Multi-layer intelligent coop based on environment feedback and temperature and humidity control system

Abstract

The invention relates to the technical field of poultry breeding environment monitoring, and discloses a multi-layer intelligent coop and temperature and humidity control system based on environment feedback, an environment prediction module predicts the temperature and humidity conditions of each layer in the future by using historical environment data of the multi-layer coop and chicken flock behavior data; the control decision module determines an adjustment strategy of the temperature and humidity of each layer according to the predicted value and outputs a preliminary control instruction; and the instruction optimization module optimizes the preliminary control instruction by comparing the predicted temperature and humidity with the actually monitored temperature and humidity, and outputs a final control instruction. The system control center combines actual data, generates a preliminary instruction through environment prediction and control decision to adjust the environment, continuously monitors actual temperature and humidity, compares the actual temperature and humidity with a prediction result, and triggers a deviation correction model to recalculate a final instruction when a significant deviation is found. According to the invention, accurate control of the temperature and humidity of each layer in the multi-layer coop is realized, and the comfort and growth efficiency of chicken flocks are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of poultry breeding environment monitoring, and particularly to a multi-layer intelligent chicken cage and temperature and humidity control system based on environmental feedback. Background Art

[0002] In modern poultry breeding industry, the control of the chicken cage environment has a crucial impact on the health, growth efficiency and product quality of the chicken flock. Especially in the multi-layer chicken cage breeding mode, due to the differences between layers and the changes of the external climate, there may be significant differences in the temperature and humidity conditions in each layer of chicken cages, bringing greater challenges to breeding management. Traditional temperature and humidity control systems usually adopt fixed control strategies and cannot dynamically adjust according to the actual environmental conditions in the chicken cage and the changes of the chicken flock behavior. This lack of flexibility in control may lead to excessive fluctuations in the chicken cage environment, affecting the comfort and growth performance of the chicken flock. In addition, traditional systems often rely on single sensor data for feedback control, and are easily affected by sensor errors and external interferences, resulting in poor control effects.

[0003] With the rapid development of Internet of Things, big data and artificial intelligence technologies, intelligent breeding has become a new trend. In multi-layer chicken cage breeding, through the integration of multiple sensors and intelligent algorithms, real-time monitoring and precise control of the chicken cage environment can be achieved. However, most of the existing intelligent chicken cage systems focus on the control of a single environmental factor, such as temperature or humidity, and lack a comprehensive control system that comprehensively considers the environmental differences between layers in the multi-layer chicken cage and the changes of the chicken flock behavior.

[0004] Therefore, there is a need for a multi-layer intelligent chicken cage and temperature and humidity control system based on environmental feedback, which can comprehensively consider the historical environmental data, chicken flock behavior data and external climate information in the multi-layer chicken cage, and through constructing an environmental prediction model and a control decision model, achieve accurate prediction and dynamic control of the temperature and humidity in each layer of chicken cages. At the same time, the system also needs to have a deviation correction function, which can adjust the control instructions in time according to the comparison between the real-time monitoring data and the prediction results to ensure the stability and suitability of the chicken cage environment. Summary of the Invention

[0005] The purpose of the present invention is to provide a multi-layer intelligent chicken cage and temperature and humidity control system based on environmental feedback to solve the problems raised in the above background art.

[0006] To achieve the above purpose, the present invention provides the following technical solution: A multi-layer intelligent chicken cage temperature and humidity control system based on environmental feedback, the system includes:

[0007] An environmental prediction module, which is used to build a chicken coop environmental prediction model. This model predicts the temperature and humidity conditions of each layer in the multi-layer chicken coop for a period of time in the future based on the input historical environmental data and chicken group behavior data of the multi-layer chicken coop, and outputs the predicted temperature and humidity values of each layer in the future. Among them, the historical environmental data includes the historical temperature, humidity, ventilation conditions and external climate information of each layer in the multi-layer chicken coop, and the chicken group behavior data includes the activity levels, drinking water and feed consumption of the chicken groups on each layer.

