with the normal sample data. The extreme value enhancement algorithm and PCA dimensionality reduction technology are used to identify the main fluctuation dimensions. Combining with the normal development temperature fluctuation pattern in the hatching cycle, the region with a significant deviation from the trajectory is defined as the potential development abnormal region, and the corresponding prediction data of the embryo development abnormal degree is output, and the heat flux disturbance intensity in the prediction region is used as the abnormal level quantification index.

[0019] Step S3: Based on the prediction data of the embryo development abnormal degree, perform phased temperature demand interval matching on the surface temperature time series fluctuation characteristic data on the outer surface of the eggshell in multiple directions to obtain the phased temperature matching optimization interval; In the embodiment of the present invention, after obtaining the prediction data of the embryo development abnormal degree, it is used as a matching parameter to perform phased temperature demand matching operation with the aforementioned surface temperature time series fluctuation characteristic data. First, perform linear normalization processing on the prediction data of the development abnormal degree, and the normalization range is set to 0 to 1. The result is used as a weight factor to multiply the temperature gradient change value in the temperature time series fluctuation data to construct a weighted temperature demand signal sequence. According to the hatching cycle, time stages are divided (such as 3 days, 10 days, 15 days, 21 days). The mean sliding analysis is performed on the weighted temperature demand signal in each stage to extract its stable interval range, and this stable interval is used as the temperature demand interval for the current stage. Through the local minimum error interpolation optimization technology, the boundaries of the preliminary matching interval are iteratively corrected to make the temperature intervals in each time period maintain the minimum error disturbance in terms of continuity and gradient transition. Finally, the temperature matching optimization intervals for each stage are output as the input basis for constructing the temperature and humidity adjustment strategy.

[0020] Step S4: According to the optimized interval of the phased temperature matching, perform spatial temperature and humidity interactive phased balance adjustment on the real-time spatial temperature and humidity data to obtain the spatial temperature and humidity phased balance adjustment data; construct an incubator heating / spatial temperature and humidity control model based on the random forest algorithm for the spatial temperature and humidity balance adjustment learning data to obtain the heating / spatial temperature and humidity control model; send the heating / spatial temperature and humidity control model to the terminal to execute the incubator environment adjustment method.

[0021] In the embodiment of the present invention, based on the above-mentioned optimized interval of the phased temperature matching, a spatial temperature and humidity interactive adjustment model is constructed on the basis of the real-time temperature and humidity data. The specific implementation method is as follows: Set the upper and lower limits of the target temperature in the current stage as the target interval value, construct a minimum interference adjustment function according to the principle of minimum information entropy, and adjust the existing spatial temperature distribution model. The adjustment method adopts the radial Gaussian weighted adjustment method with the local maximum as the core point, and adjusts the output power of the heating elements or humidifying units where the four sensors are located under the condition of minimum perturbation to realize the fine adjustment and correction of temperature and humidity. Record each adjustment behavior and the corresponding spatial distribution change, and construct a spatial temperature and humidity interactive adjustment record database. Accumulate the database data stage by stage to form a mapping relationship matrix between the adjustment behavior and the effect, and use the random forest classification algorithm for model training. The input of the model is set as the optimized interval of the phased temperature and the current temperature and humidity field data, and the output is set as the next adjustment parameter configuration vector. The number of trees in the random forest is set to 100, the maximum depth of each tree is set to 10, and the Gini coefficient is used as the division criterion. After the model training is completed, export the model to the embedded control terminal, and the terminal calls the inference module to automatically output the heating power and humidification control signal according to the real-time data input and the optimized interval, so as to realize the automatic adjustment of the incubator environment.

[0022] Step S1 includes the following steps: Step S11: Deploy electronic monitoring devices, temperature and humidity sensors, and infrared temperature sensors in multiple directions inside the incubator, and collect the all-round morphological images of the hatching eggs through the electronic monitoring devices to obtain the all-round morphological images of the hatching eggs; Step S12: According to the all-round morphological images of the hatching eggs, and monitor the heating temperature state of the outer surface of the eggshell in multiple directions through the infrared temperature sensors to obtain the monitoring data of the heating temperature state of the outer surface of the eggshell; Step S13: Perform time-series fluctuation analysis on the monitoring data of the heating temperature state of the outer surface of the eggshell to obtain the time-series fluctuation data of the surface temperature; Step S14: Perform feature analysis on the time-series fluctuation data of the surface temperature to obtain the time-series fluctuation feature data of the surface temperature.

[0023] In the embodiment of the present invention, during the implementation of step S11, in the internal space of a thermostatic incubator with a standard inner tank volume of 480 liters, following the principle of uniform spatial layout, 8 groups of electronic monitoring devices are deployed on the top, bottom, and four side walls. Each group of devices uses a CMOS image sensor with a pixel of 2592×1944, a fixed focal length of 4 mm, a horizontal viewing angle of approximately 72 degrees, and the image acquisition time interval for each frame is set to 2 seconds, with the coverage range controlled within 90% of the internal space. The image sensor is combined with a linear polarization filter and an infrared supplementary light module to achieve high-frequency image acquisition of the surface structure contour and suspicious thermal reflection areas of the hatching eggs under low visible light conditions. Each hatching egg obtains no less than 24 frames of images from different angles. After multi-view geometric reconstruction using a stereo calibration device with a marker board, a dense image registration algorithm is used to perform sub-pixel stitching and fusion on the images from all angles, and finally, an omnidirectional morphological image of a single hatching egg is generated, with the output format being a three-channel 16-bit RGB image and the image resolution not lower than 2048×2048 pixels. During the implementation of step S12, 1 to 2 infrared temperature sensors are arranged above the surface of each hatching egg to cover the temperature measurement paths with different incident angles. The sensor model used is a thermopile infrared temperature sensor with a far-infrared band response range of 8μm to 14μm and a response time of less than 0.5 seconds. The sensor sampling frequency is 1Hz, the sensing distance is set to 10 cm, and each hatching egg collects no less than 100 groups of temperature data during the acquisition period. All data is synchronized with the image acquisition time through timestamp alignment processing, and coordinate registration is performed in combination with the pixel block position of the eggshell surface area in the image. Furthermore, the infrared temperature measurement results are mapped to the hatching egg morphological image point by point to construct a temperature state monitoring data set for the outer surface of the eggshell. The data structure includes the acquisition time, three-dimensional coordinates of the acquisition point, temperature value, and mapping parameters with the image coordinate system. During the implementation of step S13, based on the temperature value sequence of each spatial point changing with time in the temperature state monitoring data of the outer surface of the eggshell, an equally spaced time series temperature fluctuation sequence is constructed. Each time series is preprocessed by denoising using the five-point moving average method, and the first-order difference method is introduced to extract the change trend to form a time-temperature change rate sequence for each sampling point. After Z-normalization processing, continuous heating and cooling intervals are extracted through local extreme value detection and variance analysis within the sliding window interval, and then a temperature fluctuation identification sequence is generated for each eggshell monitoring point. Finally, the temperature fluctuation sequences of all monitoring points are integrated according to the spatial position to generate surface temperature time series fluctuation data in three-dimensional space, with a time accuracy of 1 second and a spatial accuracy of 1 point per square centimeter. During the implementation of step S14, based on the obtained surface temperature time series fluctuation data, the main frequency, energy distribution, and periodic characteristics of each temperature time series are analyzed using the discrete Fourier transform, and the samples with the amplitude of the high-frequency component exceeding the reference threshold are identified as potential abnormal fluctuation points.The principal component analysis method is used to reduce the dimension of all time series, and combined with the region clustering algorithm based on spectral clustering to identify the temperature fluctuation characteristics of the eggshell surface. In each clustering region, indicators such as its frequency domain energy distribution, mean square fluctuation intensity, and fluctuation duration period are statistically analyzed. Finally, a surface temperature time series fluctuation characteristic data structure including spatial position, frequency component, period type, energy level, and abnormal fluctuation index is generated. This data structure takes each egg as the basic unit, is encapsulated into a time stamp synchronized data frame, and is indexed and associated with the morphological image and the original temperature monitoring data for subsequent heat conduction and development prediction analysis.

[0024] Step S2 includes the following steps: Step S21: Real-time data collection of the temperature and humidity in the incubator interior space is carried out through a temperature and humidity sensor to obtain real-time temperature and humidity data of the space; Step S22: Based on the real-time temperature and humidity data of the space and the all-round morphological image of the hatching eggs, temperature concentration recognition of the surface temperature time series fluctuation characteristic data is carried out to obtain temperature concentration data of the heat conduction region; Step S23: Distribution enhancement gradient numerical quantization is carried out on the temperature concentration data of the heat conduction region to obtain temperature distribution enhancement numerical gradient data; Step S24: Based on the temperature distribution enhancement numerical gradient data and the all-round morphological image of the hatching eggs, simulation prediction of the abnormal degree of embryo development in the eggs is carried out to obtain prediction data of the abnormal degree of embryo development.

