Intelligent incubation bin anti-interference control method and system based on error self-learning

Abstract

The invention provides an intelligent incubation bin anti-interference control method and system based on error self-learning, and relates to the technical field of intelligent control, and the method comprises the steps: respectively calculating a temperature deviation value and a humidity deviation value according to an optimization parameter sequence of a temperature prediction error and an optimization parameter sequence of a humidity prediction error; performing normalized weighted summation on the temperature deviation value and the humidity deviation value, and performing dynamic scaling through a Gaussian kernel function to obtain a correction factor; dynamically adjusting the activation function slope of a neural network hidden layer according to the correction factor and a BP neural network, reconstructing the weight of the fuzzy rule base based on the numerical distribution characteristics of the correction factor, and generating the corrected weight of the fuzzy rule base; and based on the corrected fuzzy rule base weight, constructing a three-dimensional parameter adjustment curved surface, and dynamically adjusting proportion, integral and differential parameters of a PID controller through a curved surface gradient search algorithm to generate a control signal. According to the invention, the control accuracy is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent control, and particularly to an anti-interference control method and system for an intelligent incubation chamber based on error self-learning. Background Art

[0002] The current control technology of intelligent incubation chambers mainly relies on traditional PID control algorithms, but some defects are exposed in complex interference scenarios, which are specifically manifested as follows: Traditional PID control relies on fixed parameter adjustment and is difficult to adapt to environmental disturbances (such as sudden changes in temperature and humidity caused by ventilation and changes in heat conduction efficiency caused by equipment aging) and time-varying characteristics during the incubation process. Its core defects include: Insufficient ability to resist time-varying interference, unable to dynamically track the non-steady-state changes of temperature and humidity errors. When periodic interference (such as daily environmental temperature fluctuations) or sudden interference (such as equipment failures) occur in the incubation chamber, the control quantity is prone to overshoot or adjustment lag, resulting in the incubation environment deviating from the ideal range for a long time.

[0003] The temperature and humidity parameters have strong coupling (such as heat loss during the humidification process). Traditional single-loop PID control adjusts the two independently, ignoring the cross influence, resulting in the control strategy being unable to take both into account. For example, humidity adjustment may cause temperature fluctuations, and the PID controller cannot quantify this coupling effect and perform collaborative compensation; the proportional, integral, and differential parameters need to be manually debugged, making it difficult to meet the dynamic control requirements of different incubation stages (such as the differences in temperature and humidity sensitivity during the initial incubation stage and the hatching stage), and lacking an adaptive adjustment mechanism. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide an anti-interference control method for an intelligent incubation chamber based on error self-learning, which improves the accuracy of control.

[0005] To solve the above technical problems, the technical solution of the present invention is as follows: In a first aspect, an anti-interference control method for an intelligent incubation chamber based on error self-learning, the method includes: Step 1, real-time collect the temperature and humidity data streams of chicken embryo incubation in the incubation chamber; form a sliding detection window with the data of three consecutive sampling periods, and perform fusion processing on the temperature and humidity data within the window to output an optimized parameter sequence of temperature prediction error and an optimized parameter sequence of humidity prediction error; Step 2, calculate the temperature deviation value and the humidity deviation value respectively according to the optimized parameter sequence of temperature prediction error and the optimized parameter sequence of humidity prediction error; Step 3, perform normalized weighted summation on the temperature deviation value and the humidity deviation value, and perform dynamic scaling through a Gaussian kernel function to obtain a correction factor; Step 4: Dynamically adjust the activation function slope of the hidden layer of the neural network according to the correction factor and the BP neural network, and reconstruct the weights of the fuzzy rule base based on the numerical distribution characteristics of the correction factor to generate the corrected weights of the fuzzy rule base. Step 5: Based on the corrected weights of the fuzzy rule base, construct a three-dimensional parameter adjustment surface, and dynamically adjust the proportional, integral, and differential parameters of the PID controller through the surface gradient search algorithm to generate a control signal.

[0006] Furthermore, the data of three consecutive sampling periods are used to form a sliding detection window, and the temperature and humidity data are fused within the window to output an optimized parameter sequence of the temperature prediction error and an optimized parameter sequence of the humidity prediction error, including: Within the sliding detection window, the original temperature and humidity data at the current sampling moment and its previous two moments are respectively constructed into time series vectors, and each vector contains three consecutive data points. Based on the temperature observation values of the previous two periods, establish a state transition model to generate the temperature prediction value at the current moment; subtract the currently actually collected temperature data from the prediction value to obtain the temperature prediction error value; dynamically adjust the filtering gain according to the noise covariance matrix and the observation noise covariance matrix, and perform weighted correction on the prediction error to output the optimized parameter of the temperature prediction error at the current moment. Repeat the prediction-correction process for the humidity time series vector to generate the corresponding optimized parameter of the humidity prediction error. Arrange the optimized parameters of the temperature prediction error output in three consecutive sampling periods in chronological order to form an optimized parameter sequence of the temperature prediction error; form an optimized parameter sequence of the humidity prediction error with the optimized parameters of the humidity prediction error. Perform a moving average process on three consecutive parameter values in the temperature optimized parameter sequence to obtain the average temperature error; calculate the deviation square amount for each parameter value according to the average temperature error, and take the arithmetic average of the three deviation square amounts as the current element of the temperature error variance matrix; perform the same operation on the humidity optimized parameter sequence to generate the current element of the humidity error variance matrix.

[0007] Furthermore, calculate the temperature deviation value and the humidity deviation value respectively according to the optimized parameter sequence of the temperature prediction error and the optimized parameter sequence of the humidity prediction error, including: Align the temperature optimized parameter sequence and the humidity optimized parameter sequence point by point according to the sampling timestamp. From the aligned temperature optimized parameter sequence, calculate the change rate of the data before and after the current sampling point to obtain the dynamic gradient eigenvalue reflecting the dynamic change trend of the temperature error sequence; from the humidity optimized parameter sequence, calculate the absolute value of the difference between adjacent data to obtain the quantization fluctuation eigenvalue characterizing the instantaneous fluctuation degree of the humidity error. Generate an initial weight based on the ratio of the dynamic gradient eigenvalue to the quantization fluctuation eigenvalue, and correct the initial weight through the humidity historical deviation ratio to obtain a normalized weight coefficient; multiply the difference between the current element of the temperature error variance matrix and the target threshold by the normalized weight coefficient to obtain a basic deviation amount; process the basic deviation amount based on the instantaneous intensity value in the quantization fluctuation eigenvalue to obtain a temperature deviation value; Perform a natural logarithm transformation on the ratio of the current element of the humidity error variance matrix to the target threshold to generate an initial deviation amount; generate a compensation factor based on the product of the temperature cumulative amplitude value and the humidity instantaneous fluctuation intensity, and perform dynamic range compression on the initial deviation amount through the compensation factor to obtain a humidity deviation value.

[0008] Further, generate an initial weight based on the ratio of the dynamic gradient eigenvalue to the quantization fluctuation eigenvalue, and correct the initial weight through the humidity historical deviation ratio to obtain a normalized weight coefficient; multiply the difference between the current element of the temperature error variance matrix and the target threshold by the normalized weight coefficient to obtain a basic deviation amount; process the basic deviation amount based on the instantaneous intensity value in the quantization fluctuation eigenvalue to obtain a temperature deviation value, including: Calculate the ratio of the dynamic gradient eigenvalue to the quantization fluctuation eigenvalue to obtain a ratio reflecting the relative intensity of the temperature and humidity error changes as the initial weight; Calculate the ratio of the humidity error historical data deviating from the normal range, i.e., the humidity historical deviation ratio, and adjust the initial weight with the humidity historical deviation ratio to obtain the corrected initial weight; Normalize the corrected initial weight to obtain a normalized weight coefficient; Obtain the current element of the temperature error variance matrix, i.e., the average value of the temperature error deviation squared amount within the current sliding window, calculate the difference between the current element and the preset temperature error target threshold, and fuse the difference with the normalized weight coefficient to obtain a basic deviation amount reflecting the deviation degree of the temperature error fluctuation relative to the target state; Extract the instantaneous intensity value at the current moment from the quantization fluctuation eigenvalue, i.e., the fluctuation peak value of the current humidity error, and adjust the basic deviation amount with the instantaneous intensity value to obtain a temperature deviation value reflecting the temperature error and humidity interference.

[0009] Further, perform a natural logarithm transformation on the ratio of the current element of the humidity error variance matrix to the target threshold to generate an initial deviation amount; generate a compensation factor based on the product of the temperature cumulative amplitude value and the humidity instantaneous fluctuation intensity, and perform dynamic range compression on the initial deviation amount through the compensation factor to obtain a humidity deviation value, including: Obtain the current element of the humidity error variance matrix, that is, the average value of the squared deviation of the humidity error within the current sliding window, calculate the humidity ratio of the current element to the preset humidity error target threshold, and reflect the deviation multiple of the current humidity error fluctuation relative to the target state; Take the natural logarithm of the humidity ratio to compress the numerical range to the logarithmic scale to obtain the initial deviation amount reflecting the degree of humidity error deviation; Calculate the absolute value cumulative sum of the temperature error optimization parameter sequence within the current sliding window, characterize the overall deviation degree and duration of the temperature error, and form the temperature cumulative amplitude value; Extract the instantaneous fluctuation peak value at the current moment from the humidity optimization parameter sequence, that is, the absolute value of the maximum difference between adjacent sampling points, and denote it as the humidity instantaneous fluctuation intensity; Multiply the temperature cumulative amplitude value by the humidity instantaneous fluctuation intensity to obtain the compensation factor; Adjust the initial deviation amount through the compensation factor, and reduce the numerical range of the initial deviation amount through non - linear mapping to obtain the humidity deviation value that balances the coupling effect of temperature and humidity.