[0008] A control decision-making module, which is used to build a temperature and humidity control decision-making model. This model is used to determine the adjustment strategies for the temperature and humidity of each layer in the multi-layer chicken coop. Its input data is the predicted temperature and humidity values of each layer in the multi-layer chicken coop output by the environmental prediction module, and its output data is the preliminary temperature and humidity control instructions for each layer.

[0009] An instruction optimization module, which is used to build a deviation correction model. This model optimizes the preliminary temperature and humidity control instructions for each layer output by the control decision-making module according to the differences between the predicted temperature and humidity of each layer in the multi-layer chicken coop output by the environmental prediction module and the actual temperature and humidity of each layer obtained through real-time monitoring by sensors, and outputs the final temperature and humidity control instructions for each layer after optimization.

[0010] And a system control center, which is used to combine the actual historical environmental data of the multi-layer chicken coop and the actual chicken group behavior data, generate the preliminary temperature and humidity control instructions for each layer of the multi-layer chicken coop through the environmental prediction module and the control decision-making module to adjust the environment of each layer in the multi-layer chicken coop, continuously monitor the actual temperature and humidity of each layer in the multi-layer chicken coop, and compare with the prediction results of the environmental prediction module. If a significant deviation is found, the deviation correction model is immediately triggered to recalculate the final temperature and humidity control instructions for each layer.

[0011] Preferably, the environmental prediction module uses the long short-term memory network LSTM algorithm to build the chicken coop environmental prediction model.

[0012] Preferably, the training steps of the chicken coop environmental prediction model include:

[0013] Collect training data, including historical environmental data of the multi-layer chicken coop, chicken group behavior data and the corresponding actual temperature and humidity data in the chicken coop.

[0014] Preprocess the collected training data, including data cleaning and data standardization processing.

[0015] Use the long short-term memory network LSTM algorithm to build the initial structure of the chicken coop environmental prediction model and set the hyperparameters of the model.

[0016] Use the preprocessed data set to train the initial model, and adjust the model parameters through the iterative optimization algorithm to make the model accurately predict the temperature and humidity conditions in the chicken coop for a period of time in the future. Among them, the label of the training data is the actual temperature and humidity data in the chicken coop.

[0017] During the training process, the hold-out method is used to divide the dataset to verify the model performance, and the model with the optimal performance is selected as the final chicken coop environment prediction model.

[0018] Preferably, the method for data cleaning includes:

[0019] For missing values, the forward and backward value averaging method is used for filling;

[0020] For outliers, the Isolation Forest algorithm is used to detect outliers and replace them with the median.

[0021] Preferably, the control decision-making module uses the reinforcement learning Q-learning algorithm to construct a temperature and humidity control decision-making model.

[0022] Preferably, the training steps of the temperature and humidity control decision-making model include:

[0023] S1: Define the state space, action space, and reward function of the Q-learning algorithm; among them, the state space is the predicted future temperature and humidity values output by the chicken coop environment prediction model, the action space is the possible temperature and humidity control instructions, and the reward function is set according to the temperature and humidity control effect;

[0024] S2: Initialize the Q-table and set the hyperparameters of the Q-learning algorithm, including the learning rate, discount factor, and exploration rate;

[0025] S3: During the training process, select an action according to the current state, observe the new state and reward after executing the action, and update the Q value of the corresponding state-action pair in the Q-table;

[0026] S4: Repeat S3 until the Q-table converges or reaches the preset number of training rounds to obtain the trained temperature and humidity control decision-making model.

[0027] Preferably, the instruction optimization module uses the Support Vector Machine (SVM) algorithm to construct a deviation correction model.

[0028] Preferably, the training steps of the deviation correction model include:

[0029] Collect training data, including the difference between the predicted temperature and humidity output by the chicken coop environment prediction model and the actual temperature and humidity monitored in real time by the sensor; at the same time, collect the optimal temperature and humidity control instructions marked by experts under the given difference between the predicted temperature and humidity and the actual temperature and humidity as the labels of the training data;

[0030] Construct a Support Vector Machine (SVM) algorithm model and set the parameters of the SVM, including the kernel function type, penalty parameter, and kernel parameter;

[0031] The support vector machine (SVM) algorithm model is trained using training data. Through iterative learning, the model can accurately predict the optimal temperature and humidity control instructions under given input differences.