[0025] As an example of the present invention, referring to Figure 2 As shown, in this example, step S2 includes: Step S21: Real-time data collection of the temperature and humidity in the incubator interior space is carried out through a temperature and humidity sensor to obtain real-time temperature and humidity data of the space; In the embodiment of the present invention, a micro digital temperature and humidity sensor is used to perform multi-point real-time collection of the interior space environment data of the incubator. The temperature measurement range of this type of sensor is -40 degrees Celsius to 80 degrees Celsius, the humidity measurement range is 0% to 100%RH, the temperature accuracy is ±0.5 degrees Celsius, the humidity accuracy is ±2%RH, and the sampling period is set to once every 5 seconds. A 6×4×3 regular grid layout is implemented for the three-dimensional space along the X, Y, and Z directions inside the incubator, and a total of 72 sampling points are arranged, each point is marked with an independent number. The sensor array is connected to the central data acquisition module through a serial communication interface, and the temperature and humidity values at the corresponding time stamps are recorded at each sampling point. All the data is integrated to generate a time series type real-time temperature and humidity dataset of the space. This dataset is stored in a two-dimensional table structure, each row records the three-dimensional coordinates, acquisition time, temperature value, and humidity value of a sampling point, the data frequency is 0.2Hz, and the data accuracy remains the same as the original sensor resolution.

[0026] Step S22: Based on the real-time spatial temperature and humidity data and the omnidirectional morphological images of the hatching eggs, perform thermal conduction region temperature concentration identification on the surface temperature time-series fluctuation characteristic data to obtain thermal conduction region temperature concentration data; In the embodiment of the present invention, based on the obtained real-time spatial temperature and humidity data, the overlapping region with the three-dimensional morphological image of the outer shell surface of the hatching egg is selected for corresponding processing. By establishing a unified spatial coordinate system, the sensor sampling point coordinates of the temperature and humidity data are projected and mapped to the adjacent region on the three-dimensional surface model of the eggshell, and a continuous temperature and humidity distribution map is generated on the eggshell model by using the weighted inverse distance interpolation method. Combining the surface temperature time-series fluctuation characteristic data obtained in the previous step, the temperature fluctuation value and the real-time humidity gradient change rate are compared at the same spatial point, and the bivariate correlation analysis method is used to identify the region where the temperature rapidly rises or continuously drops under specific temperature and humidity environmental conditions, and this region is defined as the heat conduction concentration area. During this process, the temperature change rate threshold is set to 0.8 degrees Celsius per minute, and the humidity change rate threshold is set to 2%RH per minute. The continuous points that meet the conditions are selected to form a heat conduction region point group, and the Marching Cubes method is used to generate the boundary body of the three-dimensional heat concentration region, and a thermal conduction region temperature concentration data structure is constructed, which includes region voxel coordinates, temperature and humidity gradient values, and time interval labels.

[0027] Step S23: Perform distribution enhancement gradient numerical quantization on the thermal conduction region temperature concentration data to obtain temperature distribution enhancement numerical gradient data; In the embodiment of the present invention, gradient enhancement processing is performed on the above thermal conduction region temperature concentration data. The central difference method is used to solve the three-dimensional direction temperature gradient vector of each spatial point, and a three-dimensional temperature gradient field is constructed based on the gradient modulus value. To highlight the local change difference, a high-pass Laplacian filter is applied in the gradient modulus map for edge enhancement, and then a piecewise linear mapping function is used to normalize the gradient value to the 8-bit gray space between 0 and 255 to construct a temperature distribution enhancement numerical map. To achieve the usability of the data structure, the numerical map is unfolded according to the spatial voxel structure to form a voxel-level numerical grid, and each voxel contains the corresponding gradient intensity value and the unit vector information of the heat conduction direction. During the numerical quantization process, the adjacent voxels with an angular difference in the gradient direction exceeding 30 degrees are merged using the spatial connectivity compression rule to form a numerical gradient data body with enhanced local consistency, and the output data is independently archived for each eggshell as a unit.

[0028] Step S24: Based on the temperature distribution enhancement numerical gradient data and the omnidirectional morphological images of the hatching eggs, perform simulation prediction on the abnormal degree of embryo development in the eggs to obtain embryo development abnormal degree prediction data.

[0029] In the embodiments of the present invention, based on the temperature distribution to strengthen the numerical gradient data and the three-dimensional morphological image of the hatching egg, a mapping relationship of the in-embryo temperature influence area is established based on the heat conduction channel on the path from the eggshell to the embryo center. In the three-dimensional morphological image, multi-directional radial path lines from the eggshell to the egg center are generated based on the radius growth method. Temperature gradient data and its direction vectors are extracted along the path lines for cumulative integration to form a path energy distribution curve. By comparing the means and standard deviations of different path energy curves, temperature conduction abnormal areas in a specific direction are identified. The single-value heteroscedasticity identification method is introduced, and the direction with the path energy curve variance exceeding twice the standard deviation of the normal sample mean is marked as an abnormal channel. Finally, an index system for the degree of abnormal embryo development is constructed by combining indicators such as the number of abnormal channels, the heat asymmetry distribution rate, and the conduction direction deviation angle of each hatching egg. The index value is normalized to between 0 and 1 to generate prediction data for the degree of abnormal embryo development. The prediction data is output in JSON format, including egg number, abnormal index, abnormal direction distribution angle set, and time tag, for use as input in the subsequent temperature control adjustment process.

[0030] Step S22 includes the following steps: Step S221: Perform a curvature structure calculation on the all-round morphological image of the hatching egg to obtain the curvature structure data of the hatching egg; perform a terminal meandering angle calculation on the all-round morphological image of the hatching egg to obtain the terminal meandering angle of the hatching egg; Step S222: Based on the curvature structure data of the hatching egg and the terminal meandering angle of the hatching egg, perform texture rough surface distribution difference identification on the all-round morphological image of the hatching egg to obtain texture rough surface distribution difference data; Step S223: Perform axial ratio analysis based on the curvature structure data of the hatching egg and the terminal meandering angle of the hatching egg to obtain the axial ratio data of the hatching egg; Step S224: Extract the humidity increment interval from the real-time space temperature and humidity data to obtain the space humidity increment interval; based on the space humidity increment interval, perform heat capacity increase interval analysis on the real-time space temperature and humidity data to obtain the heat capacity increase interval; Step S225: Evaluate the effective conversion increment of medium heat conduction according to the heat capacity increase interval to obtain medium heat conduction increment data; Step S226: Perform heat conduction region diffusion fluctuation difference clustering on the surface temperature time series fluctuation characteristic data according to the texture rough surface distribution difference data, the axial ratio data of the hatching egg, and the medium heat conduction increment data to obtain heat conduction diffusion fluctuation difference clustering data; Step S227: Perform heat conduction region temperature concentration identification on the heat conduction diffusion fluctuation difference clustering data to obtain heat conduction region temperature concentration data.

[0031] In the embodiment of the present invention, the operation of step S221 first performs three-dimensional curvature structure calculation and processing based on the omnidirectional morphological image of the hatching egg. The image data is sourced from the acquisition by CMOS industrial cameras arranged from multiple angles. The image resolution is set to 1920×1080 pixels, and the spatial contour information of the hatching egg is determined through structured light assisted calibration. The discrete Gaussian curvature approximation method is used to deduce the curvature of the image boundary contour point set, where the Gaussian curvature value of each local neighborhood is deduced by obtaining the curvature radius by the five-point method and then calculating the specific value. The processing unit area of each image is set to 1 square millimeter, and the tangent change rate boundary is delimited into five equal parts within the segmentation area to generate the curvature structure data of the hatching egg. On this basis, the contour extraction algorithm is used to enhance the end features, and the points with sharp curvature changes (the set threshold is ±20 degrees / mm) are selected to identify the end sharp corner area. By measuring the angle between the main curvature direction of this area and the main axis of the egg body, the vector projection method is used to calculate the end meandering angle, and finally the end meandering angle data of the standardized output hatching egg is formed. The identification of the texture rough surface distribution difference is carried out based on the curvature structure data and the end meandering angle obtained in step S221. The local texture direction consistency index of the hatching egg image is extracted by the gray level co-occurrence matrix method, the parameter window size is 21×21 pixels, and the step size is 7 pixels. The spatial distribution matching is carried out using the cosine similarity of the included angle between the curvature change direction and the texture gradient direction. It is defined that the area with the included angle between 0 and 30 degrees is the smooth area, and the area with the included angle greater than 60 degrees is the rough surface area, and a binary layer distribution image based on direction dissimilarity is constructed. Further, the image difference expansion algorithm is used to extract the texture density change trend in different regions and normalize the output texture rough surface distribution difference data. By analyzing the relationship between the main axis distribution and the meandering angle in the curvature structure data, the pixel span of the long axis and the short axis of the hatching egg is determined on the three-dimensional model using the central axis centroid symmetry projection method, and the unit scale is millimeter. The original axial ratio value is obtained by dividing the long axis pixel value by the short axis pixel value, and then it is normalized by the standard deviation to exclude the influence of individual scale differences, and finally the axial ratio data of the hatching egg is output.