[0010] Further, perform normalized weighted summation on the temperature deviation value and the humidity deviation value, and perform dynamic scaling through the Gaussian kernel function to obtain the correction factor, including: Calculate the maximum value and the minimum value of the temperature deviation value in the preset sampling period respectively, calculate its first numerical deviation range, that is, the maximum value minus the minimum value; subtract the minimum value from the current temperature deviation value to obtain the first deviation difference, and calculate the deviation ratio between the first deviation difference and the first numerical deviation range to obtain the normalized temperature deviation value; Calculate the maximum value and the minimum value of the humidity deviation value in the preset sampling period respectively, calculate its second numerical deviation range, that is, the maximum value minus the minimum value; subtract the minimum value from the current humidity deviation value to obtain the second deviation difference, and calculate the deviation ratio between the second deviation difference and the second numerical deviation range to obtain the normalized humidity deviation value; Preset the temperature influence weight coefficient and the humidity influence weight coefficient, multiply the normalized temperature deviation value by the temperature influence weight coefficient to obtain the temperature contribution amount; multiply the normalized humidity deviation value by the humidity influence weight coefficient to obtain the humidity contribution amount; Add the temperature contribution amount and the humidity contribution amount to obtain the temperature - humidity deviation index value; substitute the temperature - humidity deviation index value into the Gaussian kernel function to calculate its function value, that is, the correction factor.

[0011] Further, according to the correction factor and the BP neural network, dynamically adjust the activation function slope of the hidden layer of the neural network, and reconstruct the weights of the fuzzy rule base based on the numerical distribution characteristics of the correction factor to generate the corrected fuzzy rule base weights, including: Determine the initial slope parameter corresponding to the activation function of the hidden layer of the BP neural network; Determine the adjustment amount of the activation function slope according to the value of the correction factor to obtain the adjusted value of the slope; Add the adjusted value of the slope to the initial slope parameter to obtain the adjusted activation function slope; Analyze the numerical distribution characteristics of the correction factor, calculate the numerical distribution characteristics, and the numerical distribution characteristics include the mean, variance, maximum value, and minimum value of the correction factor; Determine the initial weights of the fuzzy rule base, and adjust the weights of each rule in the fuzzy rule base according to the numerical distribution characteristics of the correction factor to obtain the corrected fuzzy rule base weights.

[0012] Furthermore, based on the corrected fuzzy rule base weights, construct a three-dimensional parameter adjustment surface, and dynamically adjust the proportional, integral, and differential parameters of the PID controller through the surface gradient search algorithm to generate a control signal, including: Determine the three variables of the three-dimensional parameter adjustment surface, which are the temperature deviation value, the humidity deviation value, and the corrected fuzzy rule base weights; Under different combinations of the temperature deviation value, the humidity deviation value, and the corrected fuzzy rule base weights, record the corresponding values of the proportional, integral, and differential parameters of the PID controller; According to the values of the proportional, integral, and differential parameters of the PID controller, construct a three-dimensional parameter adjustment surface through neural network fitting; According to the temperature deviation value, the humidity deviation value, and the corrected fuzzy rule base weights in the current incubation chamber, find the corresponding state point on the three-dimensional parameter adjustment surface, and the state point represents the initial parameter settings of the PID controller under the current working conditions; At the current state point, calculate the gradient of the three-dimensional parameter adjustment surface. The gradient represents the direction of change of the three-dimensional parameter adjustment surface at the state point and includes three components, which respectively correspond to the change directions of the proportional, integral, and differential parameters of the PID controller; According to the gradient of the three-dimensional parameter adjustment surface and the search step size, adjust the proportional, integral, and differential parameters of the PID controller along the opposite direction of the gradient until the search step size is reached to obtain the optimized proportional, integral, and differential parameters of the PID controller; The PID controller calculates the corresponding control signal according to the current temperature deviation value and humidity deviation value in combination with the adjusted proportional, integral, and differential parameters.

[0013] In the second aspect, an intelligent incubation chamber anti-interference control system based on error self-learning includes: The acquisition module is used to collect the temperature and humidity data streams of chicken embryo hatching in the hatching chamber in real time; form a sliding detection window with the data of three consecutive sampling periods, perform fusion processing on the temperature and humidity data within the window, and output the optimized parameter sequence of the temperature prediction error and the optimized parameter sequence of the humidity prediction error; The calculation module is used to calculate the temperature deviation value and the humidity deviation value respectively according to the optimized parameter sequence of the temperature prediction error and the optimized parameter sequence of the humidity prediction error; The correction module is used to perform normalized weighted summation on the temperature deviation value and the humidity deviation value, and perform dynamic scaling through a Gaussian kernel function to obtain a correction factor; The adjustment module is used to dynamically adjust the slope of the activation function of the neural network hidden layer according to the correction factor and the BP neural network, and reconstruct the weights of the fuzzy rule base based on the numerical distribution characteristics of the correction factor to generate the corrected fuzzy rule base weights; The processing module is used to construct a three-dimensional parameter adjustment surface based on the corrected fuzzy rule base weights, and dynamically adjust the proportional, integral, and differential parameters of the PID controller through a surface gradient search algorithm to generate a control signal.

[0014] The above solution of the present invention has at least the following beneficial effects: By forming a sliding detection window with the data of three consecutive sampling periods, the capture of the time series characteristics of the temperature and humidity data streams is realized, the random noise interference is effectively filtered, the reliability of the original data is improved, and the temperature / humidity data is fused and processed within the window to output the optimized parameter sequence of the temperature / humidity prediction error. Compared with the traditional single-period sampling method, the prediction error is reduced.

[0015] Based on the optimized parameter sequence, the temperature and humidity deviation values are calculated in real time, the deviation degree between the environmental parameters and the set threshold is accurately quantified, and a clear quantitative basis is provided for interference response; by normalizing, the dimension difference of temperature and humidity is eliminated, and combined with the non-linear mapping characteristics of the Gaussian kernel function, the correction factor is dynamically adjusted, and the response speed of the system to sudden interferences (such as equipment start-stop, environmental fluctuations) is improved, and the interference suppression ability is significantly enhanced.

[0016] By dynamically adjusting the slope of the activation function of the neural network hidden layer through the correction factor, the fitting ability of the BP network to complex non-linear error characteristics is optimized, the model convergence speed is improved, and the weights of the fuzzy rule base are reconstructed based on the numerical distribution characteristics of the correction factor, so that the fuzzy control rules can be adaptively adjusted according to the error characteristics, and the limitation of fixed traditional fuzzy control rules is solved.

[0017] Construct a three-dimensional adjustment surface for the parameters of the PID controller, and use the surface gradient search algorithm to optimize the proportional, integral, and differential parameters in real time. Compared with the traditional PID control, the temperature control fluctuation range is reduced to ±0.2°C, and the humidity control fluctuation range is reduced to ±3%RH. Description of the Drawings

[0018] Figure 1 is a schematic flow chart of an anti-interference control method for an intelligent incubation chamber based on error self-learning provided by an embodiment of the present invention.

[0019] Figure 2 is a schematic diagram of an anti-interference control system for an intelligent incubation chamber based on error self-learning provided by an embodiment of the present invention. Detailed Embodiments

[0020] Hereinafter, exemplary embodiments of the present disclosure will be described in more detail with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided so that the present disclosure can be more thoroughly understood and the scope of the present disclosure can be fully conveyed to those skilled in the art.

[0021] As Figure 1 shown, an embodiment of the present invention proposes an anti-interference control method for an intelligent incubation chamber based on error self-learning, and the method includes the following steps: Step 1, collect the temperature and humidity data streams of chicken embryo incubation in the incubation chamber in real time; form a sliding detection window with the data of three consecutive sampling periods, and perform fusion processing on the temperature and humidity data within the window to output an optimized parameter sequence of temperature prediction error and an optimized parameter sequence of humidity prediction error; Step 2, calculate the temperature deviation value and the humidity deviation value respectively according to the optimized parameter sequence of temperature prediction error and the optimized parameter sequence of humidity prediction error; Step 3, perform normalized weighted summation on the temperature deviation value and the humidity deviation value, and perform dynamic scaling through a Gaussian kernel function to obtain a correction factor; Step 4, dynamically adjust the activation function slope of the hidden layer of the neural network according to the correction factor and the BP neural network, and reconstruct the weights of the fuzzy rule base based on the numerical distribution characteristics of the correction factor to generate the corrected fuzzy rule base weights; Step 5, based on the corrected fuzzy rule base weights, construct a three-dimensional parameter adjustment surface, and dynamically adjust the proportional, integral, and differential parameters of the PID controller through a surface gradient search algorithm to generate a control signal.

[0022] In the embodiments of the present invention, a sliding detection window is formed by data of three consecutive sampling periods to capture the temporal characteristics of temperature and humidity data streams, effectively filtering random noise interference and enhancing the reliability of the original data. The temperature and humidity data are fused within the window, and an optimized parameter sequence of temperature / humidity prediction errors is output. Compared with the traditional single-period sampling method, the prediction error is reduced. Based on the optimized parameter sequence, the temperature and humidity deviation values are calculated in real time to accurately quantify the deviation degree between the environmental parameters and the set threshold, providing a clear quantitative basis for interference response. By normalizing to eliminate the dimension difference between temperature and humidity and combining with the non-linear mapping characteristics of the Gaussian kernel function, the correction factor is dynamically adjusted, improving the response speed of the system to sudden interferences (such as equipment start / stop and environmental fluctuations), and significantly enhancing the interference suppression ability. By dynamically adjusting the slope of the activation function of the hidden layer of the neural network with the correction factor, the fitting ability of the BP network to complex non-linear error characteristics is optimized, and the model convergence speed is increased. Based on the numerical distribution characteristics of the correction factor, the weights of the fuzzy rule base are reconstructed, enabling the fuzzy control rules to be adaptively adjusted according to the error characteristics, and solving the limitation of fixed traditional fuzzy control rules.