[0032] After the training is completed, a deviation correction model is obtained.

[0033] Preferably, the method for determining whether the deviation between the actual temperature and humidity and the predicted temperature and humidity is significant is as follows: calculate the absolute difference between the actual temperature and humidity and the predicted temperature and humidity, set a difference threshold, compare the calculated absolute difference with the set difference threshold. If the absolute difference exceeds the set threshold, it is determined as a significant deviation.

[0034] Preferably, a multi-layer intelligent chicken coop based on environmental feedback includes a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, it implements the above-mentioned multi-layer intelligent chicken coop temperature and humidity control system.

[0035] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0036] By integrating the environmental prediction module, the present invention can comprehensively consider the historical environmental data, chicken flock behavior data, and external climate information in the multi-layer chicken coop, and construct an accurate chicken coop environmental prediction model. This model can accurately predict the temperature and humidity conditions of each layer in the multi-layer chicken coop in the future for a period of time, providing a reliable basis for control decisions, thereby significantly improving the accuracy of environmental control.

[0037] According to the output of the environmental prediction module, the control decision module determines the temperature and humidity adjustment strategies for each layer in the multi-layer chicken coop and outputs preliminary temperature and humidity control instructions. The instruction optimization module further optimizes the preliminary control instructions based on the comparison between the real-time monitoring data and the prediction results to ensure that the control instructions are more in line with the actual environmental requirements. This dynamic adjustment and optimization mechanism enables the environment in the chicken coop to always maintain the most suitable state. The present invention has a deviation correction function. When a significant deviation occurs between the real-time monitoring data and the prediction results, it can immediately trigger the deviation correction model to recalculate the final temperature and humidity control instructions for each layer. This function effectively enhances the stability and adaptability of the system, enabling the system to cope with various emergencies and external interferences, and ensuring the stability and suitability of the environment in the chicken coop.

[0038] By precisely controlling the temperature and humidity of each layer in the multi-layer chicken coop, the present invention provides a more comfortable living environment for the chicken flock. This helps to reduce the stress response of the chicken flock caused by environmental discomfort, improve the health level and growth efficiency of the chicken flock. At the same time, the optimized environmental control also helps to improve the quality and market competitiveness of chicken products.

[0039] The proposal and implementation of the present invention provide a new intelligent solution for modern poultry farming. By integrating Internet of Things, big data, and artificial intelligence technologies, real-time monitoring and precise control of the chicken coop environment are achieved, promoting the intelligent, information-based, and modern development of the poultry farming industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 It is the working principle diagram of the multi-layer intelligent chicken coop temperature and humidity control system based on environmental feedback according to the present invention;

[0041] Figure 2 It is the flow chart of constructing the chicken coop environment prediction model by using the long short-term memory network (LSTM) algorithm;

[0042] Figure 3 It is the flow chart of constructing the temperature and humidity control decision model by using the reinforcement learning Q-learning algorithm. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0043] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0044] Please refer to Figures 1 - 3 , the present invention provides a technical solution: a multi-layer intelligent chicken coop temperature and humidity control system based on environmental feedback, and the system includes:

[0045] Environmental prediction module: responsible for constructing the chicken coop environment prediction model. This model uses machine learning algorithms and takes the historical environmental data of the multi-layer chicken coop and the chicken flock behavior data as inputs. The historical environmental data includes, but is not limited to, the historical temperature, humidity, ventilation conditions of each layer in the multi-layer chicken coop, and external climate information, such as outdoor temperature, humidity, wind speed, and sunshine time, etc. The chicken flock behavior data includes the activity level, drinking water frequency, feed consumption situation, etc. of each layer of the chicken flock. During the model training process, these historical data are preprocessed, including data cleaning, missing value filling, and normalization processing. The processed data is input into the prediction model, and the model learns the mapping relationship between the historical data and the future temperature and humidity conditions. The model can output the temperature and humidity prediction values of each layer in the multi-layer chicken coop for a future period of time (such as the next 24 hours).