[0032] The extraction of the humidity increment interval is based on the incubator space humidity data recorded at a frequency of once every 3 seconds. A sequence is formed every 60 seconds. Local peak detection and moving average difference processing are performed on each humidity sequence to identify humidity sudden increase points above 1.5%, which are defined as the start of the humidity increment interval. The position where the humidity growth rate is less than 0.2% / second in the subsequent time period is set as the end of the increment interval. Each extracted humidity increment interval is accompanied by the recording of the start, peak, and end time points, constituting structured space humidity increment interval data. Based on this data, using the air specific heat capacity derivation formula, combined with the measured humidity concentration and temperature base value, the total heat capacity increment change amount per unit air mass is calculated, and the heat capacity rising interval is calculated in a step-by-step cumulative manner with a sliding time window (60 seconds), so as to output the heat capacity rising interval data. Using the time coincidence segment of the heat capacity rising interval data and the humidity increment interval data, the Fourier heat conduction approximation formula is used to inversely deduce the change rate of thermal energy in unit volume of air under the influence of humidity change. The heat conduction increment is calculated by converting the latent heat carried by humidity molecules. In the calculation process, first, the change in the water vapor content in the air is extracted, and combined with the air thermal conductivity and the space volume, the increase ratio of the overall thermal energy diffusion speed by water vapor is calculated. The unit is joules per cubic meter per degree Celsius, and it is recorded as the medium heat conduction increment data. The heat conduction diffusion fluctuation difference clustering adopts a three-channel data fusion strategy. First, the texture rough surface distribution difference data, the axial ratio data, and the medium heat conduction increment data are respectively normalized to the range of 0 to 1. The Euclidean distance is used to construct the feature space, and high-density point division processing is carried out. The density peak clustering method is used to extract high-similar clusters with concentrated heat conduction fluctuation trends in the feature space. The local fluctuation extreme difference of heat conduction diffusion is calculated for each cluster and marked in the eggshell image in the form of three-dimensional coordinates to generate heat conduction diffusion fluctuation difference clustering data. The local heat aggregation index extraction method is used for the heat conduction diffusion fluctuation difference clustering data to identify the temperature concentration in the heat conduction region. This method is based on the central point of the spatial distribution of each clustering cluster, and the fluctuation gradient range within 5 mm outward is extended to extract the region where the gradient change rate is not less than 0.15 degrees Celsius per millimeter. All regions that meet the aggregation conditions are encapsulated by the polygon convex hull algorithm to form the heat conduction region boundary data. Combining the corresponding temperature interval ranges of each clustering cluster, they are uniformly converted into a spatial heat intensity layer, and finally the heat conduction region temperature concentration data is output.

[0033] Step S24 includes the following steps: Step S241: Perform embryo tissue protein denaturation entropy change analysis based on the temperature distribution enhanced numerical gradient data to obtain embryo protein denaturation entropy change data; Step S242: Simulate the embryo heat damage cascade amplification effect on the embryo protein denaturation entropy change data to obtain the embryo heat damage cascade amplification effect; Step S243: Estimate the exponential growth of the oxygen consumption demand of the embryo based on the enhanced numerical gradient data of the temperature distribution to obtain the exponential growth data of the oxygen consumption demand; Step S244: Identify the density of eggshell pores in the omnidirectional morphological image of the hatched egg to obtain the eggshell pore density data; Step S245: Simulate and predict the degree of embryonic hypoxic metabolic disorder of the exponential growth data of the oxygen consumption demand according to the eggshell pore density data to obtain the predicted data of the degree of hypoxic metabolic disorder; Step S246: Simulate and predict the degree of abnormal development of the embryo in the egg based on the embryonic heat damage cascade amplification effect and the predicted data of the degree of hypoxic metabolic disorder to obtain the predicted data of the degree of abnormal embryo development.

[0034] In the embodiment of the present invention, in step S241, based on the enhanced numerical gradient data of the temperature distribution obtained in step S23, first, a two-dimensional thermal projection map with a time resolution of 10 seconds and a spatial resolution of 0.5 mm is used as the basic analysis unit to construct an equivalent thermal energy field inside the embryo. The thermal energy field is bounded by the three-dimensional surface reconstruction data of the external image of the hatched egg, and the temperature information of the hollow area is filled by the two-way Laplace interpolation method inside to ensure the continuity and curvature adaptability of the temperature distribution. After the temperature field is constructed, for the local area with a large heat conduction intensity, that is, the pixel block area where the temperature numerical gradient is greater than 0.7 °C / mm, the equivalent tissue blocks are divided by the method of regional segmentation, and the single block area is fixed at 4 mm². For each tissue block, a unit thermal perturbation response curve is constructed according to the temperature change rate of the block under the time series. Further, the Markov state transition theory is used to model the thermal state evolution sequence of the unit area, and a state transition probability matrix is constructed. The state space is divided into four levels of temperature perturbation levels, namely low perturbation (<0.1 °C / s), medium perturbation (0.1 - 0.3 °C / s), high perturbation (0.3 - 0.6 °C / s), and extremely high perturbation (>0.6 °C / s). Based on the transition probability distribution, the path uncertainty of the unit tissue block is calculated, and the entropy value of the elements outside the main diagonal of the transition matrix is deduced using the Shannon entropy formula to obtain the thermal state perturbation entropy change value of the tissue block during the hatching cycle. This entropy change value is used to evaluate the randomization trend of the spatial configuration of the protein chain caused by high-frequency thermal perturbation, that is, the energy manifestation form of the protein denaturation behavior. To correct the influence of the boundary curvature on the local thermal perturbation focusing effect, a regulation factor based on the curvature radius r needs to be introduced. In the area where r is less than 10 mm, an enhancement coefficient r - ¹ is used to weight the entropy change value. Subsequently, all tissue blocks with entropy change values greater than 1.2 × the average entropy change value of the whole region are marked as high denaturation regions, and their spatial position indexes and corresponding entropy change values are output to form complete "embryonic protein denaturation entropy change data". This data participates in the cumulative simulation analysis as a basic variable for simulating thermal damage and metabolic loss in subsequent steps.

[0035] When simulating the embryo thermal damage cascade amplification effect on the embryo protein denaturation entropy change data obtained in step S241, first, the internal structure of the entire hatching egg is segmented according to a three-dimensional spatial grid, and the segmentation unit is a 3mm×3mm×2mm cube. The entropy change value within each cube is continuously adjusted with the surrounding area through bicubic interpolation to generate a complete spatial entropy change field distribution map. Subsequently, based on the principle of thermal damage cascade amplification, a response mapping relationship between instantaneous thermal perturbation and weakened structural stability is established in the time axis direction. Specifically, a sliding time window technique is used to construct the thermal stress accumulation factor of the embryo tissue unit under the background of continuous entropy change increase. The window length is set to 60 seconds, and the step size is 10 seconds. For each tissue unit, the entropy change rising rate is compared with the entropy change accumulation of the adjacent six faces. If the entropy change increment of the current unit is greater than 1.5 times the average entropy change increment of the surrounding area, then this unit is determined as a potential thermal damage amplification center and is recorded as entering the cumulative activation state. For all activated state units, an extended action domain is formed by expanding 1.5 times the cubic size outward from its center, and the entropy change increment of all adjacent units is detected. If there are more than two entropy change synchronous growth phenomena within three consecutive time windows, it is regarded as constituting a thermal damage amplification chain. The length, number of nodes, and propagation path direction of each chain are calibrated in space, and the chain coupling amplification coefficient is calculated for each chain. This coupling amplification coefficient is composed of the product of the ratio of the maximum entropy change to the minimum entropy change in the chain and the propagation time span, so as to quantitatively measure the degree of gradually irreversible damage caused by the thermal perturbation propagating from the initial point to multi-level tissues. Finally, all regions constituting an effective chain response are marked as the "thermal damage cascade amplification effect area", and structural description data such as the starting time point, spatial coordinates, chain length, coupling amplification coefficient, and number of chain branches corresponding to each region are output to form the data result of the "embryo thermal damage cascade amplification effect".

[0036] When estimating the exponential growth of the embryo's oxygen consumption demand based on the temperature distribution enhanced numerical gradient data obtained in step S23, first, a heat flux density distribution map is constructed on the three-dimensional projection grid of the surface of the hatching egg according to the enhanced numerical gradient data. The heat flux calculation is based on the temperature gradient per unit area per unit time multiplied by the heat conduction constant. The selected heat conduction constant is 0.55 W / m·K, and the unit of heat flux is W / m². Based on the heat flux density map, the area with a heat flux higher than 1.3 times the average heat flux is selected and marked as the "metabolic stress response area". The surface points of this area are projected vertically inward in the direction of the internal curvature of the hatching egg to obtain the predicted active metabolic units inside. For each active metabolic unit, the rate of change of its temperature gradient is extracted as the metabolic activation factor. The exponential growth evaluation period is set to 120 seconds. During this period, the time t experienced by the temperature gradient rising from the initial level to the peak, and the ratio r of the peak value to the initial value are statistically analyzed. Taking r as the base and t as the exponential growth power, the oxygen consumption demand growth factor is calculated. All active units are evaluated one by one in this way. According to the relationship between the physiological oxygen demand of the embryo, the basic unit oxygen consumption is set to 0.38 μmol per second. Finally, the exponential growth factor is multiplied by the basic oxygen consumption value to obtain the predicted oxygen consumption rate of the unit per unit time. The sum of the inside of the entire hatching egg is calculated, and the total oxygen consumption demand growth data is output. To express the non-linear change trend of the oxygen consumption demand with the expansion of the metabolic area, it is also necessary to record the expansion speed and the area increment rate of the metabolic stress response area in the time axis direction, and draw an exponential fitting curve. Variables such as the exponential growth coefficient, the growth starting point coordinates, the growth radius change rate, and the total oxygen consumption estimate are used to construct the "exponential growth data of oxygen consumption demand" for subsequent modeling of the hypoxia risk state and simulation of metabolic disorders.