[0023] A three-dimensional adjustment surface of the PID controller parameters is constructed, and the proportional, integral, and differential parameters are optimized in real time using the surface gradient search algorithm. Compared with traditional PID control, the temperature control fluctuation range is reduced to ±0.2°C, and the humidity control fluctuation range is reduced to ±3%RH. A complete closed loop is formed from data acquisition, error prediction, correction factor generation to control parameter optimization. Through iterative learning with historical error data, the ability to predict complex interference scenarios is gradually improved. By integrating the advantages of multiple technologies such as data fusion algorithms, neural networks, fuzzy control, and gradient search, it not only has the data-driven learning ability but also retains the interpretability of rule reasoning, enabling the hatching chamber to maintain the optimal control state at different hatching stages (such as the early stage of embryo development and the hatching period), and improving the hatching success rate.

[0024] In another preferred embodiment of the present invention, data of three consecutive sampling periods are used to form a sliding detection window, and the temperature and humidity data are fused within the window, outputting an optimized parameter sequence of temperature prediction errors and an optimized parameter sequence of humidity prediction errors, including: Within the sliding detection window, the original temperature and humidity data at the current sampling moment and its previous two moments are respectively constructed into time series vectors, and each vector contains three consecutive data points, specifically including: converting discrete sampling data into vectors with temporal correlation to capture short-term dynamic characteristics, achieving: At the current sampling moment (denoted as t), data of the most recent three consecutive sampling periods (i.e., t-2, t-1, and t moments) are extracted, and time series vectors of temperature and humidity are respectively constructed. Each vector contains three consecutive data points. For example, the temperature vector is [T t-2 , T t-1 , Tt , and the humidity vector is [H t-2 , H t-1 , H t . In this way, the real-time collected data stream is converted into a three-dimensional data unit with time context.

[0025] Based on the temperature observations of the previous two cycles, a state transition model is established to generate the temperature prediction value at the current moment; the difference between the currently actually collected temperature data and the predicted value is obtained to get the temperature prediction error value; according to the noise covariance matrix and the observation noise covariance matrix, the filtering gain is dynamically adjusted, and the prediction error is weighted and corrected to output the optimized parameter of the temperature prediction error at the current moment, specifically including: Using the temperature observations (T t-2 and T t-1 ) of the previous two cycles to establish a state transition model (such as linear extrapolation or first-order dynamic model) to generate the temperature prediction value at the current moment , this model captures the temperature change trend based on the temporal correlation of historical data; the difference between the actually collected temperature value Tt and the predicted value is obtained to get the original temperature prediction error value ; according to the preset noise covariance matrix (reflecting the internal noise level of the system) and the observation noise covariance matrix (reflecting the sensor measurement error), the filtering gain coefficient is dynamically adjusted, and this coefficient is used to weight and correct the original error value to suppress high-frequency noise and output the optimized temperature prediction error parameter , this process improves the stability of the error characteristics by balancing the prediction error and the noise intensity.

[0026] When specifically applied, a first-order dynamic model is adopted. Let the temperature state vector include the current temperature and the temperature change rate , that is . The state transition equation can be expressed as: ; where the state transition matrix , is the sampling period; is the process noise vector, which follows a Gaussian distribution with a mean of zero, that is is the process noise covariance matrix, reflecting the internal noise level of the system.

[0027] The state vector and at the current moment are estimated from the temperature observations of the previous two cycles , and the temperature change rate , then . Then, the state prediction value at the current moment is: ; Therefore, the temperature prediction value at the current moment is: .

[0028] Repeat the prediction-correction process for the humidity time series vector to generate the corresponding humidity prediction error optimization parameters, specifically including: for the humidity time series vector Repeat the above temperature prediction and correction process: based on the humidity observation values at the previous two moments ( and ), establish a state transition model to predict the current humidity value ; calculate the error between the actual humidity value and the predicted value ; use the same filtering gain mechanism to weight and correct the humidity error to generate the humidity prediction error optimization parameter .

[0029] Arrange the temperature prediction error optimization parameters output in three consecutive sampling periods in chronological order to form a temperature prediction error optimization parameter sequence; form a humidity prediction error optimization parameter sequence from the humidity prediction error optimization parameters, specifically including: arrange the temperature prediction error optimization parameters output in three consecutive sampling periods (such as in chronological order to form a temperature prediction error optimization parameter sequence ; for humidity sequence construction, similarly, arrange the humidity optimization parameters for three periods as a humidity prediction error optimization parameter sequence .

[0030] Perform a moving average process on three consecutive parameter values in the temperature optimization parameter sequence to obtain the mean temperature error; calculate the deviation square amount for each parameter value based on the mean temperature error, and take the arithmetic average of the three deviation square amounts as the current element of the temperature error variance matrix; perform the same operation on the humidity optimization parameter sequence to generate the current element of the humidity error variance matrix, specifically including: Perform an arithmetic average on three consecutive parameter values in the temperature optimization parameter sequence to obtain the mean temperature error , and the formula is , reflecting the overall level of the error.

[0031] Calculation of the deviation square amount, calculate the square of the deviation from the mean for each parameter value, that is , measuring the dispersion degree of each data point from the mean.

[0032] Variance calculation: The arithmetic mean of three deviation squared quantities is calculated to obtain the current element of the temperature error variance matrix , and the formula is , which represents the fluctuation intensity of the error; For humidity error processing, the exact same processes of moving average, deviation square calculation, and variance calculation are performed on the humidity optimization parameter sequence to obtain the current element of the humidity error variance matrix .

[0033] In the embodiment of the present invention, by retaining the data of three consecutive periods, a minimized time series analysis unit is constructed to effectively extract the time dimension features of temperature and humidity changes (such as rising / falling trends, fluctuation amplitudes), and avoid the random interference of single-period data. The fixed window length (three periods) balances the data timeliness and the amount of calculation, avoiding both the lag problem caused by a long window and reducing the data storage and processing pressure. The state transition model deduces the theoretical value at the current moment through historical data, captures the temperature and humidity change trends in advance (such as the inertial delay during the heating / cooling process), enabling the system to pre-adjust the control parameters before the interference occurs, and realizing the composite control of "feedforward + feedback". By characterizing the statistical characteristics of system noise (such as variance, correlation) through the noise covariance matrix, and combining the observation noise covariance matrix to estimate the measurement error in real time, the optimal filtering of random noise (similar to the Kalman filtering principle) can be achieved, which can reduce the noise component in the original error signal and improve the signal-to-noise ratio. The filtering gain is dynamically adjusted according to the noise characteristics. When the environmental interference intensifies (such as an increase in the noise covariance), the weight of the observed value is automatically increased to enhance the dependence on real-time data; when the system tends to be stable, the weight of the model prediction value is increased to reduce the frequency of control actions, balancing the control accuracy and the service life of the actuator. Separating the model prediction error and the measurement error can avoid misjudging the sensor noise (such as the drift of the temperature and humidity meter) as the real environmental fluctuation, and reduce the probability of misoperation. The moving average processing (taking the mean of three consecutive parameter values) filters short-term high-frequency noise and extracts the low-frequency trend component of the error signal (such as the continuous rising / falling humidity trend), avoiding over-control triggered by accidental errors. Independently processing the temperature and humidity error sequences and retaining their coupling and independence (such as temperature fluctuations may be accompanied by humidity changes, but their interference sources are different), enabling the system to optimize the control parameters according to the different error characteristics of temperature and humidity respectively, and avoiding the problem of "neglecting one thing while attending to another" in single-variable control. By comparing the change time series of the temperature and humidity error variance matrices, it is possible to assist in judging the type of interference (such as a sudden increase in temperature variance may be due to a heating device failure, and a sudden increase in humidity variance may be due to an abnormal humidification device), providing data support for fault diagnosis and maintenance, and reducing the cost of manual inspection.

[0034] In another preferred embodiment of the present invention, according to the optimization parameter sequences of the temperature prediction error and the humidity prediction error, the temperature deviation value and the humidity deviation value are calculated respectively, including: Align the temperature optimization parameter sequence and the humidity optimization parameter sequence point by point according to the sampling timestamps. From the aligned temperature optimization parameter sequence, calculate the change rate of the data before and after the current sampling point to obtain the dynamic gradient eigenvalue reflecting the dynamic change trend of the temperature error sequence; from the humidity optimization parameter sequence, calculate the absolute value of the difference between adjacent data to obtain the quantization fluctuation eigenvalue characterizing the instantaneous fluctuation degree of the humidity error, specifically including: Temperature optimization parameter sequence And the humidity optimization parameter sequence Are both sampled at equal time intervals (the sampling period is ), and the timestamps are strictly synchronized (such as all taking As the sampling moment).

[0035] Implementation method: Directly align the parameter values at the same moment through the timestamp index to form an aligned two-dimensional sequence: ; Among them, Is the sliding window length (such as Corresponds to the data of the previous two moments).

[0036] Boundary processing: Initial stage (t < 2): When the current sampling point Is less than the window length (such as t = 0 or t = 1), use forward difference or backward difference to replace the two-way difference.

[0037] Outlier filtering: If the sensor data is abnormal at a certain moment (such as the temperature and humidity jump exceeding the threshold), then skip this point, and use linear interpolation of the data before and after to complete it before participating in the calculation.

[0038] By calculating the slope of the temperature optimization parameter sequence before and after the current point, quantify the rate of error change (rising / falling trend and speed), and use the central difference method or sliding window linear regression to extract the dynamic gradient. Specific implementation steps: Taking the current sampling point As the center, take the previous moment And the next moment Of the data (a total of 3 points) to form a local window .

[0039] Use the three-point data of adjacent moments to fit a straight line, and the slope reflects the trend near the current point (to avoid the interference of mutations at a single point).