[0046] Control Decision-making Module: Responsible for constructing the temperature and humidity control decision-making model. This model determines the temperature and humidity adjustment strategies for each layer in the multi-layer chicken cage based on the predicted future temperature and humidity values output by the environmental prediction module. In rule-based control, a series of temperature and humidity control rules are preset in advance, such as turning on the ventilation equipment when the predicted temperature is higher than the set threshold, and increasing the humidification amount when the predicted humidity is lower than the set threshold. Fuzzy logic control can handle more complex control logics and generate control instructions through steps such as fuzzification, rule inference, and defuzzification. The reinforcement learning method finds the optimal control strategy through trial-and-error learning to maximize the comfort and growth efficiency of the chicken flock.

[0047] The output of the control decision-making model is the preliminary temperature and humidity control instructions for each layer, such as adjusting the ventilation rate, the working power of heating or cooling equipment, and the operating status of humidifying or dehumidifying equipment.

[0048] Instruction Optimization Module: Responsible for constructing the deviation correction model. This model optimizes the preliminary control instructions output by the control decision-making module based on the difference between the predicted temperature and humidity output by the environmental prediction module and the actual temperature and humidity obtained through real-time sensor monitoring.

[0049] The optimized control instructions are used as the final temperature and humidity control instructions for each layer and are sent to the corresponding execution devices, such as ventilation fans, heaters, coolers, and humidifiers, to adjust the environment of each layer in the multi-layer chicken cage.

[0050] System Control Center: Responsible for coordinating and managing the work of each module. The system control center first combines the historical environmental data of the actual multi-layer chicken cage and the actual chicken flock behavior data, and generates the preliminary temperature and humidity control instructions for each layer of the multi-layer chicken cage through the environmental prediction module and the control decision-making module. The system control center continuously monitors the actual temperature and humidity of each layer in the multi-layer chicken cage and obtains these data in real time through sensors. By comparing the actual temperature and humidity with the prediction results of the environmental prediction module, if a significant deviation (such as exceeding the preset threshold) is found, the deviation correction model is immediately triggered to recalculate the final temperature and humidity control instructions for each layer.

[0051] In this way, the system control center can ensure that the temperature and humidity of each layer in the multi-layer chicken cage are always maintained within an appropriate range, providing a comfortable and healthy growth environment for the chicken flock. At the same time, the system control center also has data storage and analysis functions, and can record and analyze data such as historical environmental data, chicken flock behavior data, and the operating status of the control system, providing decision-making support for breeding managers.

[0052] The present invention will be further described below in conjunction with Embodiments 1 to 3:

[0053] Embodiment 1:

[0054] This embodiment details how the environmental prediction module uses the Long Short-Term Memory (LSTM) algorithm to construct a chicken coop environmental prediction model. The following is the specific implementation method of this module:

[0055] The training steps of the chicken coop environmental prediction model include:

[0056] (1) Collect training data: Collect historical environmental data, chicken group behavior data, and corresponding actual temperature and humidity data inside the chicken coop from a multi-layer chicken coop breeding environment. The historical environmental data includes, but is not limited to, the temperature, humidity, ventilation status of each layer, and external climate information (such as outdoor temperature, humidity, wind speed, etc.). The chicken group behavior data includes the activity level, drinking frequency, feed consumption of the chicken group on each layer, etc. The actual temperature and humidity data is used as the label for model training to verify the prediction accuracy of the model.

[0057] Suppose there is a three-layer chicken coop breeding system, and each layer is equipped with temperature and humidity sensors, while recording the activity level and feed consumption of the chicken group. The system continuously collects data for one month, including hourly temperature and humidity readings, the number of activities of the chicken group, and daily feed consumption.