[0037] In the specific implementation process of identifying the density of eggshell pores from the omnidirectional morphological images of hatching eggs, first, a multi-angle synchronous surrounding lighting device is used to collect 360-degree rotation images of a single hatching egg. The image acquisition resolution is set to 2048×2048 pixels, and the acquisition angle interval does not exceed 4 degrees to ensure that there is no visually occluded area on the eggshell surface. All the collected images are subjected to distortion correction processing. The correction algorithm uses spherical unfolding mapping combined with a consistency filtering mechanism based on the gradient direction to restore the true geometric texture distribution of the eggshell surface. In the processed images, an algorithm based on Laplace edge enhancement combined with gray-level local extreme point detection is used to extract the micropore regions on the eggshell surface. To ensure the extraction accuracy, a local contrast enhancement function is introduced at the pixel level to overcome the edge determination deviation caused by uneven illumination. In each single frame of the image, a binary segmentation layer is constructed, and the dynamic threshold is adjusted by analyzing the mean and standard deviation of the gray-level distribution in the 8-neighborhood around each pore pixel point to effectively exclude the interference of misidentification of non-genuine pores. After the identification is completed, an image grid is divided for each unfolded eggshell image, with each grid being 100×100 pixels. The number of effective pores in each grid is counted, the number of pores per unit area is calculated, and the average pore diameter is recorded. The unit area is defined as 1 square millimeter, and the corresponding pixel scale is obtained through an image calibration board. Finally, information such as the mean value, standard deviation, maximum and minimum values of the pore density in each grid is integrated into a structured matrix to form the "eggshell pore density data", and the coordinates of the abnormally high-density or low-density regions are marked for subsequent judgment of the regional limitation of oxygen exchange ability. When simulating and predicting the degree of embryonic hypoxic metabolic disorder based on the exponentially growing data of oxygen consumption demand according to the eggshell pore density data, first, the eggshell pore density data obtained in step S244 is projected onto a three-dimensional eggshell model corresponding to the space of the embryonic active metabolic region in step S243. A resistance function of the pore spatial distribution to the embryonic oxygen diffusion path is constructed. The resistance function is constructed based on the number of pores per unit area and the reciprocal of the average pore diameter. The critical density of pore permeability is set to 20 pores per square millimeter, and the region with an average diameter less than 20 microns is marked as a high-diffusion resistance area. The volume distribution of all the marked high-resistance areas is measured. If its coverage ratio on the outer shell of the embryonic high-oxygen consumption region exceeds 40%, it is recorded as a potential hypoxic region of the embryo. Further, a path function for oxygen diffusion to the oxygen-consuming unit is established, and the average diffusion distance between the embryonic oxygen consumption center and the point of maximum pore density is measured by the Euclidean shortest path. The oxygen flux is calculated in combination with the permeability resistance function of the pore region. If the oxygen flux is lower than 75% of the oxygen consumption demand per unit of basic metabolism and lasts for more than 180 seconds, it is determined that metabolic oxygen supply imbalance occurs in this region. The difference trajectory between the oxygen consumption index and the actual oxygen supply of each oxygen-consuming unit under different temperature gradient perturbations is simulated on the time axis, the maximum difference point is recorded, and the cumulative value of the difference over time is used as the metabolic disorder factor.During the entire simulation period, if the metabolic disorder factor exceeds 10 times the basal metabolic threshold, the output is a high-risk area of hypoxic metabolic disorder, and multiple indicators such as the time-dynamic curve, spatial coordinates, and hypoxia level of this area are generated, and integrated to form "prediction data on the degree of hypoxic metabolic disorder". When simulating and predicting the degree of abnormal development of the embryo in the egg based on the embryo heat damage cascade amplification effect obtained in step S242 and the prediction data on the degree of hypoxic metabolic disorder obtained in step S245, first align the data spaces of the two, and construct a superimposed layer of dangerous areas in the three-dimensional hatching egg body. By analyzing the spatial intersection degree between the terminal position of the heat damage chain and the metabolic disorder area, the synergy degree of the multi-factor stress area is judged. If the spatial coincidence degree of the two exceeds 60%, it is constructed as a high-composite stress area. In the high-composite stress area, an embryo stress response index is generated according to the weighted average of the heat damage chain coupling amplification coefficient and the metabolic disorder factor. This index is processed by moving average in the time series, and the window length is set to 240 seconds to smooth the dynamic response curve. In each stress response area, four indicators including its volume, duration, response peak value, and propagation direction are calculated to construct a regional influence factor. According to the stress response index and regional influence factor of all areas, a rank label is constructed using the regional ranking method, classified from level one (low influence) to level five (extremely high influence), and combined with the embryo development time axis to deduce the probability threshold of physiological function disorder during its development process. Finally, all information such as the abnormal development level, position coordinates, area and volume, and index dynamic curve of all areas is sorted into "prediction data on the degree of abnormal embryo development" for the basis of feedback on the internal adjustment strategy of the incubator.

[0038] Step S242 includes the following steps: Perform entropy flux density matrix reconstruction processing on the embryo protein denaturation entropy change data to obtain an entropy flux spatial distribution tensor; Perform enzyme inactivation equivalent change analysis on the embryo protein denaturation entropy change data according to the entropy flux spatial distribution tensor to obtain enzyme inactivation equivalent change data; Perform multi-phase kinetic weakening rate deduction on the enzyme inactivation equivalent change data to obtain the enzyme inactivation multi-phase weakening rate; Perform numerical integration of the cell division arrest time sequence based on the enzyme inactivation multi-phase weakening rate to obtain the cell division arrest time sequence value; Perform simulation of the embryo heat damage cascade amplification effect on the embryo protein denaturation entropy change data according to the enzyme inactivation multi-phase weakening rate and the cell division arrest time sequence value to obtain the embryo heat damage cascade amplification effect.