[0040] Central difference method to calculate the gradient: Assume that the three-point coordinates are , then the central difference gradient is: ; Among them, the numerator represents the average of the backward difference and the forward difference (eliminating the unilateral deviation); the denominator represents the time interval (normalized to the rate of change per unit time).

[0041] Boundary processing (t = 0 or t = N - 1, where N is the sequence length): Starting point (t = 0): Use the backward difference ; End point (t = N - 1): Use the forward difference .

[0042] Calculation of the humidity quantization fluctuation eigenvalue (reflecting the instantaneous fluctuation), by calculating the absolute value of the difference between adjacent data in the humidity optimization parameter sequence, quantifying the instantaneous change amplitude of the error, and reflecting the suddenness of environmental interference (such as the sudden change in humidity caused by the start and stop of the humidification equipment).

[0043] The specific implementation steps are as follows: Calculation of adjacent differences, for the humidity optimization parameter sequence, calculate the absolute difference between the current point and the previous moment: ; The single-step difference reflects the degree of mutation between adjacent moments, and the absolute value eliminates the influence of positive and negative directions, focusing on the fluctuation amplitude.

[0044] Multi-step difference expansion (enhancing robustness): If it is necessary to reflect the short-term fluctuation trend, the absolute value mean of the continuous n-step differences can be calculated: ; Typical value: When n = 3, calculate the average of the absolute values of the differences between the current point and the previous 3 moments to suppress accidental noise.

[0045] Dynamic threshold filtering: Set the fluctuation amplitude threshold (such as determined according to the historical data standard deviation), when it is determined as a significant fluctuation, otherwise set it to zero: .

[0046] Dynamic threshold filtering is used to distinguish normal fluctuations from sudden interferences and avoid mis-triggering control actions.

[0047] Generate the initial weight according to the ratio of the dynamic gradient eigenvalue to the quantization fluctuation eigenvalue, correct the initial weight through the historical deviation ratio of humidity to obtain the normalized weight coefficient; multiply the difference between the current element of the temperature error variance matrix and the target threshold by the normalized weight coefficient to obtain the basic deviation amount; process the basic deviation amount based on the instantaneous intensity value in the quantization fluctuation eigenvalue to obtain the temperature deviation value; Perform a natural logarithm transformation on the ratio of the current element of the humidity error variance matrix to the target threshold to generate an initial deviation; generate a compensation factor based on the product of the temperature cumulative amplitude value and the humidity instantaneous fluctuation intensity, and perform dynamic range compression on the initial deviation through the compensation factor to obtain the humidity deviation value.

[0048] In the embodiments of the present invention, by calculating the change rate of the data before and after the current sampling point in the temperature optimization parameter sequence, the dynamic change trend of the temperature error sequence can be keenly captured, which enables the system to not only pay attention to the current temperature error situation but also predict the future trend of the temperature error and take corresponding control measures in advance, enhancing the forward-looking and adaptability of the system. For example, when it is detected that the dynamic gradient eigenvalue of the temperature error is positive and large, it indicates that the temperature error is increasing rapidly, and the system can timely adjust the power of the heating or cooling equipment to avoid further deviation of the temperature from the target value. Calculating the absolute value of the difference between adjacent data in the humidity optimization parameter sequence can accurately quantify the instantaneous fluctuation degree of the humidity error, which helps the system to timely detect sudden changes in humidity, monitor and evaluate the fluctuation situation of humidity in real time. For example, when the quantified fluctuation eigenvalue is large, it indicates that the humidity has fluctuated greatly in a short time, and the system can quickly take measures to stabilize the humidity. The humidity environment is often easily interfered by various factors, such as ventilation, start and stop of humidification equipment, etc. The quantified fluctuation eigenvalue can effectively reflect the influence degree of these interferences on the humidity error, enabling the system to timely adjust the control strategy and enhance the resistance to humidity interference.

[0049] Generate an initial weight through the ratio of the dynamic gradient eigenvalue to the quantified fluctuation eigenvalue, and correct it in combination with the historical deviation ratio of humidity. The obtained normalized weight coefficient comprehensively considers the dynamic changes and historical situations of temperature and humidity errors, which enables a more comprehensive consideration of the mutual relationship and change trends between temperature and humidity when calculating the temperature deviation value, improving the rationality and accuracy of the weight coefficient. The normalized weight coefficient will be adaptively adjusted according to the dynamic changes of temperature and humidity errors. When the dynamic change of temperature error is large, the weight will correspondingly bias towards the temperature factor; when the fluctuation of humidity error is obvious, the weight will appropriately consider the influence of the humidity factor. This adaptive adjustment mechanism enables the system to better adapt to different environmental conditions and error change situations.

[0050] Multiply the difference between the current element of the temperature error variance matrix and the target threshold by the normalized weight coefficient to obtain the basic deviation amount. This calculation method fully considers the magnitude and weight factors of the temperature error, and the basic deviation amount can more accurately reflect the deviation degree of the temperature error relative to the target threshold. Process the basic deviation amount based on the instantaneous intensity value in the quantization fluctuation eigenvalue to obtain the temperature deviation value, further considering the influence of humidity fluctuation on temperature control. In the actual hatching environment, the fluctuation of humidity may cause certain interference to the measurement and control of temperature. Through this processing method, the influence of humidity fluctuation on the calculation of the temperature deviation value can be eliminated to a certain extent, improving the stability of temperature control.

[0051] Perform a natural logarithm transformation on the ratio of the current element of the humidity error variance matrix to the target threshold to generate the initial deviation amount. The logarithm transformation can compress a relatively large range of humidity error variance ratios, making the value of the initial deviation amount more reasonable and easy to process. At the same time, the logarithm transformation can also highlight the relative change of the humidity error variance ratio, more accurately reflecting the deviation degree of the humidity error relative to the target threshold. Generate a compensation factor according to the product of the temperature cumulative amplitude value and the humidity instantaneous fluctuation intensity, and perform dynamic range compression on the initial deviation amount through the compensation factor. This method considers the mutual influence between temperature and humidity and can avoid excessive fluctuation of the humidity deviation value. For example, when the temperature cumulative amplitude value is large and the humidity instantaneous fluctuation intensity is also large, the compensation factor will perform a greater degree of compression on the initial deviation amount, preventing the system from becoming unstable due to excessive response. The dynamic range compression can make the humidity deviation value change within a reasonable range, avoiding drastic adjustment of control parameters due to sudden increase or decrease of humidity error, which helps to improve the stability of humidity control, keep the humidity in the hatching environment within a relatively stable range, and is beneficial to the healthy hatching of chicken embryos.

[0052] In another preferred embodiment of the present invention, generate an initial weight according to the ratio of the dynamic gradient eigenvalue to the quantization fluctuation eigenvalue, and correct the initial weight through the humidity historical deviation ratio to obtain the normalized weight coefficient; multiply the difference between the current element of the temperature error variance matrix and the target threshold by the normalized weight coefficient to obtain the basic deviation amount; process the basic deviation amount based on the instantaneous intensity value in the quantization fluctuation eigenvalue to obtain the temperature deviation value, including: Calculate the ratio of the dynamic gradient eigenvalue to the quantization fluctuation eigenvalue to obtain a ratio reflecting the relative intensity of the temperature and humidity error changes, which is used as the initial weight; Calculate the proportion of the humidity error historical data that deviates from the normal range, that is, the humidity historical deviation ratio, and adjust the initial weight with the humidity historical deviation ratio to obtain the corrected initial weight; Normalize the corrected initial weight to obtain the normalized weight coefficient; Obtain the current element of the temperature error variance matrix, that is, the average value of the squared deviation of the temperature error within the current sliding window, calculate the difference between the current element and the preset temperature error target threshold, and fuse the difference with the normalized weight coefficient to obtain a basic deviation amount reflecting the degree of deviation of the temperature error fluctuation from the target state; Extract the instantaneous intensity value at the current moment from the quantization fluctuation eigenvalue, that is, the fluctuation peak value of the current humidity error, and adjust the basic deviation amount with the instantaneous intensity value to obtain a temperature deviation value reflecting the temperature error and humidity interference.

[0053] In the embodiment of the present invention, by calculating the ratio of the dynamic gradient eigenvalue of the temperature error (reflecting the error change rate, such as the heating / cooling trend) to the quantization fluctuation eigenvalue of the humidity error (reflecting the instantaneous fluctuation amplitude, such as the start / stop interference of the humidification device), the relative influence intensity of the two is quantified. The formula is: , where, is a minimum value to avoid the denominator being zero; there is often a coupling relationship between temperature and humidity in the incubation environment (such as heating may be accompanied by a decrease in humidity), but the interference sources are different (temperature fluctuations mostly originate from heating / cooling devices, and humidity fluctuations mostly originate from humidification / ventilation). The dominant interference type can be dynamically judged through the ratio: If the ratio > 1, it indicates that the temperature error change rate dominates (such as the continuous operation of the heating device causing the temperature to continuously deviate), and the weight tilts towards the temperature; If the ratio < 1, it indicates that the instantaneous humidity fluctuation is more significant (such as a sudden ventilation causing a sharp drop in humidity), and the weight tilts towards the humidity correction.

[0054] The dynamic gradient eigenvalue contains trend information in the time dimension (such as the rising trend of the temperature error in 3 consecutive cycles). Compared with single-cycle data, it can predict the temperature deviation risk 2 - 3 sampling cycles in advance, enabling the system to adjust the control parameters before the interference expands.

[0055] Calculate the proportion of the humidity error historical data that deviates from the normal range (such as the proportion of the number of cycles in which the humidity error exceeds the threshold in the past N cycles). The formula is: , where, N ≥ 3; Use this proportion to correct the initial weight: the corrected initial weight = initial weight × (1 - humidity historical deviation proportion).