[0058] (2) Data preprocessing: Preprocess the collected training data to ensure the quality and consistency of the data. The preprocessing steps include data cleaning and data standardization.

[0059] Data cleaning: For missing values, use the method of averaging the previous and next values for filling. For example, if the humidity data for a certain hour is missing, take the average of the humidity values of the two hours before and after this hour as the filling value. For outliers, use the Isolation Forest algorithm to detect outliers. Isolation Forest is a tree-based unsupervised outlier detection algorithm that can effectively identify outliers in the data. After detecting outliers, replace them with the median of this feature. For example, if the temperature reading for a certain hour significantly deviates from the normal range, it is marked as an outlier and replaced with the median of the temperature readings for this month.

[0060] Data standardization: Standardize all features so that they have the same scale. This is usually achieved by subtracting the mean of the feature and dividing by its standard deviation. Standardization helps to accelerate the convergence speed of the model and improve the prediction accuracy of the model.

[0061] (3) Construct an initial model and set hyperparameters: Use the LSTM algorithm to construct the initial structure of the chicken coop environmental prediction model. LSTM is a neural network algorithm suitable for processing and predicting time series data, which can capture long-term dependencies in the data. When constructing the model, a series of hyperparameters need to be set, such as the number of LSTM layers, the number of neurons in each layer, the learning rate, the batch size, etc. The selection of these hyperparameters has an important impact on the performance of the model.

[0062] (4) Model training: Use the preprocessed dataset to train the initial model. During the training process, adjust the model parameters through an iterative optimization algorithm (such as the Adam optimizer) to minimize the error between the predicted value and the actual value. The goal is to enable the model to accurately predict the temperature and humidity conditions in the chicken coop within a future period of time (such as the next 24 hours). During the training process, the holdout method is used to divide the dataset. That is, the dataset is divided into a training set, a validation set, and a test set. The training set is used to train the model, the validation set is used to evaluate the performance of the model during training, and select the model with the best performance. The test set is used to finally evaluate the model after training ends.

[0063] (5) Model selection and validation: During the training process, select the model with the best performance by monitoring the error on the validation set. Once the performance of the model on the validation set reaches a satisfactory level, the training can be stopped, and this model is selected as the final chicken coop environment prediction model.

[0064] During the training process, it is found that when the number of LSTM layers is 2, the number of neurons in each layer is 128, and the learning rate is 0.001, the prediction accuracy of the model on the validation set is the highest. Therefore, this configuration is selected as the final model parameters.

[0065] Example 2:

[0066] The control decision-making module constructs a temperature and humidity control decision-making model using the reinforcement learning Q-learning algorithm. The specific steps include:

[0067] S1: Define the state space, action space, and reward function of the Q-learning algorithm

[0068] State space: The state space consists of the future temperature and humidity prediction values output by the chicken coop environment prediction model. For example, the state can be represented as a tuple (temperature prediction value, humidity prediction value). For multi-layer chicken coops, each layer has its corresponding state.

[0069] Action space: The action space contains possible temperature and humidity control instructions. These instructions can be to adjust the ventilation rate, the working power of heating or cooling equipment, the operating state of humidifying or dehumidifying equipment, etc. For example, the action can be "increase the ventilation rate by 10%", "turn on the heater", or "turn off the humidifier", etc.

[0070] Reward function: The reward function is set according to the temperature and humidity control effect. The reward can be positive (indicating good control effect) or negative (indicating bad control effect). For example, if the actual temperature and humidity in the chicken coop are closer to the set target temperature and humidity after performing an action, a positive reward is given; conversely, if the temperature and humidity deviate from the target, a negative reward is given. The design of the reward function needs to consider the comfort and growth efficiency of the chicken flock.

[0071] Suppose there is a two-layer chicken coop system, and each layer has a temperature sensor and a humidity sensor. The state space can be defined as the predicted values of temperature and humidity for the next 24 hours for each layer. The action space can include "increase / decrease ventilation rate", "turn on / off heater", and "turn on / off humidifier". The reward function can be set according to the difference between the actual temperature and humidity and the target temperature and humidity. The smaller the difference, the greater the reward.