[0039] In the embodiments of the present invention, based on the entropy change data of embryonic protein denaturation obtained in step S241, a three-dimensional tensor expression construction is carried out for the distribution state of the entropy change value in different time sections and spatial regions. First, taking the temperature distribution enhanced numerical gradient data corresponding to each embryonic structure surface image as the horizontal axis input, taking the entropy change value per unit area corresponding to each temperature point as the vertical axis value, and taking the entropy change coefficient corresponding to different time sections as the tensor dimension, the three-dimensional tensor is reconstructed into an entropy flux density matrix. The dimension of this matrix is set to 64×64×t, where t is the number of sampling times, the total time span is set to 600 seconds, the sampling interval is 10 seconds, and each element of the matrix is the entropy flux intensity of protein denaturation per unit area under unit temperature gradient, in units of joules per kilogram kelvin per second. By performing entropy flux gradient fitting on the continuous tensor dimension and introducing fifth-order cubic b-spline interpolation to fit and smooth the entropy flux distribution differences at different time points, an entropy flux spatial distribution tensor is obtained, ensuring that the curvature continuity of the changes in the components in each spatial direction in adjacent time segments is below the third derivative, so as to meet the stable call of various subsequent differential operators. Based on the aforementioned obtained entropy flux spatial distribution tensor, a dissection of the proportion of entropy change loss in the main metabolic enzyme active regions related to embryonic proteins is carried out, and the dissection range is the region in the tensor where the temperature gradient is greater than 5 degrees Celsius per millimeter. An isentropic surface fitting is performed on the entropy flux in each region, and the Gaussian curvature weighted average method is used to solve the thermal response threshold of the enzymatic hydrolysis path in this region. Then, based on the mutation points of the change in the thermal response threshold, the critical stability coefficient of local enzyme activity is calibrated. For the regions where the stability coefficient is lower than 0.5, according to the local enzyme mass conservation law, a conversion coefficient equal to the influence of unit entropy change on enzyme activity is introduced, with the unit of moles per joule. According to this coefficient, a weighted integral is performed on the entire entropy flux spatial distribution tensor, and finally the equivalent change data of enzyme inactivation corresponding to each region is obtained, and the result is output in the form of a percentage change in enzyme concentration. The equivalent change data of enzyme inactivation is used for the deduction of the multi-phase kinetic weakening rate. The specific operation is as follows: First, the change trend of the time gradient in the data is differentiated, and the five-point central difference calculation is performed on the enzyme concentration change rate between any two time points, and the time window is set to 50 seconds. According to the obtained time series rate curve, it is divided into three segments: a linear change period, a plateau period, and a sharp change period, and the rate values in different regions are divided into three categories: fast change, medium change, and slow change by the K-means clustering method. Each type of change rate is respectively fitted into three function forms: exponential, logarithmic, and linear. The optimal fitting form is determined by the minimum sum of squared residuals criterion. Finally, the multi-phase weakening rate of enzyme activity with time in each spatial region is obtained, and the result is output as a concentration decay function in units of seconds and is used for the subsequent construction of the time series cascade of the thermal damage amplification effect. This process strictly depends on the coincidence of the tensor partition boundary and the actual spatial variation distribution of the temperature gradient, and it is necessary to ensure the continuity of the first derivative of the fitting function at the boundary between each weakening region to prevent jump errors in subsequent coupled modeling.In the implementation process of numerically integrating the timing of cell division arrest based on the multi-phase decay rate of enzyme inactivation, first obtain the multi-phase decay rate data of enzyme inactivation obtained in the previous step. The data structure is a three-dimensional spatio-temporal tensor with dimensions of 64×64×t, where the value of each unit represents the decay rate of unit enzyme concentration per unit time at that spatial position, and the unit is the change in molar concentration per second. According to the phase of the dependence of cells on metabolic enzyme activity during the embryonic cell division cycle, set the minimum enzyme activity threshold required for each cell division cycle to 0.65 mol / L. Search for the time node when the enzyme concentration drops below this threshold point by point in the tensor data, and record the duration between this node time and the initial moment, which is defined as the first cell division arrest time corresponding to that position. Perform a weighted average of the arrest times at all spatial positions, with the weight being the magnitude of the gradient value in the gradient data of the temperature distribution enhancement value at that position, to reflect the dominance of the temperature rise intensity on the cell cycle. Then perform a time-domain integration of the average arrest time series over the entire embryonic region based on the fourth-order Runge-Kutta method, with a step size set to 5 seconds and an integration upper limit of the total experimental duration of 600 seconds. The output result is the numerical value of the cell division arrest timing, whose unit is the cumulative arrest duration in seconds. This value represents the total delay time of the overall embryonic cell division process due to insufficient metabolic enzyme activity under the given temperature and enzyme inactivation conditions. In the implementation process of simulating the cascade amplification effect of embryonic thermal damage on the embryonic protein denaturation entropy change data based on the multi-phase decay rate of enzyme inactivation and the numerical value of the cell division arrest timing, first multiply the unit entropy change rate at each spatial position in the entropy change spatial distribution tensor by the enzyme inactivation rate corresponding to that position to obtain the weighted entropy change data based on the enzyme activity reduction weight. This data can be used to deduce the non-linear response degree of protein secondary structure damage in the spatial and time dimensions. Then use the numerical value of the cell division arrest timing as an enhancement factor in the time dimension to perform exponential weighted cumulative processing on the above weighted entropy change data. The specific method is to multiply the weighted entropy change at each position by e to the power of the cell division arrest time multiple. This processing logic is based on the biological mechanism that the lag of the cell repair mechanism verified by experiments leads to an accelerated rate of thermal damage accumulation. The entropy change data obtained after processing is the thermal damage response intensity tensor after cascade amplification processing, with the unit of joules per kilogram per second. Perform a maximum normalization operation on this tensor in the time dimension and extract an isosurface in three-dimensional space to indicate the propagation range and degree of the thermal damage response intensity within the three-dimensional structure. Finally, by statistically calculating the volume ratio of the isosurface and mapping it back to the time evolution axis, form the simulation output data of the embryonic thermal damage cascade amplification effect, which is expressed as the entropy change enhancement magnification curve per unit time per unit embryonic volume and is used for subsequent reverse derivation processing of embryonic development abnormality prediction and environmental regulation strategies. In all the above operations, no prediction model is introduced, and only two physical quantity transformation mechanisms, namely time series numerical integration and exponential weighted accumulation, are used to ensure that the process is traceable and the parameters have clear biophysical meanings, and to avoid interference from non-deterministic functions in the causal chain analysis.

[0040] Step S245 includes the following steps: Perform a calculation and estimation of the eggshell gas exchange capacity per unit time based on the eggshell pore density data to obtain the estimated eggshell gas exchange capacity per unit time; Based on the estimated eggshell gas exchange capacity, perform a quantitative assessment of the lack of embryo oxygen content for the exponentially growing data of oxygen consumption demand to obtain the quantitative data of the lack of embryo oxygen content; Based on the quantitative data of the lack of embryo oxygen content, perform an assessment of the compensatory enhancement of glycolysis under the stress state of the embryo to obtain the data of the compensatory enhancement of glycolysis; Perform an analysis of the lactate accumulation increment rate ratio on the data of the compensatory enhancement of glycolysis to generate the lactate accumulation increment rate ratio; Based on the lactate accumulation increment rate ratio, estimate the probability of embryo myocardial contractility loss to obtain the probability of embryo myocardial contractility loss; Based on the lactate accumulation increment rate ratio and the probability of embryo myocardial contractility loss, simulate and predict the degree of embryo hypoxic metabolic disorder to obtain the predicted data of the degree of hypoxic metabolic disorder.

[0041] In the embodiment of the present invention, during the implementation process of performing a calculation and estimation of the eggshell gas exchange capacity per unit time based on the eggshell pore density data, first, based on the eggshell pore density data extracted in step S244, perform a block counting process on the pore regions identified by the edge enhancement + region growing joint segmentation algorithm in the image on a two-dimensional plane. Each block has a size of 5 mm × 5 mm. Count the number of pores and the total pore area per unit area, and divide the total pore area by the actual projected area of the region to obtain the local pore opening rate data, with the unit being a dimensionless ratio. Subsequently, combined with the diffusion coefficient values of oxygen and carbon dioxide in the porous calcium matrix under standard atmospheric pressure, which are respectively and Square centimeters per second. Substitute the stomatal density, average pore diameter, and pore depth (derived from the cross-sectional image, with the average value taken as 0.3 mm) and the gas diffusion coefficient into the classical Stefan-Maxwell diffusion flux expression logic. Multiply the gas flux per unit area per second by the effective ventilation area of the whole shell to obtain the estimated capacity of eggshell gas exchange per unit time, with the unit of millimoles per second. This data structure is a multi-channel tensor, and the number of channels corresponds to two types of gases, oxygen and carbon dioxide respectively. Its spatial distribution reflects the heterogeneity of gas exchange capabilities in each region. During the implementation of quantitatively evaluating the lack of embryonic oxygen content based on the exponentially increasing data of oxygen consumption demand with the estimated capacity of eggshell gas exchange, extract the exponentially increasing data of embryonic oxygen consumption demand output in step S243. The data structure is a two-dimensional time series, and each time point is marked with the demand flux value of oxygen per unit mass of the corresponding embryo, with the unit of millimoles per gram per second. Divide the total value of the eggshell oxygen exchange capacity per unit time by the total mass of the embryo to obtain the oxygen supply capacity per unit mass, with the unit of millimoles per gram per second. Then compare it hour by hour with the oxygen consumption demand data per unit mass, take the difference and limit it within the positive and negative intervals. A negative difference represents hypoxia, and a positive difference represents surplus oxygen supply. Multiply the negative difference at each time point by the corresponding time duration and accumulate it, which is the quantitative data of embryonic oxygen content lack, with the unit of millimoles per gram, representing the total cumulative hypoxia faced by the whole embryo during the critical hatching stage. This processing adopts the time-weighted integration logic and does not introduce fitting or regression models to ensure that the calculation process corresponds one by one with physical phenomena. During the implementation of evaluating the compensatory enhancement of glycolysis under the embryonic stress state based on the quantitative data of embryonic oxygen content lack, convert the quantitative data of oxygen content lack into a driving factor for metabolic pathway adjustment. According to the relationship between the glycolysis flow rate response of embryonic tissues under hypoxia in the literature, set a proportional adjustment coefficient of 0.015 millimoles per gram per minute for the upregulation of the lactate pathway flow rate induced by every millimole per gram of oxygen deficiency. Multiply this ratio by the original basal rate of glycolysis (obtained from the metabolic flow experiment of embryonic tissues under normal temperature and pressure, with the value taken as approximately 0.08 millimoles per gram per minute) to obtain the compensatory enhancement rate of glycolysis of the embryonic tissue per unit mass, and conduct a weighted average of the overall embryonic spatial region. The weight is based on the dual-fusion tensor of the temperature enhancement gradient field and the stomatal distribution density to ensure that the compensatory response distribution has dual dependence on thermal and oxygen factors. Finally, output the compensatory enhancement data of glycolysis, with the unit of millimoles per gram per minute, for subsequent evaluation of the occurrence probability and potential level of hypoxic metabolic disorders. The whole processing logic is constructed based on three basic operation methods: proportional adjustment, time integration, and weighted average, without the need to introduce a fitting model, and has clear physiological basis and mathematical logic support.

