Livestock breeding constant temperature control method and system

Abstract

This invention relates to the field of livestock breeding technology and discloses a method and system for constant temperature control in livestock breeding. The method includes: extracting environmental characteristic data; performing evolution trend analysis on the environmental characteristic data to predict the dynamic change trend of the breeding environment under different control conditions; identifying key factor data affecting constant temperature through pattern matching; matching and executing corresponding preset control strategies based on the key factor data; and adjusting the operating status of the environmental control equipment through real-time feedback. This invention, through environmental trend analysis, can predict the changing trends under different control conditions in advance, thereby matching and executing targeted control strategies, providing reliable data-driven decision support for constant temperature control in livestock breeding environments.

Description

Technical Field

[0001] This invention relates to the field of livestock breeding technology, and more specifically, to a method and system for constant temperature control in livestock breeding. Background Technology

[0002] In recent years, the livestock farming industry has rapidly developed towards large-scale and intensive operations, with the number and scale of livestock increasing continuously. The importance of farming efficiency and quality control has become increasingly prominent. Among these factors, the temperature of the livestock sheds is a key environmental factor affecting livestock growth, development, physiological health, and survival rates. This is especially true in northern regions where winters are extremely cold. Inadequate insulation measures can easily lead to excessively low temperatures in the sheds, which is particularly detrimental to livestock, especially young animals with weaker immune systems. This not only causes a significant increase in morbidity among young animals but also directly impacts the economic benefits and sustainable development of the livestock industry. Therefore, there is an urgent need for a precise and stable temperature control method adapted to large-scale farming scenarios to ensure reasonable temperature regulation in the farming environment.

[0003] However, existing technologies rely heavily on human intervention and lack the ability to proactively predict environmental changes. They are difficult to regulate before temperature anomalies occur and are not easy to accurately pinpoint the key factors causing temperature fluctuations. This results in delayed regulatory responses, crude strategies, and an inability to adapt to the dynamic needs of the aquaculture environment.

[0004] No effective solutions have yet been proposed to address the problems in the relevant technologies. Summary of the Invention

[0005] In view of the problems in related technologies, this invention proposes a method and system for constant temperature control in livestock breeding, so as to overcome the above-mentioned technical problems existing in the existing related technologies.

[0006] Therefore, the specific technical solution adopted by the present invention is as follows: According to one aspect of the present invention, a method for constant temperature control in livestock farming is provided, comprising: S1. Obtain real-time environmental data of the aquaculture environment, and preprocess the real-time environmental data to extract environmental feature data; S2. Analyze the evolution trend of environmental characteristic data to predict the dynamic change trend of the aquaculture environment under different control conditions; S3. Based on dynamic change trends, a stability and anomaly analysis subspace is constructed by extracting stable features, key change patterns are separated and their contribution is quantified, state representation is generated by feature mapping, and key factor data affecting environmental constant temperature are identified by pattern matching. S4. Based on key factor data, match and execute corresponding preset control strategies, and adjust the operating status of environmental control equipment through real-time feedback to achieve constant temperature control of the breeding environment.

[0007] Furthermore, real-time environmental data of the aquaculture environment is acquired, and the real-time environmental data is preprocessed to extract environmental feature data, including: S11. Perform noise reduction, filtering, and smoothing on duplicate data, missing values, and outliers in the real-time environmental data to obtain standardized real-time environmental data. S12. Reorganize and align the standardized real-time environmental data according to the monitoring location and time dimension to construct a multi-dimensional environmental status dataset; S13. Using each current environmental state point in the multidimensional environmental state dataset as a reference point, calculate the environmental feature correlation between the current environmental state point and other environmental state points. S14. Based on a preset feature similarity threshold, select the environmental state points with the highest similarity to the benchmark point to form a local environmental feature set. S15. Extract the most representative feature combination from the local environmental feature set, and finally fuse them to generate environmental feature data.