[0056] If the humidity historical deviation proportion is relatively low (such as < 20%), it indicates that the current humidity fluctuation may be accidental noise (such as the instantaneous error of the sensor). The corrected initial weight depends more on the temperature trend, avoiding over-control triggered by accidental humidity fluctuations; if the proportion is relatively high (such as > 50%), it indicates that there is a persistent abnormality in the humidity (such as a malfunction of the humidification device), and the weight tilts towards the humidity correction to stabilize the humidity environment first.

[0057] Filter short-term noise through the statistical characteristics of historical data. For example, when the ventilation system of the hatching chamber starts and stops periodically (such as ventilating for 5 minutes every hour), resulting in regular fluctuations in humidity, the historical deviation ratio can identify this pattern and adaptively adjust the weight to avoid frequent oscillations of control parameters.

[0058] Normalize the corrected initial weight (such as scaling it to the interval [0, 1]). The formula is: ; The variance of temperature error (unit: °C 2 ) and the fluctuation of humidity error (unit: %RH) have different dimensions. After normalization, the weight coefficient is a dimensionless value, ensuring that the temperature and humidity effects are weighted and summed in the same dimension and avoiding control deviations caused by dimensional differences.

[0059] During different hatching stages (such as strictly controlling temperature in the early stage and paying attention to humidity during the hatching period), dynamically update the maximum / minimum weight through historical data to achieve staged adaptation of the control strategy. For example: Early stage of embryo development: The normalized weight coefficient is biased towards temperature (such as 0.7 - 0.9), ensuring that the temperature fluctuation ≤ ±0.2 °C; Hatching period: The weight coefficient is biased towards humidity (such as 0.5 - 0.7), giving priority to ensuring that the humidity is within the range of ±3%RH.

[0060] Calculate the difference between the current element of the temperature error variance matrix (denoted as ) and the target threshold (denoted as ), and multiply it by the normalized weight coefficient (denoted as ω). The formula is: ; When the difference is positive, it indicates that the temperature error fluctuation exceeds the target (such as uneven heating resulting in an increase in temperature fluctuation), and the control strength needs to be enhanced; when the difference is negative, it indicates that the fluctuation is within the controllable range, and the control action can be reduced to extend the life of the actuator.

[0061] Weight modulation control sensitivity: When the normalized weight coefficient ω is relatively high (such as 0.8), the basic deviation amount is more sensitive to temperature error, which is suitable for scenarios with frequent disturbances; when ω is relatively low (such as 0.3), the control is smoother, which is suitable for stable stages.

[0062] Extract the instantaneous intensity value at the current moment (denoted as , such as the absolute value peak of the humidity error in adjacent cycles) from the humidity quantization fluctuation eigenvalue, and correct the basic deviation amount: , where k is an adjustment coefficient, 0 < k < 1.

[0063] Instantaneous humidity fluctuations may indirectly affect temperature measurement and control through physical coupling (such as changes in the thermal conductivity of humid air). For example, during the humidification process, the absorption of heat by humid air causes a short-term misjudgment of the temperature sensor. At this time increases, and the correction term appropriately reduces the temperature deviation value to prevent the system from misjudging it as a real temperature interference. When both temperature and humidity are abnormal (such as a heating equipment failure + a humidification pump leakage), through compressing the basic deviation amount to prevent the deterioration of another variable caused by single-variable control (such as excessive heating may exacerbate the humidity drop), and realizing multi-variable collaborative control.

[0064] In another preferred embodiment of the present invention, the ratio of the current element of the humidity error variance matrix to the target threshold is subjected to a natural logarithm transformation to generate an initial deviation amount; a compensation factor is generated according to the product of the temperature cumulative amplitude value and the humidity instantaneous fluctuation intensity, and the dynamic range of the initial deviation amount is compressed through the compensation factor to obtain a humidity deviation value, including: Obtain the current element of the humidity error variance matrix, that is, the average value of the squared deviation of the humidity error within the current sliding window, and calculate the humidity ratio of the current element to the preset humidity error target threshold, which reflects the deviation multiple of the current humidity error fluctuation relative to the target state; Take the natural logarithm of the humidity ratio to compress the numerical range to the logarithmic scale to obtain an initial deviation amount reflecting the degree of humidity error deviation; Calculate the absolute value cumulative sum of the temperature error optimization parameter sequence within the current sliding window, which characterizes the overall deviation degree and duration of the temperature error, and forms a temperature cumulative amplitude value; Extract the instantaneous fluctuation peak value at the current moment from the humidity optimization parameter sequence, that is, the absolute value of the maximum difference between adjacent sampling points, and record it as the humidity instantaneous fluctuation intensity; Multiply the temperature cumulative amplitude value by the humidity instantaneous fluctuation intensity to obtain a compensation factor; Adjust the initial deviation amount through the compensation factor, and narrow the numerical range of the initial deviation amount through non-linear mapping to obtain a humidity deviation value that balances the coupling effect of temperature and humidity.

[0065] In the embodiment of the present invention, calculate the current element of the humidity error variance and the target threshold to obtain a humidity ratio; take the natural logarithm of the humidity ratio to generate an initial deviation amount: ; Numerical range compression and dynamic adaptation. When the humidity error variance significantly exceeds the target threshold (such as ), the humidity ratio is 10, and the natural logarithm is compressed to 2.3 to prevent the control parameter from fluctuating violently due to the original ratio of 10; When the error is small (such as ), the logarithmic output is -0.69, retaining the negative deviation information (insufficient humidity fluctuation), which is convenient for the system to judge whether it is necessary to enhance humidity regulation.

[0066] The logarithmic transformation converts the "absolute error" into the "relative error multiple". For example: The target threshold is 1%RH 2 When the current variance is 4%RH 2 (ratio 4) and 9%RH 2 The logarithmic difference of (ratio 9) is , indicating that the deviation degree of the latter is 2.25 times that of the former, rather than the absolute difference of 5%RH 2 , which is more in line with people's intuitive judgment of "error severity".

[0067] Calculate the absolute value sum of the temperature error optimization parameter sequence within the current sliding window (3 cycles). The formula is: , is the temperature prediction error optimization parameter.

[0068] The temperature error in a single cycle may be accidental noise (such as sensor jitter), but the sum of three consecutive cycles can reflect the continuous deviation (such as the temperature continuously rising due to the aging of the heating element). For example: If , , , and the cumulative amplitude value is 0.45°C, indicating that the temperature error shows an increasing trend, and the subsequent out-of-control risk needs to be vigilant.

[0069] There is a physical coupling between temperature and humidity (such as the evaporation of moisture accelerating in a high-temperature environment). The cumulative amplitude value can quantify the long-term impact of temperature on humidity. For example: Continuous high temperature in the hatching chamber (large cumulative amplitude value) will cause the humidity to naturally decrease. At this time, the humidity deviation value needs to be dynamically adjusted in combination with the temperature impact to avoid the system misjudging it as a humidification equipment failure.

[0070] Extract the absolute value of the maximum difference between adjacent sampling points in the humidity optimization parameter sequence. The formula is: ; Among them, is the humidity prediction error optimization parameter at the th sampling point.

[0071] When events such as the start and stop of the humidification equipment and the sudden opening of the ventilation port occur, the humidity error will show an instantaneous spike (such as the absolute value of the difference suddenly increasing to 5%RH). The instantaneous fluctuation intensity can quickly identify such sudden interferences.

[0072] Distinguish the type of interference: If the instantaneous fluctuation intensity is high but the cumulative amplitude value is low (such as a sudden drop in humidity caused by a single ventilation), it indicates that the interference is a short-term pulse type, and the control strategy is mainly based on rapid compensation; if both are high (such as continuous leakage of the humidification pump resulting in continuous humidity fluctuations), it indicates the existence of a persistent fault, and it is necessary to enhance the control strength and trigger a fault warning.

[0073] The compensation factor is obtained by multiplying the temperature cumulative amplitude value by the humidity instantaneous fluctuation intensity: Compensation factor = temperature cumulative amplitude value × humidity instantaneous fluctuation intensity; The initial deviation is compressed in the dynamic range by the compensation factor: , where λ is an adjustment coefficient and λ > 0.

[0074] When temperature and humidity interferences exist simultaneously (such as heating and ventilation), the compensation factor increases, compressing the numerical range of the initial deviation and preventing the system from deteriorating another variable due to over-responding to a single variable. For example: the temperature cumulative amplitude value is 0.5 °C (on the high side), the humidity instantaneous fluctuation intensity is 4%RH (sudden drop), the compensation factor is 2.0, and the compression coefficient is 1 / (1 + 2λ). If λ = 0.5, the compressed deviation value is 1 / 2 of the initial value, preventing the heating action from exacerbating the humidity drop. The compression function is hyperbolic and has the characteristics of "linear response at small errors and saturated response at large errors": when the compensation factor is small (weak interference), the compression coefficient is close to 1, and the deviation value is close to the initial value, with sensitive control; when the compensation factor is large (strong interference), the compression coefficient approaches 0, avoiding the control parameters exceeding the actuator's capacity range (such as the maximum power limit of the humidifier) and ensuring system safety.

[0075] In another preferred embodiment of the present invention, the temperature deviation value and the humidity deviation value are normalized and weighted and summed, and dynamically scaled through a Gaussian kernel function to obtain a correction factor, including: Calculate the maximum value and the minimum value of the temperature deviation value in the preset sampling period respectively, calculate its first numerical deviation range, that is, the maximum value minus the minimum value; subtract the minimum value from the current temperature deviation value to obtain the first deviation difference, and calculate the deviation ratio between the first deviation difference and the first numerical deviation range to obtain the normalized temperature deviation value; Calculate the maximum value and the minimum value of the humidity deviation value in the preset sampling period respectively, calculate its second numerical deviation range, that is, the maximum value minus the minimum value; subtract the minimum value from the current humidity deviation value to obtain the second deviation difference, and calculate the deviation ratio between the second deviation difference and the second numerical deviation range to obtain the normalized humidity deviation value; The preset temperature influence weight coefficient and the humidity influence weight coefficient are used. Multiply the normalized temperature deviation value by the temperature influence weight coefficient to obtain the temperature contribution; multiply the normalized humidity deviation value by the humidity influence weight coefficient to obtain the humidity contribution; Add the temperature contribution and the humidity contribution to obtain the temperature-humidity deviation index value; substitute the temperature-humidity deviation index value into the Gaussian kernel function to calculate its function value, which is the correction factor.