[0042] In the implementation process of analyzing the incremental rate ratio of lactic acid accumulation for the data of compensatory enhancement of glycolysis, first extract the data of compensatory enhancement of glycolysis obtained in step S245, with the data unit of millimoles per gram per minute, and combine it with the basic value of normal glycolysis rate (obtained through a micro-metabolic flux measurement experiment under a constant oxygen supply state, and the typical value is 0.08 millimoles per gram per minute). Subtract the basic value from the compensatory enhancement rate to obtain the net incremental rate of lactic acid production, and the unit is still millimoles per gram per minute. Subsequently, through the measured cumulative value of lactic acid concentration at a constant temperature, derive the actual rate of lactic acid accumulation from the lactic acid concentrations at multiple time points, and calculate the ratio of this rate to the above net incremental rate to obtain the incremental rate ratio of lactic acid accumulation. This ratio is used to measure whether the metabolic removal ability of lactic acid increases synchronously with its production rate. If the ratio is greater than 1, it means that the accumulation is faster than the metabolism, otherwise the accumulation is effectively cleared. The ratio calculation adopts a per-minute window smoothing process, and performs a weighted moving average on all rate data within a 10-minute sliding period. The weights are set according to the integral value of the temperature distribution gradient tensor of the embryonic tissue and the glycolytic activity response curve to ensure that the ratio has tissue thermal response consistency. In the implementation process of estimating the probability of embryonic myocardial contractility loss based on the incremental rate ratio of lactic acid accumulation, first establish a mapping between the accumulation rate ratio and the intracellular lactic acid accumulation concentration in embryonic myocardial cells according to the experimental data of myocardial lactic acid sensitivity. Among them, for every 1 millimole per liter increase in lactic acid concentration, the regulation rate of intracellular calcium ion concentration in the myocardium will decrease by 2.1%, and the decrease in calcium ion transport rate will affect the amplitude of calcium transient and the myocardial contraction frequency. Through the above mapping logic, perform piecewise normalization processing on the incremental rate ratio of lactic acid accumulation in the range of [1.0, 3.5], and set a linear threshold function. When the ratio is greater than 2.0, the probability of intracellular calcium homeostasis imbalance in the myocardium increases exponentially. Refer to the experimental data of myocardial contractility in the embryonic stage to construct a probability conversion table of lactic acid concentration - calcium response - contractility decline. Based on this table, interpolate according to the ratio data to obtain the corresponding probability of myocardial contractility loss, with the unit of percentage, reflecting the statistical possibility of abnormal contraction function in the embryonic myocardial cell population under a given lactic acid accumulation rate. The probability distribution values are finally allocated according to the spatial thermal blocks and embedded in the myocardial position map tensor to construct a two-dimensional probability matrix for subsequent simulation processing. In the implementation process of simulating and predicting the degree of embryonic hypoxic metabolic disorder based on the incremental rate ratio of lactic acid accumulation and the probability of embryonic myocardial contractility loss, first embed the data of the two types of indicators into a unified response space by constructing a joint response function. Among them, map the incremental rate ratio of lactic acid accumulation to the metabolic side pressure factor, and map the probability of myocardial contractility loss to the oxygen transport attenuation factor of the circulatory system. The weighted product of the two is used as the weight factor of the metabolic disorder composite index, and the value range is limited to 0 to 1.Subsequently, sampling is performed by time axis partitioning. For each interval, the weighted average of the metabolic side pressure factor and the circulating oxygen transport factor in that interval is taken, and weight distribution is carried out with the thermal field response weight and the embryo tissue density distribution weight as coefficients respectively. Then, the composite indexes of all intervals are accumulated to construct the integral value of the degree of hypoxic metabolic disorder. Finally, the prediction data of the degree of hypoxic metabolic disorder is expressed in the form of dimensionless entropy difference units, which is used to describe the trend intensity of multi-system metabolic coordination disorder of the whole embryo in a hypoxic environment. No fitting or deep network structure is used in the whole processing process. Only a simulation system is constructed through linear interpolation, weighted integration, and probability mapping to ensure that the prediction logic is consistent and effective within the physiological response and mathematical framework.

[0043] Step S3 includes the following steps: Step S31: Normalize the prediction data of the degree of embryonic developmental abnormality to obtain the normalized data of the degree of developmental abnormality; Step S32: Based on the normalized data of the degree of embryonic developmental abnormality, perform multi-faceted matching of the interval of the phased temperature demand on the outer surface of the eggshell for the characteristic data of the temporal fluctuation of the surface temperature to obtain the interval of the phased temperature demand; Step S33: Iteratively simulate and optimize the interval of the phased temperature demand to obtain the optimized interval of the phased temperature matching.

[0044] As an example of the present invention, refer to Figure 3 As shown, in this example, step S3 includes: Step S31: Normalize the prediction data of the degree of embryonic developmental abnormality to obtain the normalized data of the degree of developmental abnormality; In the embodiment of the present invention, first, the prediction data of the degree of embryonic developmental abnormality formed in the previous step is extracted. The data dimension is a three-dimensional tensor structure, corresponding to the time series dimension, the embryo internal space point cloud distribution dimension, and the developmental abnormality entropy weight dimension respectively. The minimum-maximum linear normalization method is used to project the abnormality degree data at each time point to between 0 and 1. To ensure the dynamic structure stability between the normalized data, the normalization interval is determined by the global minimum and maximum values in the full-cycle development stage, and the local sliding window minimum and maximum values are not used. During the normalization process, all measurement noise critical value samples are removed. The specific removal standard is the time point data with a measurement error exceeding ±0.003 entropy units, and the three-point median filtering noise reduction process is implemented on the sequence after removal. Finally, a normalized data tensor of the degree of developmental abnormality in a unified format is constructed. The normalized data at each time point is remapped into a unit vector with a length of 128, and the position weights inside the vector remain unchanged according to the abnormal space distribution structure, which is used for multi-channel comparison and matching during subsequent temperature demand deduction.

[0045] Step S32: Based on the normalized data of the degree of developmental abnormality, perform multi-faceted matching of the surface temperature temporal fluctuation characteristic data with the phased temperature demand intervals on the outer surface of the eggshell to obtain phased temperature demand intervals; In the embodiment of the present invention, the normalized data of the degree of developmental abnormality obtained in step S31 is used as the matching driving parameter, and the surface temperature temporal fluctuation characteristic data recorded in each time period is sequentially matched. The acquisition method of the fluctuation characteristic data is the temperature mean sequence extracted after pixel averaging of the eggshell surface temperature images collected by the thermal imager at a sampling frequency of 0.5 seconds, and the unit is degrees Celsius. To achieve multi-faceted matching, first divide the eggshell surface into 36×18 grid regions in the longitude and latitude directions respectively. Each grid records its own temperature fluctuation function, and this function is statistically divided into a window of 4 hours per time period. Four types of statistical features, namely the maximum, minimum, average, and standard deviation, are extracted within each window to construct a local fluctuation feature vector. By comparing the normalized abnormal data with the fluctuation characteristics of each grid region through cosine similarity, the characteristic time period with the maximum response to the abnormal data is extracted, and the temperature fluctuation range within this characteristic time period is used as the phased temperature demand interval required by this grid. Each demand interval is represented by the upper and lower bounds of Celsius temperature, and an example value is [36.1, 37.3] degrees Celsius. After the matching is completed, the phased temperature demand intervals of all grids are stored in a spherical grid tensor structure for reference in subsequent optimization steps.

[0046] Step S33: Iteratively simulate and optimize the phased temperature demand intervals to obtain phased temperature matching optimization intervals.

[0047] In the embodiments of the present invention, the obtained phased temperature requirement interval in step S32 is iteratively simulated and optimized. The optimization objective is to minimize the spatial temperature gradient on the premise of ensuring that the temperature intervals in each local area meet the consistency of the embryo normalization anomaly response threshold. The constraint condition optimization algorithm is adopted. Under the initial setting conditions, a multivariable function based on the Lagrange multiplier method is introduced to construct the optimization objective function. The objective function is set to minimize the second-order difference between the upper and lower temperature bounds of all grids, that is, the sum of the squares of the interval boundary differences between adjacent grids is used as the cost function. In the iterative process, the optimization path is solved in a gradient descent manner. The step size of each iteration is controlled at 0.01 degrees Celsius, and the maximum number of iterations is set to 500 times. The termination condition is that the boundary temperature difference between all grids is less than 0.2 degrees Celsius and the overall cost function decrease rate is lower than 0.001 square degrees Celsius. In each round of the optimization iteration, a normalization anomaly response reverse mapping verification needs to be performed again according to the optimized temperature interval. If the matching response is lost, one step is rolled back to reset the penalty weight factor and restart the optimization process. The finally formed phased temperature matching optimization interval retains three decimal places and is organized into a two-dimensional interval matrix data according to the spherical grid coding for output, which is directly used for the local thermal regulation interval instruction input of the incubator micro temperature control system. Throughout the process, no probability model or fitting strategy is used, and the interval adjustment is only completed through function construction and gradient iteration.

[0048] Step S4 includes the following steps: Step S41: Perform spatial temperature and humidity interactive phased balance adjustment on the real-time spatial temperature and humidity data according to the phased temperature matching optimization interval to obtain spatial temperature and humidity phased balance adjustment data; Step S42: Perform logical learning on the spatial temperature and humidity phased balance adjustment data to obtain spatial temperature and humidity balance adjustment learning data; Step S43: Based on the random forest algorithm, construct an incubator heating / spatial temperature and humidity control model for the phased temperature matching optimization interval and the spatial temperature and humidity balance adjustment learning data to obtain a heating / spatial temperature and humidity control model; send the heating / spatial temperature and humidity control model to the terminal to execute the incubator environment adjustment method.