[0008] Furthermore, an evolution trend analysis of environmental characteristic data is conducted to predict the dynamic change trends of the aquaculture environment under different control conditions, including: S21. Construct an environmental state evolution network based on environmental feature data. Each network node corresponds to the aquaculture environment state under preset control conditions at a certain moment. Initially, all network node states are marked as unclassified, and the overlapping relationships between network nodes are marked as non-overlapping. S22. Calculate the ratio of environmental stability to regulation response based on the network node status, and use this ratio as an indicator to select the most representative network node status as the central network node status. Assign the remaining nodes whose status differs from the central network node status by less than a set threshold to the same cluster to form an environmental evolution pattern cluster. S23. Match the current environmental feature data with the environmental evolution pattern cluster, match the evolution pattern to which the current environmental state belongs, and extract the complete state transition change law of the evolution pattern under the same historical control conditions. S24. Based on the matching evolution pattern and the corresponding state transition change law, establish a time series prediction model to predict the dynamic change trend of the aquaculture environment under different control conditions.

[0009] Furthermore, based on the matching evolution patterns and corresponding state transition change laws, a time-series prediction model is established to predict the dynamic change trends of the aquaculture environment under different regulatory conditions, including: S241. Obtain the environmental characteristic time series corresponding to the historical time period and the target control condition covariate series corresponding to the future target time period, respectively. S242. Input the historical environmental feature sequence into the encoding network, and encode the environmental features of each moment in the sequence into a high-dimensional feature vector through the encoding network to generate an environmental evolution feature representation sequence that reflects the historical dynamic evolution law. S243. Based on the attention mechanism, the control variables at each time step in the covariate sequence of the target control conditions corresponding to the future target time period are fused with the feature representations of the corresponding time in the historical environmental evolution feature representation sequence to generate a future environmental feature sequence with fused control information. S244. Input the future environmental feature sequence into the prediction network for time series modeling and multi-step forward extrapolation, and output the predicted value of the dynamic change trend of the aquaculture environment under different control conditions in the future target period to form a visualized trend map.

[0010] Furthermore, based on the attention mechanism, the control variables at each time step in the target control condition covariate sequence corresponding to the future target time period are fused with the feature representations at the corresponding time in the historical environmental evolution feature representation sequence to generate a future environmental feature sequence with fused control information, including: S2431. Using the sequence of historical environmental evolution characteristics as keys and values, and the sequence of covariates of future target regulation conditions as queries, calculate the feature similarity between the query vector of each future time step and the key vector of all historical time steps, and generate a cross-time period attention weight matrix. S2432. Use the attention weight matrix to perform a weighted summation on the sequence of historical environment evolution feature representations to obtain the attention-weighted environment feature representation corresponding to each future time step; S2433. The regulation condition covariate of each future time step is vector-concatenated with its corresponding attention-weighted environmental feature representation to form a preliminary fused joint feature representation of the regulation environment. S2434. Arrange the joint feature representations of the regulatory environment for all future time steps in chronological order, and perform feature transformation and normalization through a feedforward network to output a future environment feature sequence that integrates regulatory information.

[0011] Furthermore, based on dynamic change trends, a stability and anomaly analysis subspace is constructed through stable feature extraction, key change patterns are separated and their contribution is quantified, state representations are generated using feature mapping, and key factors affecting environmental isothermal conditions are identified through pattern matching, including: S31. Analyze the dynamic change trend and calculate the characteristic components corresponding to the unchanging direction as the basic characteristics of the steady state. S32. Construct a feature set based on the basic features of the stable state, and establish a trend stability analysis matrix based on the feature set. Obtain the structural composition of the stability subspace and the anomaly subspace through feature decomposition. S33. Construct a trend component separation matrix, extract stable change patterns and abnormal fluctuation patterns through feature decomposition, and calculate the contribution of stable change patterns and abnormal fluctuation patterns to trend changes respectively. S34. Select the stable change pattern and the abnormal fluctuation pattern with the greatest contribution, construct feature mapping rules, and convert the dynamic change trend into a feature representation that distinguishes between stable and abnormal states through the feature mapping rules. S35. By analyzing the matching degree between the feature representation and the known state pattern features, the feature pattern with the highest correlation to the current dynamic change trend is identified as the key factor data affecting the constant temperature of the breeding environment.