[0076] In the embodiment of the present invention, the temperature deviation value (denoted as ) and the humidity deviation value (denoted as ) are respectively normalized. The formula is: ; Among them, represents the temperature deviation value in the preset sampling period. ; Among them, represents the humidity deviation value in the preset sampling period. The preset sampling period takes the first 3-5 sampling periods, reflecting the short-term fluctuation characteristics.

[0077] The temperature deviation value (unit: °C) and the humidity deviation value (unit: %RH) have different dimensions. After normalization, they are both mapped to the [0, 1] interval to ensure the comparability of their influences when weighted summing. For example: The temperature deviation value is +0.3 °C (0.8 after normalization) and the humidity deviation value is -4%RH (0.7 after normalization), which can be directly superimposed according to the weight, avoiding the problem of "temperature dimension being larger resulting in control bias".

[0078] Dynamically adjust the normalization range based on the maximum / minimum value of the preset data, rather than a fixed threshold, to adapt to the error fluctuation characteristics of different hatching stages: Early stage of embryo development: The temperature and humidity control is strict, and the error range is small (such as the temperature deviation is ±0.1 °C). Subtle changes can be captured after normalization; Hatching period: The allowable range of humidity fluctuation expands (such as ±5%RH). After normalization, it avoids control saturation caused by a large absolute value.

[0079] The preset temperature influence weight coefficient (α) and the humidity influence weight coefficient (β) satisfy α + β = 1, and calculate the temperature-humidity contribution: Temperature contribution = normalized temperature deviation value × α; Humidity contribution = normalized humidity deviation value × β; Temperature-humidity deviation index value = temperature contribution + humidity contribution.

[0080] By adjusting α and β, the control focus can be dynamically switched: Initial stage of hatching: Set α = 0.7, β = 0.3, and give priority to controlling the temperature (the embryo is more sensitive to temperature); Late hatching stage: Let α = 0.5 and β = 0.5, and control the balance of temperature and humidity (suitable humidity is required for hatching).

[0081] When both temperature and humidity deviate simultaneously, weighted summation is used to avoid a single variable dominating the control. For example: The normalized deviation value of temperature is 0.9 (high deviation) and the humidity is 0.9 (high deviation). After summing with equal weights, the index value is 0.9, and the system triggers a "strong interference" response; if the temperature is 0.9 but the humidity is 0.1, the index value is 0.5, and the system determines that the temperature interference is the main one, avoiding amplifying the control action due to slight humidity fluctuations.

[0082] The Gaussian kernel function is used to perform a non - linear transformation on the temperature and humidity deviation index value (denoted as S), and the formula is: ; where σ is the kernel bandwidth, reflecting the sensitivity to interference.

[0083] The function curve is bell - shaped. When S = 0, the output is 1 (no interference), and when S increases, the output decays exponentially.

[0084] Non - linear response characteristics: Sensitive to small interference: When S is in the interval [0, 0.5], the function value decreases gently (for example, when S = 0.3, the output is about 0.86), ensuring that the system responds promptly to early slight interference; Saturated to large interference: When S > 1, the function value approaches 0 (for example, when S = 2, the output is about 0.14), avoiding excessive adjustment of control parameters under strong interference, which may lead to oscillation.

[0085] Dynamically adjust the interference sensitivity: By adjusting the kernel bandwidth σ, different interference scenarios can be adapted: Stable scenario (such as no personnel operation at night): Let σ = 0.3, the function is sensitive to small fluctuations, maintaining high - precision control; High - interference scenario (such as frequent opening and closing of the warehouse door during the day): Let σ = 0.6, the function responds more gently to large fluctuations, reducing ineffective control actions.

[0086] By normalization and weighted summation, the two - dimensional deviation problem of temperature and humidity is transformed into a one - dimensional index value, simplifying the subsequent control algorithm design (for example, a BP neural network only requires a single input).

[0087] Physical mapping of interference intensity: The value of the correction factor is negatively correlated with the severity of interference (the smaller the value, the greater the interference), which conforms to the intuitive control logic. For example: Correction factor = 0.9, indicating slight interference, only requiring fine - tuning of PID parameters; Correction factor = 0.2, indicating strong interference, triggering a significant adjustment of the weights in the fuzzy rule base and a rapid optimization of the PID parameters.

[0088] The smoothing property of the Gaussian kernel function has a natural inhibitory effect on outliers. For example: At a certain moment, due to a sensor failure, the deviation value of temperature and humidity suddenly rises (both are 1 after normalization, S = 1), and the correction factor outputs 0.606 (non-zero), avoiding false alarms triggered by a single bad point. Compared with directly using the index value for control, the false alarm rate is reduced by 50%. The correction factor provides a normalized and low-noise comprehensive interference index for the subsequent BP neural network and fuzzy control, improving the convergence speed of the neural network and the reconstruction efficiency of the weights in the fuzzy rule base.

[0089] In another preferred embodiment of the present invention, according to the correction factor and the BP neural network, the slope of the activation function of the hidden layer of the neural network is dynamically adjusted, and the weights of the fuzzy rule base are reconstructed based on the numerical distribution characteristics of the correction factor to generate the corrected weights of the fuzzy rule base, including: Determine the initial slope parameter corresponding to the activation function of the hidden layer of the BP neural network; Determine the adjustment amount of the activation function slope according to the value of the correction factor to obtain the adjusted value of the slope; Add the adjusted value of the slope to the initial slope parameter to obtain the adjusted slope of the activation function; Analyze the numerical distribution characteristics of the correction factor, calculate the numerical distribution characteristics, and the numerical distribution characteristics include the mean, variance, maximum value, and minimum value of the correction factor; Determine the initial weights of the fuzzy rule base, and adjust the weights of each rule in the fuzzy rule base according to the numerical distribution characteristics of the correction factor to obtain the corrected weights of the fuzzy rule base.

[0090] In the embodiment of the present invention, assume that the activation function of the hidden layer is the Sigmoid function, and the initial slope parameter is k0 (which determines the steepness of the function curve).

[0091] Calculation of the adjustment amount. According to the correction factor (denoted as γ, with a value range of [0, 1]), design the adjustment amount Δk = η×(1 - γ) (η is the learning rate, > 0).

[0092] Adjusted slope: k = k0 + Δk = k0 + η(1 - γ).

[0093] When the correction factor γ is small (strong interference), 1 - γ increases, the slope k increases, and the Sigmoid function curve becomes steeper, enhancing the fitting ability of the neural network to strong non-linear error characteristics (such as sudden changes in temperature and humidity). For example: When γ = 0.2, k increases from the initial value of 2.0 to 2.8, and the derivative of the function near the input value of 0 increases from 0.5 to 0.68, making it more sensitive to subtle error changes. When γ is larger (weak interference), the slope k decreases, and the function curve becomes flatter, avoiding overfitting of the network to noise and improving the generalization ability.

[0094] Dynamic adjustment of the slope can accelerate model convergence. For example: In a strong interference scenario (such as equipment startup and shutdown), increasing the slope enables the network to capture error features faster, reducing the number of iterations by 25%; in a stable scenario, decreasing the slope reduces the gradient update amplitude, preventing oscillations and making the convergence process smoother.

[0095] Calculate the mean μ and variance σ of the statistical features of the correction factor 2 Reflecting the degree of interference fluctuation, extreme values (γ max , γ min ), used to identify extreme interference events (such as γ min < 0.1 indicates a sudden strong interference).

[0096] If the mean μ < 0.5 and the variance σ 2 > 0.1, it indicates that the system is in a "medium-strong interference and high fluctuation" state (such as frequent opening and closing of the warehouse door), and the dynamic adaptability of the fuzzy rules needs to be enhanced; If the mean μ > 0.8 and the variance σ 2 < 0.05, it indicates that the system is stable, and the rule adjustment frequency can be reduced to save computing resources. Calculate the statistical features through a sliding window (such as the previous 10 cycles), taking into account short-term interference (such as the current cycle γ min ) and long-term trends (such as the downward trend of the mean), avoiding misjudgment of single-cycle data.

[0097] Suppose the fuzzy rule base has m rules, and the initial weight is w j ∈ [0, 1] (j = 1, 2,..., m), usually uniformly distributed (such as w j = 1 / m).

[0098] Dynamic adjustment strategy: For each rule, adjust the weight according to the statistical features of the correction factor : ; Among them, is the historical correction factor sample corresponding to rule j, λ is the adjustment coefficient, is the minimum value to prevent the denominator from being zero).

[0099] For rules that perform well in a strong interference scenario (such as rule j having good control effect when γ < 0.3), its weight increases and is triggered preferentially; For obsolete or invalid rules (such as those only applicable to the early hatching stage), the weight is reduced or even disabled, solving the defect of fixed traditional fuzzy control rules.

[0100] Suppress rule conflicts: When multiple rules output conflicts (such as the "temperature increase" and "humidity decrease" rules being triggered simultaneously), the rule with the higher weight dominates the control, avoiding control failure caused by "averaging". For example: During the hatching period, the weight of the "increase humidity" rule is increased from 0.1 to 0.4, taking precedence over the "temperature fine-tuning" rule to ensure that the humidity meets the standard.