[0072] S2: Initialize the Q-table and set the hyperparameters of the Q-learning algorithm

[0073] Q-table: The Q-table is a two-dimensional table where the rows represent states and the columns represent actions. The value in each cell represents the expected long-term reward for performing a specific action in a specific state.

[0074] Hyperparameters: Set the learning rate, discount factor, and exploration rate of the Q-learning algorithm.

[0075] Learning rate: Determines the speed at which new information overrides old information. The higher the learning rate, the faster the model adapts to the new environment, but it may also cause the model to be unstable.

[0076] Discount factor: Represents the importance of future rewards in the current decision. The closer the discount factor is to 1, the more important future rewards are.

[0077] Exploration rate: Determines the balance between the model's exploration (trying new actions) and exploitation (performing the known best action). The higher the exploration rate, the more the model tends to try new actions.

[0078] S3: During the training process, select an action according to the current state, observe the new state and reward after performing the action, and update the Q value of the corresponding state-action pair in the Q-table

[0079] Select action: Select an action according to the current state and exploration rate. You can select the action with the highest Q value (exploitation), or randomly select an action (exploration).

[0080] Execute action: Execute the selected action in the simulated environment or the actual chicken coop system, and observe the new state and reward.

[0081] Update the Q value: According to the update formula of Q-learning, update the Q value of the corresponding state-action pair in the Q-table. The update formula is: Q(s,a) = Q(s,a) + α * [r + γ * max(Q(s',a')) - Q(s,a)], where s is the current state, a is the selected action, r is the reward obtained after executing the action, s' is the new state, α is the learning rate, and γ is the discount factor.

[0082] S4: Repeat S3 until the Q-table converges or reaches the preset number of training rounds to obtain the trained temperature and humidity control decision-making model.

[0083] Convergence judgment: The sign that the Q-table converges is that the Q value changes very little in consecutive multiple iterations or reaches the preset convergence threshold.

[0084] Number of training rounds: If the Q-table does not converge within the preset number of training rounds, it is also possible to stop training and use the current Q-table as the temperature and humidity control decision-making model.

[0085] Embodiment 3:

[0086] The instruction optimization module uses the support vector machine (SVM) algorithm to construct a deviation correction model. The following is the specific implementation method of this module:

[0087] The training steps of the deviation correction model include:

[0088] (1) Collect training data: Collect the difference data between the predicted temperature and humidity output by the chicken coop environment prediction model and the actual temperature and humidity monitored by the sensor in real time. These data reflect the deviation between the prediction model and the actual environment. At the same time, in order to train the deviation correction model, it is also necessary to collect the optimal temperature and humidity control instructions marked by experts under the given difference between the predicted temperature and humidity and the actual temperature and humidity. These optimal instructions are used as the labels of the training data to guide the model to learn how to make the best control decisions under different deviation situations.

[0089] Suppose in a three-layer chicken coop system, each layer is equipped with temperature and humidity sensors, and an environment prediction model regularly outputs the predicted temperature and humidity values for a period of time in the future. By comparing the predicted values with the actual sensor monitoring values, the temperature and humidity difference data of each layer can be obtained. At the same time, invite breeding experts to mark the temperature and humidity control instructions under the optimal control strategy according to these difference data, such as "increase the ventilation volume by 20%", "turn on the heater to medium gear", etc.

[0090] (2) Construct the Support Vector Machine (SVM) algorithm model: Select the Support Vector Machine (SVM) as the algorithm basis for the deviation correction model because it is good at handling classification and regression problems of small samples, non - linearity, and high - dimensional spaces. When constructing the SVM model, a series of parameters need to be set, including the kernel function type, penalty parameter, and kernel parameter. Specifically as follows:

[0091] Kernel function type: Select an appropriate kernel function according to the characteristics of the data, such as linear kernel, polynomial kernel, or Radial Basis Function (RBF) kernel. In the temperature and humidity control problem, since there may be non - linear relationships in the data, the RBF kernel is a common choice.