[0049] In the embodiments of the present invention, first, the phased temperature matching optimization interval data generated in the previous step S33 is called as the boundary input constraint for space temperature and humidity regulation. Through a three-dimensional temperature and humidity acquisition network internally deployed in the incubator space, real-time temperature and humidity sampling are performed at each deployment point every 3 seconds. The temperature measurement accuracy is controlled within ±0.05 degrees Celsius, and the humidity measurement accuracy is controlled within ±1.2% relative humidity. The obtained real-time space temperature and humidity data are mapped into the spherical coordinate system of the hatching chamber to form a three-dimensional grid tensor. To achieve the interactive balance regulation of space temperature and humidity, the temperature and humidity state values of each sampling point are respectively compared with the target values of the current period in the phased temperature matching optimization interval. A local regulation function is calculated based on the temperature deviation value and the humidity gradient direction. The regulation function adopts a linear superposition structure. Among them, the temperature regulation component adjusts the on-off time ratio of the temperature control unit according to the unit temperature difference, and the humidity regulation component is controlled by adjusting the operation cycle ratio of the micro evaporator or the dehumidification fan. The above regulation actions are synchronously implemented in units of grids, and the regulation period is 15 seconds. To achieve the balance control of the space interaction effect, after each regulation cycle ends, the latest temperature and humidity state values of the grid and its six neighboring areas are collected, and difference analysis is performed to dynamically adjust the regulation function gain coefficient to ensure that there is no local heat accumulation or humidity stagnation in the space regulation. The temperature and humidity regulation parameter values executed by each grid point in each regulation cycle are recorded and summarized as phased regulation data, constituting a phased balance regulation data tensor of space temperature and humidity, and its size is expanded in real time along the time dimension. Based on the phased balance regulation data tensor of space temperature and humidity formed in step S41, logical learning processing based on a time window is performed. First, the space temperature and humidity regulation data of each time period are classified according to regional block division. The regional block division is based on the spatial thermodynamic flow field structure, and the entire incubator space is divided into 9 temperature flow dominant units. The regulation data is organized in a spatial grid distribution of 5×5×5 inside each unit. Taking 10 minutes as a time window, the average regulation parameter value, regulation direction volatility, regulation frequency density, and temperature and humidity response delay time within each unit are respectively statistically analyzed to construct a multi-dimensional logical state vector. Through sequence analysis methods, the transition logical relationship between state vectors in consecutive time windows is established, and the space temperature and humidity change response paths under different regulation strategies are extracted in the form of Boolean logic. The response paths form a learning sample set to characterize the logical coupling characteristics between space regulation behaviors and temperature and humidity evolution trends. To ensure that the logical learning has global representativeness, a set of training segments containing multiple phased temperature and humidity anomalies is introduced in each learning process, covering different development stages from the 4th day to the 18th day of the embryo, and controlling the environmental parameter consistency of the hatching chamber load density at about 400 eggs per cubic meter. The obtained logical state transition data is uniformly integrated into a space temperature and humidity balance regulation learning data tensor.The stage temperature matching optimization interval data obtained in the previous step S33 and the space temperature and humidity balance adjustment learning data generated in step S42 are jointly used as the input feature group, and the incubator heating / space temperature and humidity control model is constructed by building a multi-dimensional random forest algorithm. The feature input dimensions include regional space identification, time period coding, temperature target upper and lower boundary values, current humidity change trend, historical adjustment path hash coding, and logical status tags. During the construction of the random forest, the forest scale is set to 150 decision trees, the maximum depth of a single tree is 12, the minimum number of samples in a leaf node is 16, the feature division criterion uses the information gain ratio calculation method, and the Bootstrap resampling mechanism is enabled to enhance sample diversity. During the model training process, the full-cycle adjustment sample data tensor constructed in steps S41 and S42 is used as the training set, and the training and validation data are divided in a 7:3 ratio. The training cycle is set to 200 rounds, the tree structure is reconstructed once per round, and the sample feature importance ranking is synchronously executed. The control output includes three types of decision commands: the first type is the on-off cycle time series of the heaters in each area of the hatching chamber temperature control module; the second type is the working intensity and time series of the water vapor adjustment device in the humidity control module; the third type is the intermittent opening rhythm of the ventilation fan and the wind direction feedback adjustment factor. After the model is constructed, the heating / space temperature and humidity control model is loaded into the embedded incubator environment control terminal controller in binary coding. The control terminal re-executes the model inference every 30 seconds based on the latest environmental data and outputs a new round of control parameters to continuously perform the incubator environment adjustment task in a closed-loop manner.

[0050] The present invention also provides an incubator environment adjustment system for performing the incubator environment adjustment method as described above. The incubator environment adjustment system includes: A temperature time series fluctuation monitoring module for deploying electronic monitoring devices, temperature and humidity sensors, and infrared temperature sensors in multiple directions inside the incubator, collecting omnidirectional morphological images of the hatching eggs through the electronic monitoring devices to obtain omnidirectional morphological images of the hatching eggs; and monitoring the surface heating temperature state of the hatching eggs in multiple directions through the infrared temperature sensors based on the omnidirectional morphological images of the hatching eggs to obtain surface temperature time series fluctuation characteristic data. An abnormal embryo development degree prediction module for collecting real-time space temperature and humidity data inside the incubator through the temperature and humidity sensors to obtain real-time space temperature and humidity data; identifying the temperature concentration in the heat conduction area of the surface temperature time series fluctuation characteristic data based on the real-time space temperature and humidity data to obtain heat conduction area temperature concentration data; and simulating and predicting the abnormal degree of embryo development inside the eggs based on the heat conduction area temperature concentration data to obtain abnormal embryo development degree prediction data. The stage temperature requirement matching module for the outer surface of the eggshell is used to perform multi-faceted matching of the stage temperature requirement intervals on the outer surface of the eggshell for the surface temperature time-series fluctuation characteristic data based on the prediction data of the abnormal degree of embryo development, so as to obtain an optimized stage temperature matching interval; The heating / space temperature and humidity control model construction module is used to perform space temperature and humidity interactive stage balance adjustment on the real-time data of space temperature and humidity according to the optimized stage temperature matching interval, so as to obtain space temperature and humidity stage balance adjustment data; construct an incubator heating / space temperature and humidity control model based on the random forest algorithm for the space temperature and humidity balance adjustment learning data, so as to obtain a heating / space temperature and humidity control model; send the heating / space temperature and humidity control model to the terminal to execute the incubator environment adjustment method.

[0051] The above are only specific embodiments of the present invention, enabling those skilled in the art to understand or implement the present invention. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A method for adjusting the incubator environment, characterized in that: The following steps are involved: Step S1: electronic monitoring equipment, temperature and humidity sensors and infrared temperature sensors are deployed in multiple positions inside the incubator, and all-round morphological images of the hatching eggs are collected by the electronic monitoring equipment to obtain all-round morphological images of the hatching eggs; based on the all-round morphological images of the hatching eggs, the heating temperature state of the outer surface of the hatching eggs is monitored in multiple positions by infrared temperature sensors to obtain surface temperature time series fluctuation characteristic data; Step S2: collecting real-time data of the temperature and humidity in the incubator space through a temperature and humidity sensor to obtain real-time data of the temperature and humidity in the space; Based on the real-time data of space temperature and humidity, the surface temperature time series fluctuation characteristic data is used to identify the concentrated temperature of the heat conduction area, and the concentrated temperature data of the heat conduction area is obtained; According to the concentrated temperature data of the heat conduction area, the abnormal degree of embryonic development in the egg is simulated and predicted to obtain the predicted data of the abnormal degree of embryonic development; Step S3: Based on the embryonic development abnormality prediction data, the surface temperature time series fluctuation characteristic data are matched with the stage temperature requirement interval of the outer surface of the eggshell in multiple directions to obtain the stage temperature matching optimization interval; Step S4: According to the stage-by-stage temperature matching optimization interval, the spatial temperature and humidity real-time data are interactively and stage-by-stage balanced adjustment is performed on the spatial temperature and humidity to obtain the spatial temperature and humidity stage-by-stage balance adjustment data; based on the random forest algorithm, an incubator heating / space temperature and humidity control model is constructed for the spatial temperature and humidity balance adjustment learning data to obtain the heating / space temperature and humidity control model; the heating / space temperature and humidity control model is sent to the terminal to execute the incubator environment adjustment method.

2. The incubator environment adjustment method according to claim 1, characterized in that: Step S1 includes the following steps: Step S11: electronic monitoring equipment, temperature and humidity sensors and infrared temperature sensors are deployed in multiple positions inside the incubator, and all-round morphological images of the hatching eggs are collected by the electronic monitoring equipment to obtain all-round morphological images of the hatching eggs; Step S12: Based on the omnidirectional morphological image of the hatching egg, the outer surface heating temperature state of the hatching egg is monitored in all directions by an infrared temperature sensor to obtain the outer surface temperature state monitoring data of the egg shell; Step S13: performing time series fluctuation analysis on the eggshell outer surface temperature state monitoring data to obtain surface temperature time series fluctuation data; Step S14: performing characteristic analysis on the surface temperature time series fluctuation data to obtain surface temperature time series fluctuation characteristic data.