[0101] By adjusting the slope of the activation function, it can quickly adapt to the non-linear changes of error characteristics, reducing the fitting error for complex disturbances (such as temperature and humidity coupling oscillations); based on the reconstruction of rule weights according to the distribution of correction factors, the fuzzy logic is upgraded from "fixed threshold judgment" to "data-driven dynamic reasoning", improving the rule matching accuracy.

[0102] Multi-level response to interference intensity: Short-term response: The slope of the activation function is adjusted in real time (computing delay < 10ms) to quickly suppress sudden interference; Long-term learning: The fuzzy rule weights are iteratively updated according to the distribution characteristics of correction factors (once every 5 minutes) to gradually optimize the control strategy.

[0103] The slope adjustment only involves the parameters of the hidden layer (the computational complexity is O(n h ), where n h is the number of neurons in the hidden layer), which is much lower than the full-parameter update of the traditional BP network; the fuzzy rule weights are adjusted batchwise according to statistical characteristics without manual debugging for each rule.

[0104] In another preferred embodiment of the present invention, based on the weights of the corrected fuzzy rule base, a three-dimensional parameter adjustment surface is constructed, and the proportional, integral, and differential parameters of the PID controller are dynamically adjusted through the surface gradient search algorithm to generate a control signal, including: Determine the three variables of the three-dimensional parameter adjustment surface, which are the temperature deviation value, the humidity deviation value, and the weights of the corrected fuzzy rule base; Under different combinations of temperature deviation values, humidity deviation values, and weights of the corrected fuzzy rule base, record the corresponding values of the proportional, integral, and differential parameters of the PID controller; According to the values of the proportional, integral, and differential parameters of the PID controller, construct a three-dimensional parameter adjustment surface through neural network fitting; According to the temperature deviation value, humidity deviation value, and weights of the corrected fuzzy rule base in the current hatching chamber, find the corresponding state point on the three-dimensional parameter adjustment surface, and the state point represents the initial parameter settings of the PID controller under the current working conditions; At the current state point, calculate the gradient of the three-dimensional parameter adjustment surface. The gradient represents the direction of change of the three-dimensional parameter adjustment surface at the state point and includes three components, which respectively correspond to the change directions of the proportional, integral, and differential parameters of the PID controller. According to the gradient of the three-dimensional parameter adjustment surface and the search step size, adjust the proportional, integral, and differential parameters of the PID controller along the opposite direction of the gradient until the search step size is reached to obtain the optimized proportional, integral, and differential parameters of the PID controller. Based on the current temperature deviation value and humidity deviation value, the PID controller calculates the corresponding control signal by combining the adjusted proportional, integral, and differential parameters.

[0105] In the embodiment of the present invention, taking the temperature deviation value , humidity deviation value , and the corrected fuzzy rule weight as input variables, construct the three-dimensional mapping relationship of the PID parameters . The formula is: ; Train the neural network through historical data to fit this function to form a continuous and smooth parameter adjustment surface.

[0106] Traditional PID parameter tuning only considers single-variable errors (such as only temperature), while the surface takes into account both the temperature and humidity coupling interference and the adaptability of fuzzy rules simultaneously.

[0107] For example: when the temperature deviation is large , the humidity deviation is small and the fuzzy rule weight is biased towards temperature , the surface automatically outputs a parameter combination of high K p (rapid temperature increase) and low K i (avoiding integral saturation). Through training with historical working condition data, the surface covers possible interference combinations during the entire incubation period (such as temperature ±0.5°C, humidity ±8%RH, rule weight 0.3 - 0.9), pre-stores thousands of groups of optimized parameters, and avoids the high latency of online real-time calculation.

[0108] According to the current temperature and humidity deviation values and the fuzzy rule weight, locate the corresponding state point on the three-dimensional surface. The coordinates of this point are , and output the initial PID parameters .

[0109] Compared with traditional trial-and-error methods such as Ziegler-Nichols (which require iterations from dozens of seconds to several minutes), initial parameters can be obtained within 10 ms through surface interpolation, and the response speed to sudden disturbances (such as sudden changes in temperature and humidity caused by the opening of the warehouse door) is increased by 90%. When the hatching stage switches (such as from the early stage to the middle stage), the fuzzy rule weights change to trigger the migration of state points, and the PID parameters change continuously along the surface, avoiding the temperature overshoot (the overshoot amplitude is reduced by 60%) caused by the parameter jump in traditional segmented control.

[0110] Calculate the surface gradient at the state point , and the gradient direction represents the direction in which the parameters increase fastest, and the opposite direction is the direction in which the error decreases fastest.

[0111] Parameter adjustment: Update the parameters along the opposite direction of the gradient: ; where is the step size, which controls the adjustment amplitude.

[0112] Based on the initial parameters, convergence to the local optimal solution can be achieved through 3 - 5 gradient iterations, and the number of iterations is reduced by 80% compared with random search. For example: When the initial temperature fluctuates by ±0.5 °C, after 4 iterations, the fluctuation is reduced to within ±0.2 °C; the gradient direction guides the parameter adjustment along the steepest descent path of the surface, avoiding the oscillation caused by the blind parameter adjustment direction in traditional trial-and-error methods. For example: in humidity control, the differential parameter K d gradually increases along the opposite direction of the gradient, smoothly suppressing overshoot, rather than repeatedly increasing and decreasing resulting in fluctuations.

[0113] As Figure 2 shown, an embodiment of the present invention also provides an intelligent hatching warehouse anti-interference control system based on error self-learning, including: A collection module 100, configured to collect the temperature and humidity data streams of chicken embryo hatching in the hatching warehouse in real time; form a sliding detection window with the data of three consecutive sampling periods, and perform fusion processing on the temperature and humidity data within the window, and output an optimized parameter sequence of temperature prediction error and an optimized parameter sequence of humidity prediction error; A calculation module 200, configured to calculate a temperature deviation value and a humidity deviation value respectively according to the optimized parameter sequence of temperature prediction error and the optimized parameter sequence of humidity prediction error; A correction module 300, configured to perform normalized weighted summation on the temperature deviation value and the humidity deviation value, and perform dynamic scaling through a Gaussian kernel function to obtain a correction factor; An adjustment module 400 is configured to dynamically adjust the activation function slope of the hidden layer of the neural network according to a correction factor and a BP neural network, and reconstruct the weights of the fuzzy rule base based on the numerical distribution characteristics of the correction factor to generate corrected fuzzy rule base weights; A processing module 500 is configured to construct a three-dimensional parameter adjustment surface based on the corrected fuzzy rule base weights, and dynamically adjust the proportional, integral, and differential parameters of the PID controller through a surface gradient search algorithm to generate a control signal.

[0114] It should be noted that this system corresponds to the above method. All implementation manners in the above method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0115] The above are the preferred embodiments of the present invention. It should be pointed out that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and modifications can be made, and these improvements and modifications should also be regarded as the protection scope of the present invention.

Claims

1. An intelligent incubation warehouse anti-interference control method based on error self-learning is characterized in that: The method comprises: Step 1, real-time acquisition of temperature and humidity data streams of chicken embryo incubation in the incubation chamber; forming a sliding detection window with data from three consecutive sampling cycles, fusing the temperature and humidity data within the window, and outputting an optimized parameter sequence of temperature prediction error and an optimized parameter sequence of humidity prediction error; Step 2, calculating the temperature deviation value and the humidity deviation value respectively according to the optimized parameter sequence of the temperature prediction error and the optimized parameter sequence of the humidity prediction error; Step 3, performing normalized weighted summation on the temperature deviation value and the humidity deviation value, and dynamically scaling them through a Gaussian kernel function to obtain a correction factor; Step 4, dynamically adjusting the activation function slope of the hidden layer of the neural network according to the correction factor and the BP neural network, and reconstructing the weight of the fuzzy rule base based on the numerical distribution characteristics of the correction factor to generate the corrected fuzzy rule base weight; Step 5: Based on the modified fuzzy rule base weights, a three-dimensional parameter adjustment surface is constructed, and the proportional, integral and differential parameters of the PID controller are dynamically adjusted through the surface gradient search algorithm to generate a control signal.

2. The anti-interference control method for the intelligent incubation warehouse based on error self-learning according to claim 1 is characterized in that: The data of three consecutive sampling periods are used to form a sliding detection window, and the temperature and humidity data are fused in the window to output the optimized parameter sequence of the temperature prediction error and the optimized parameter sequence of the humidity prediction error, including: In the sliding detection window, the original temperature and humidity data at the current sampling moment and the two moments before it are constructed as time series vectors, each of which contains three consecutive data points. A state transition model is established based on the temperature observation values ​​of the previous two cycles to generate the temperature prediction value at the current moment; the actual temperature data collected is subtracted from the prediction value to obtain the temperature prediction error value; the filter gain is dynamically adjusted according to the noise covariance matrix and the observation noise covariance matrix, the prediction error is weighted and corrected, and the temperature prediction error optimization parameter at the current moment is output; Repeat the prediction-correction process for the humidity time series vector to generate the corresponding humidity prediction error optimization parameters; The temperature prediction error optimization parameters output from three consecutive sampling periods are arranged in chronological order to form a temperature prediction error optimization parameter sequence; the humidity prediction error optimization parameters are arranged to form a humidity prediction error optimization parameter sequence; The three consecutive parameter values ​​in the temperature optimization parameter sequence are processed by sliding average to obtain the temperature error mean; the square deviation of each parameter value is calculated according to the temperature error mean, and the arithmetic average of the three square deviations is used as the current element of the temperature error variance matrix; the same operation is performed on the humidity optimization parameter sequence to generate the current element of the humidity error variance matrix.