[0092] Penalty parameter: Used to control the degree of penalty for training errors. The larger the penalty parameter, the higher the fitting degree of the model to the training data, but it may lead to overfitting.

[0093] Kernel parameter: Such as the γ parameter of the RBF kernel, which affects the complexity and generalization ability of the model.

[0094] (3) Train the SVM model using the training data: Input the collected training data (including temperature and humidity differences and corresponding optimal control instructions) into the SVM model. Through iterative learning, the model can accurately predict the optimal temperature and humidity control instructions under the given input differences. During the training process, the model will continuously adjust its parameters to minimize the difference between the predicted instructions and the actual optimal instructions.

[0095] (4) After training, obtain the deviation correction model: After sufficient training, the SVM model will be able to output the corresponding optimal temperature and humidity control instructions based on the input temperature and humidity differences. This model is the required deviation correction model, which can provide accurate control instruction suggestions in practical applications when there is a deviation between the predicted temperature and humidity and the actual temperature and humidity.

[0096] To determine whether the deviation between the actual temperature and humidity and the predicted temperature and humidity is significant, the following method is adopted:

[0097] Calculate the absolute difference between the actual temperature and humidity and the predicted temperature and humidity: For each monitoring point, calculate the absolute difference between the actual temperature and humidity value and the predicted temperature and humidity value.

[0098] Set the difference threshold: According to breeding experience and actual needs, set a reasonable difference threshold. This threshold is used to determine whether the deviation has reached the level that requires corrective measures.

[0099] Compare the absolute difference with the threshold: Compare the calculated absolute difference with the set difference threshold. If the absolute difference exceeds the set threshold, it is determined as a significant deviation, and the deviation correction model needs to be triggered to output the corresponding control instructions.

[0100] The present invention further includes a multi-layer intelligent chicken coop based on environmental feedback, which includes a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the above-mentioned multi-layer intelligent chicken coop temperature and humidity control system is implemented.

[0101] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or device.

[0102] Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A multi-layer intelligent chicken cage temperature and humidity control system based on environmental feedback, characterized in that: The system comprises: The environmental prediction module is used to build a chicken cage environmental prediction model. The model predicts the temperature and humidity conditions of each layer in the multi-layer chicken cage in the future based on the input multi-layer chicken cage historical environmental data and chicken flock behavior data, and outputs the future temperature and humidity prediction values ​​of each layer; among which, the historical environmental data includes the historical temperature, humidity, ventilation conditions and external climate information of each layer in the multi-layer chicken cage, and the chicken flock behavior data includes the activity level, drinking water and feed consumption of the chickens on each layer; The control decision module is used to build a temperature and humidity control decision model, which is used to determine the temperature and humidity adjustment strategy of each layer in the multi-layer chicken cage. Its input data is the future temperature and humidity prediction value of each layer of the multi-layer chicken cage output by the environmental prediction module, and its output data is the preliminary temperature and humidity control instructions for each layer; The instruction optimization module is used to build a deviation correction model. The model optimizes the preliminary temperature and humidity control instructions for each layer output by the control decision module according to the difference between the predicted temperature and humidity of each layer of the multi-layer chicken cage output by the environmental prediction module and the actual temperature and humidity of each layer obtained by real-time monitoring of the sensor, and outputs the optimized final temperature and humidity control instructions for each layer; And the system control center is used to combine the actual historical environmental data of multi-layer chicken cages and the actual chicken behavior data, generate preliminary temperature and humidity control instructions for each layer of the multi-layer chicken cage through the environmental prediction module and the control decision module to adjust the environment of each layer in the multi-layer chicken cage, continuously monitor the actual temperature and humidity of each layer in the multi-layer chicken cage, and compare them with the prediction results of the environmental prediction module. If a significant deviation is found, the deviation correction model will be immediately triggered to recalculate the final temperature and humidity control instructions for each layer.