3. The incubator environment adjustment method according to claim 1, characterized in that: Step S2 includes the following steps: Step S21: collecting real-time data of the temperature and humidity in the incubator space through a temperature and humidity sensor to obtain real-time data of the temperature and humidity in the space; Step S22: performing concentrated identification of the temperature in the heat conduction area based on the real-time data of space temperature and humidity and the omnidirectional morphological image of the hatching eggs on the surface temperature time series fluctuation characteristic data to obtain concentrated temperature data in the heat conduction area; Step S23: numerically quantifying the distribution enhancement gradient of the concentrated temperature data in the heat conduction area to obtain temperature distribution enhancement numerical gradient data; Step S24: simulate and predict the degree of abnormal development of the embryo in the egg based on the temperature distribution enhanced numerical gradient data and the omnidirectional morphological image of the hatching egg to obtain the predicted data of the degree of abnormal development of the embryo.

4. The incubator environment adjustment method according to claim 3, characterized in that: Step S22 includes the following steps: Step S221: performing hatching egg curvature structure calculation on the hatching egg omnidirectional morphology image to obtain hatching egg curvature structure data; performing terminal winding angle calculation on the hatching egg omnidirectional morphology image to obtain the terminal winding angle of the hatching egg; Step S222: identifying texture rough surface distribution difference of the hatching egg omnidirectional morphological image based on the hatching egg curvature structure data and the winding angle of the hatching egg end, and obtaining texture rough surface distribution difference data; Step S223: performing axial ratio analysis according to the curvature structure data of the hatching egg and the winding angle of the end of the hatching egg to obtain axial ratio data of the hatching egg; Step S224: extracting the humidity increment interval of the space temperature and humidity real-time data to obtain the space humidity increment interval; performing heat capacity increase interval analysis on the space temperature and humidity real-time data based on the space humidity increment interval to obtain the heat capacity increase interval; Step S225: evaluating the effective conversion increment of medium heat conduction according to the heat capacity increase interval to obtain medium heat conduction increment data; Step S226: performing heat conduction regional diffusion fluctuation difference clustering on the surface temperature time series fluctuation characteristic data according to the texture rough surface distribution difference data, the hatching egg axial ratio data and the medium heat conduction increment data, to obtain heat conduction diffusion fluctuation difference clustering data; Step S227: performing heat conduction area temperature concentration identification on the heat conduction diffusion fluctuation difference clustering data to obtain heat conduction area temperature concentration data.

5. The incubator environment adjustment method according to claim 3, characterized in that: Step S24 includes the following steps: Step S241: performing embryonic tissue protein denaturation entropy change analysis based on temperature distribution enhanced numerical gradient data to obtain embryonic protein denaturation entropy change data; Step S242: simulating the embryo thermal damage cascade amplification effect on the embryo protein denaturation entropy change data to obtain the embryo thermal damage cascade amplification effect; Step S243: estimating the exponential growth of the oxygen consumption demand of the embryo according to the temperature distribution enhanced numerical gradient data, and obtaining exponential growth data of the oxygen consumption demand; Step S244: performing eggshell pore density recognition on the hatching egg omnidirectional morphological image to obtain eggshell pore density data; Step S245: simulating and predicting the degree of hypoxia and metabolic disorder of the embryo based on the eggshell pore density data and the exponential growth data of oxygen consumption demand, to obtain hypoxia and metabolic disorder degree prediction data; Step S246: Based on the embryonic thermal damage cascade amplification effect and the hypoxia metabolic disorder degree prediction data, a simulation prediction of the degree of embryonic developmental abnormality in the egg is performed to obtain embryonic developmental abnormality degree prediction data.

6. The incubator environment adjustment method according to claim 4, characterized in that: Step S242 includes the following steps: The entropy flow density matrix is ​​reconstructed on the entropy change data of embryonic protein denaturation to obtain the entropy flow spatial distribution tensor; According to the entropy flow spatial distribution tensor, the entropy change data of embryonic protein denaturation is analyzed for enzyme inactivation equivalent changes, and the enzyme inactivation equivalent change data is obtained; The enzyme inactivation isotropic change data were subjected to multiphase dynamic decay rate deduction to obtain the enzyme inactivation multiphase decay rate; Based on the multi-phase decay rate of enzyme inactivation, the cell division arrest time series numerical integration is performed to obtain the cell division arrest time series numerical value; The cascade amplification effect of embryonic thermal damage was simulated based on the entropy change data of embryonic protein denaturation according to the multiphase decay rate of enzyme inactivation and the timing values ​​of cell division arrest, and the cascade amplification effect of embryonic thermal damage was obtained.

7. The incubator environment adjustment method according to claim 6, characterized in that: Step S245 includes the following steps: The eggshell gas exchange capacity per unit time is calculated and estimated based on the eggshell pore density data to obtain the estimated eggshell gas exchange capacity per unit time; The quantitative assessment of embryonic oxygen deficiency was performed based on the exponential growth data of oxygen demand based on the estimated eggshell gas exchange capacity, and the quantitative data of embryonic oxygen deficiency were obtained; Based on the lack of quantitative data on embryonic oxygen content, the compensatory enhancement of glycolysis under embryonic stress was evaluated to obtain compensatory enhancement data of glycolysis; The lactate accumulation increment rate ratio was analyzed on the compensatory enhancement data of glycolysis to generate the lactate accumulation increment rate ratio; The probability of loss of myocardial contractility of the embryo is estimated according to the incremental rate ratio of lactate accumulation, and the probability of loss of myocardial contractility of the embryo is obtained; Based on the lactate accumulation incremental rate ratio and the probability of embryonic myocardial contractility loss, the degree of embryonic hypoxia and metabolic disorder was simulated and predicted to obtain the predicted data of the degree of hypoxia and metabolic disorder.

8. The incubator environment adjustment method according to claim 1, characterized in that: Step S3 includes the following steps: Step S31: normalizing the embryonic development abnormality degree prediction data to obtain normalized development abnormality degree data; Step S32: performing multi-directional matching of the staged temperature requirement interval of the outer surface of the eggshell based on the normalized data of the degree of developmental abnormality with the surface temperature time series fluctuation characteristic data to obtain the staged temperature requirement interval; Step S33: performing iterative simulation optimization on the staged temperature requirement interval to obtain the staged temperature matching optimization interval.

9. The incubator environment adjustment method according to claim 1, characterized in that: Step S4 includes the following steps: Step S41: performing interactive phased balance adjustment of space temperature and humidity on the real-time data of space temperature and humidity according to the phased temperature matching optimization interval to obtain phased balance adjustment data of space temperature and humidity; Step S42: performing logic learning on the spatial temperature and humidity stage-by-stage balance adjustment data to obtain spatial temperature and humidity balance adjustment learning data; Step S43: Based on the random forest algorithm, the incubator heating / space temperature and humidity control model is constructed for the stage temperature matching optimization interval and the space temperature and humidity balance adjustment learning data to obtain the heating / space temperature and humidity control model; the heating / space temperature and humidity control model is sent to the terminal to execute the incubator environment adjustment method.

10. An incubator environment adjustment system, characterized in that: Used to perform the incubator environment adjustment method according to claim 1, the incubator environment adjustment system comprises: The temperature time series fluctuation monitoring module is used to deploy electronic monitoring equipment, temperature and humidity sensors and infrared temperature sensors in multiple positions inside the incubator, collect all-round morphological images of the hatching eggs through the electronic monitoring equipment, and obtain all-round morphological images of the hatching eggs; based on the all-round morphological images of the hatching eggs, the infrared temperature sensor is used to monitor the heating temperature state of the outer surface of the hatching eggs in multiple positions, and obtain the surface temperature time series fluctuation characteristic data; The module for predicting the degree of abnormal embryo development is used to collect real-time data of temperature and humidity in the incubator space through the temperature and humidity sensor to obtain real-time data of temperature and humidity in the space; to identify the temperature concentration of the heat conduction area based on the time series fluctuation characteristic data of the surface temperature according to the real-time data of temperature and humidity in the space to obtain the concentrated temperature data of the heat conduction area; to simulate and predict the degree of abnormal embryo development in the egg based on the concentrated temperature data of the heat conduction area to obtain the predicted data of the degree of abnormal embryo development; The module for matching the stage temperature requirement of the outer surface of the eggshell is used to match the stage temperature requirement interval of the outer surface of the eggshell in all directions based on the prediction data of the abnormal degree of embryonic development with the surface temperature time series fluctuation characteristic data, so as to obtain the stage temperature matching optimization interval; The heating / space temperature and humidity control model construction module is used to perform interactive phased balance adjustment of space temperature and humidity on the real-time data of space temperature and humidity according to the phased temperature matching optimization interval, and obtain the phased balance adjustment data of space temperature and humidity; based on the random forest algorithm, the incubator heating / space temperature and humidity control model is constructed for the space temperature and humidity balance adjustment learning data to obtain the heating / space temperature and humidity control model; the heating / space temperature and humidity control model is sent to the terminal to execute the incubator environment adjustment method.

Images