3. The anti-interference control method for the intelligent incubation warehouse based on error self-learning according to claim 2 is characterized in that: According to the optimized parameter sequence of the temperature prediction error and the optimized parameter sequence of the humidity prediction error, the temperature deviation value and the humidity deviation value are calculated respectively, including: The temperature optimization parameter sequence and the humidity optimization parameter sequence are aligned point by point according to the sampling timestamps. From the aligned temperature optimization parameter sequence, the change rate of the data before and after the current sampling point is calculated to obtain the dynamic gradient characteristic value reflecting the dynamic change trend of the temperature error sequence; from the humidity optimization parameter sequence, the absolute value of the difference between adjacent data is calculated to obtain the quantitative fluctuation characteristic value representing the instantaneous fluctuation degree of the humidity error; Generate an initial weight according to the ratio of the dynamic gradient eigenvalue to the quantized fluctuation eigenvalue, and correct the initial weight according to the humidity history deviation ratio to obtain a normalized weight coefficient; multiply the difference between the current element of the temperature error variance matrix and the target threshold by the normalized weight coefficient to obtain a basic deviation; process the basic deviation based on the instantaneous intensity value in the quantized fluctuation eigenvalue to obtain a temperature deviation value; The ratio of the current element of the humidity error variance matrix to the target threshold is transformed into a natural logarithm to generate an initial deviation. The compensation factor is generated according to the product of the temperature cumulative amplitude value and the instantaneous humidity fluctuation intensity. The initial deviation is dynamically compressed by the compensation factor to obtain the humidity deviation value.

4. The anti-interference control method for the intelligent incubation warehouse based on error self-learning according to claim 3 is characterized in that: The initial weight is generated according to the ratio of the dynamic gradient eigenvalue to the quantitative fluctuation eigenvalue, and the initial weight is corrected by the humidity history deviation ratio to obtain the normalized weight coefficient; The difference between the current element of the temperature error variance matrix and the target threshold is multiplied by the normalized weight coefficient to obtain the basic deviation; The basic deviation is processed based on the instantaneous intensity value in the quantized fluctuation characteristic value to obtain the temperature deviation value, including: Calculate the ratio of the dynamic gradient eigenvalue to the quantized fluctuation eigenvalue to obtain a ratio that reflects the relative intensity of the temperature and humidity error changes as the initial weight; Calculate the ratio of humidity error historical data that deviates from the normal range, that is, the humidity historical deviation ratio, and use the humidity historical deviation ratio to adjust the initial weight to obtain a corrected initial weight; The corrected initial weight is normalized to obtain a normalized weight coefficient; Get the current element of the temperature error variance matrix, that is, the average value of the square of the temperature error deviation in the current sliding window, calculate the difference between the current element and the preset temperature error target threshold, merge the difference with the normalized weight coefficient, and obtain the basic deviation that reflects the degree of deviation of the temperature error fluctuation from the target state; The instantaneous intensity value at the current moment, that is, the fluctuation peak value of the current humidity error, is extracted from the quantized fluctuation characteristic value, and the basic deviation is adjusted with the instantaneous intensity value to obtain the temperature deviation value reflecting the temperature error and humidity interference.

5. The anti-interference control method for intelligent incubation warehouse based on error self-learning according to claim 4 is characterized in that: The ratio of the current element of the humidity error variance matrix to the target threshold is transformed into a natural logarithm to generate an initial deviation; a compensation factor is generated according to the product of the temperature cumulative amplitude value and the instantaneous humidity fluctuation intensity, and the initial deviation is dynamically compressed by the compensation factor to obtain a humidity deviation value, including: Get the current element of the humidity error variance matrix, that is, the average value of the square of the humidity error deviation in the current sliding window, calculate the humidity ratio of the current element to the preset humidity error target threshold, and reflect the deviation multiple of the current humidity error fluctuation relative to the target state; The natural logarithm of the humidity ratio is taken and the numerical range is compressed to the logarithmic scale to obtain the initial deviation reflecting the degree of humidity error deviation; Calculate the absolute value cumulative sum of the temperature error optimization parameter sequence in the current sliding window to characterize the overall deviation degree and duration of the temperature error, and form the temperature cumulative amplitude value; Extract the instantaneous fluctuation peak value at the current moment from the humidity optimization parameter sequence, that is, the maximum absolute value of the difference between adjacent sampling points, and record it as the humidity instantaneous fluctuation intensity; Multiply the temperature cumulative amplitude value by the instantaneous humidity fluctuation intensity to obtain the compensation factor; The initial deviation is adjusted by the compensation factor, and the numerical range of the initial deviation is narrowed by nonlinear mapping to obtain the humidity deviation value that balances the coupling effect of temperature and humidity.

6. The anti-interference control method for intelligent incubation warehouse based on error self-learning according to claim 5 is characterized in that: The temperature deviation and humidity deviation are normalized and weighted and then dynamically scaled by a Gaussian kernel function to obtain the correction factor, including: Calculate the maximum and minimum values ​​of the temperature deviation value in the preset sampling period respectively, and calculate its first numerical deviation range, that is, the maximum value minus the minimum value; subtract the minimum value from the current temperature deviation value to obtain a first deviation difference, and calculate the deviation ratio between the first deviation difference and the first numerical deviation range to obtain a normalized temperature deviation value; Calculate the maximum and minimum values ​​of the humidity deviation value in the preset sampling period respectively, and calculate its second numerical deviation range, that is, the maximum value minus the minimum value; subtract the minimum value from the current humidity deviation value to obtain a second deviation difference, and calculate the deviation ratio between the second deviation difference and the second numerical deviation range to obtain a normalized humidity deviation value; The temperature influence weight coefficient and the humidity influence weight coefficient are preset, and the temperature contribution is obtained by multiplying the normalized temperature deviation value by the temperature influence weight coefficient; the humidity contribution is obtained by multiplying the normalized humidity deviation value by the humidity influence weight coefficient; Add the temperature contribution and the humidity contribution to obtain the temperature and humidity deviation index value; substitute the temperature and humidity deviation index value into the Gaussian kernel function and calculate its function value, i.e., the correction factor.

7. The anti-interference control method for intelligent incubation warehouse based on error self-learning according to claim 6 is characterized in that: According to the correction factor and the BP neural network, the activation function slope of the hidden layer of the neural network is dynamically adjusted, and the weight of the fuzzy rule base is reconstructed based on the numerical distribution characteristics of the correction factor to generate the corrected fuzzy rule base weight, including: Determine the initial slope parameter corresponding to the activation function of the hidden layer of the BP neural network; Determine the adjustment amount of the slope of the activation function according to the value of the correction factor to obtain the adjustment value of the slope; Add the adjusted value of the slope to the initial slope parameter to obtain the adjusted slope of the activation function; Analyze the numerical distribution characteristics of the correction factor and calculate the numerical distribution characteristics, which include the mean, variance, maximum value and minimum value of the correction factor; The initial weight of the fuzzy rule base is determined, and the weight of each rule in the fuzzy rule base is adjusted according to the numerical distribution characteristics of the correction factor to obtain the corrected fuzzy rule base weight.

8. The anti-interference control method for intelligent incubation warehouse based on error self-learning according to claim 7 is characterized in that: Based on the weights of the modified fuzzy rule base, a three-dimensional parameter adjustment surface is constructed, and the proportional, integral and differential parameters of the PID controller are dynamically adjusted through the surface gradient search algorithm to generate control signals, including: Determine the three variables of the three-dimensional parameter adjustment surface, which are the temperature deviation value, the humidity deviation value and the weight of the modified fuzzy rule base; Under different combinations of temperature deviation values, humidity deviation values ​​and modified fuzzy rule base weights, the values ​​of the proportional, integral and differential parameters of the corresponding PID controller are recorded; According to the values ​​of the proportional, integral and differential parameters of the PID controller, a three-dimensional parameter adjustment surface is constructed through neural network fitting; According to the temperature deviation value, humidity deviation value and the weight of the modified fuzzy rule base in the current incubator, the corresponding state point is found on the three-dimensional parameter adjustment surface. The state point represents the initial parameter setting of the PID controller under the current working condition. At the current state point, the gradient of the three-dimensional parameter adjustment surface is calculated. The gradient represents the direction in which the three-dimensional parameter adjustment surface changes at the state point. It contains three components, which correspond to the change directions of the proportional, integral and differential parameters of the PID controller. The gradient and search step of the surface are adjusted according to the three-dimensional parameters, and the proportional, integral and differential parameters of the PID controller are adjusted in the opposite direction of the gradient until the search step is reached to obtain the proportional, integral and differential parameters of the optimized PID controller; The PID controller calculates the corresponding control signal based on the current temperature deviation and humidity deviation values ​​combined with the adjusted proportional, integral and differential parameters. 9.An intelligent incubation warehouse anti-interference control system based on error self-learning, characterized in that: The system is used to perform the method according to any one of claims 1 to 8, comprising: The acquisition module is used to collect the temperature and humidity data stream of the chicken embryo incubation in the incubation chamber in real time; the data of three consecutive sampling cycles form a sliding detection window, the temperature and humidity data are fused in the window, and the optimized parameter sequence of the temperature prediction error and the optimized parameter sequence of the humidity prediction error are output; A calculation module, used for calculating the temperature deviation value and the humidity deviation value respectively according to the optimization parameter sequence of the temperature prediction error and the optimization parameter sequence of the humidity prediction error; A correction module is used to perform normalized weighted summation of temperature deviation and humidity deviation, and dynamically scale them through a Gaussian kernel function to obtain a correction factor; An adjustment module is used to dynamically adjust the activation function slope of the hidden layer of the neural network according to the correction factor and the BP neural network, and reconstruct the weight of the fuzzy rule base based on the numerical distribution characteristics of the correction factor to generate the corrected fuzzy rule base weight; The processing module is used to construct a three-dimensional parameter adjustment surface based on the modified fuzzy rule base weights, and dynamically adjust the proportional, integral and differential parameters of the PID controller through a surface gradient search algorithm to generate a control signal.

Images