2. The multi-layer intelligent chicken cage temperature and humidity control system based on environmental feedback according to claim 1 is characterized in that: The environmental prediction module uses a long short-term memory network (LSTM) algorithm to build a chicken cage environment prediction model.

3. The multi-layer intelligent chicken cage temperature and humidity control system based on environmental feedback according to claim 2 is characterized in that: The training steps of the chicken cage environment prediction model include: Collect training data, including historical environmental data of multi-layer chicken cages, chicken behavior data, and actual temperature and humidity data in the corresponding chicken cages; Preprocess the collected training data, including data cleaning and data standardization; The initial structure of the chicken cage environment prediction model was constructed using the long short-term memory network (LSTM) algorithm, and the hyperparameters of the model were set; The initial model is trained using the preprocessed data set, and the model parameters are adjusted through an iterative optimization algorithm so that the model can accurately predict the temperature and humidity conditions in the chicken coop in the future. The labels of the training data are the actual temperature and humidity data in the chicken coop. During the training process, the holdout method was used to partition the data set to verify the model performance, and the model with the best performance was selected as the final chicken cage environment prediction model.

4. The multi-layer intelligent chicken cage temperature and humidity control system based on environmental feedback according to claim 3 is characterized in that: The data cleaning method comprises: For missing values, the average of the previous and next values ​​was used to fill them; For outliers, the isolation forest algorithm is used to detect outliers and replace them with the median.

5. The multi-layer intelligent chicken cage temperature and humidity control system based on environmental feedback according to claim 1 is characterized in that: The control decision module uses the reinforcement learning Q-learning algorithm to build a temperature and humidity control decision model.

6. The multi-layer intelligent chicken cage temperature and humidity control system based on environmental feedback according to claim 5 is characterized in that: The training steps of the temperature and humidity control decision model include: S1: Define the state space, action space and reward function of the Q-learning algorithm; the state space is the future temperature and humidity prediction value output by the chicken cage environment prediction model, the action space is the possible temperature and humidity control instructions, and the reward function is set according to the temperature and humidity control effect; S2: Initialize the Q table and set the hyperparameters of the Q-learning algorithm, including the learning rate, discount factor, and exploration rate; S3: During training, select an action based on the current state, observe the new state and reward after executing the action, and update the Q value of the corresponding state-action pair in the Q table; S4: Repeat S3 until the Q table converges or reaches a preset training round, and obtain a trained temperature and humidity control decision model.

7. The multi-layer intelligent chicken cage temperature and humidity control system based on environmental feedback according to claim 1 is characterized in that: The instruction optimization module uses a support vector machine (SVM) algorithm to construct a deviation correction model.

8. The multi-layer intelligent chicken cage temperature and humidity control system based on environmental feedback according to claim 7 is characterized in that: The training steps of the bias correction model include: Collect training data, including the difference between the predicted temperature and humidity output by the chicken cage environment prediction model and the actual temperature and humidity obtained by real-time monitoring by the sensor; at the same time, collect the optimal temperature and humidity control instructions marked by experts under the given difference between the predicted temperature and humidity and the actual temperature and humidity as the labels of the training data; Build a support vector machine (SVM) algorithm model and set SVM parameters, including kernel function type, penalty parameters, and kernel parameters; Use the training data to train the support vector machine (SVM) algorithm model. Through iterative learning, the model can accurately predict the optimal temperature and humidity control instructions under given input differences. After the training is completed, the bias correction model is obtained.

9. The multi-layer intelligent chicken cage temperature and humidity control system based on environmental feedback according to claim 1 is characterized in that: The method for determining whether the deviation between the actual temperature and humidity and the predicted temperature and humidity is significant is as follows: calculating the absolute difference between the actual temperature and humidity and the predicted temperature and humidity, setting a difference threshold, comparing the calculated absolute difference with the set difference threshold, and if the absolute difference exceeds the set threshold, determining it as a significant deviation.

10. A multi-layer intelligent chicken cage based on environmental feedback, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the multi-layer intelligent chicken cage temperature and humidity control system according to any one of claims 1 to 9 is implemented.

Images