[0012] Furthermore, the stable change patterns and abnormal fluctuation patterns with the greatest contribution are selected to construct feature mapping rules. The dynamic change trends are then transformed into feature representations that distinguish between stable and abnormal states through these rules, including: S341. Construct a key pattern basis vector set for state discrimination by using the stable change pattern and the abnormal fluctuation pattern that contribute the most. S342. Based on the key pattern basis vector set, construct a linear projection mapping rule from the original dynamic change trend to the low-dimensional feature space; S343. Project the dynamic trend data using linear projection mapping rules to generate feature representations that can clearly distinguish between stable and abnormal states.

[0013] Furthermore, the dynamically changing trend data is projected and transformed using linear projection mapping rules to generate feature representations that can clearly distinguish between stable and abnormal states, including: S3431. Standardize the dynamic change trend and spatially align it with the key mode basis vector group to ensure the consistency of projection calculation. S3432. According to the linear projection mapping rule, calculate the projection coefficients of the standardized dynamic change trend on each key mode basis vector to form the preliminary projection feature vector. S3433. Normalize and enhance the projected feature vectors to generate low-dimensional feature representations, so as to clearly distinguish the feature representations of stable and abnormal states.

[0014] Furthermore, based on the linear projection mapping rule, the projection coefficients of the standardized dynamic change trend on each key mode basis vector are calculated to form the preliminary projection feature vector, including: S34321. Calculate the inner product of the standardized dynamic change trend and each key mode basis vector respectively to obtain the projection component values ​​of the standardized dynamic change trend in each basis vector direction. S34322. Arrange all projection component values ​​in the order of their corresponding basis vectors to form a projection coefficient vector that represents the distribution of the trend in the key mode space. S34323. Confirm the dimensions and standardize the format of the projection coefficient vector to form a preliminary projection feature vector.

[0015] According to another aspect of the present invention, a livestock breeding temperature control system is also provided, the system comprising: The data acquisition module is used to acquire real-time environmental data of the aquaculture environment and preprocess the real-time environmental data to extract environmental feature data. The trend analysis module is used to analyze the evolution trend of environmental characteristic data and predict the dynamic change trend of the aquaculture environment under different control conditions. The anomaly analysis module is used to construct a stability and anomaly analysis subspace based on dynamic change trends through stable feature extraction, separate key change patterns and quantify their contribution, generate state representations using feature mapping, and identify key factor data affecting environmental constant temperature through pattern matching. The constant temperature control module is used to match and execute corresponding preset control strategies based on key factor data, and adjust the operating status of environmental control equipment through real-time feedback to achieve constant temperature control of the breeding environment.

[0016] The beneficial effects of this invention are as follows: 1. This invention can predict the changing trends under different control conditions in advance through environmental trend analysis. Based on the feature mapping mechanism that separates stable and abnormal modes, it can accurately identify the key factors that cause temperature fluctuations, thereby matching and executing targeted control strategies, improving the temperature control response speed and accuracy, enhancing the system's adaptability and stability under complex working conditions, and providing reliable data-driven decision support for constant temperature control in aquaculture environments.

[0017] 2. By constructing an environmental state evolution network and a pattern matching mechanism, this invention can accurately identify the current environmental evolution mode and extract its state transition rules. Based on the temporal modeling method of encoding network and attention fusion, it realizes multi-step accurate prediction of environmental change trends under different control conditions, thereby improving the predictability and strategy adaptability of constant temperature control in aquaculture.

[0018] 3. This invention constructs a stability and anomaly analysis subspace, accurately separates key change patterns and quantifies their contributions, generates feature representations that can clearly distinguish between stable and abnormal states based on linear projection and feature enhancement, and accurately identifies the core factors affecting isothermal conditions through pattern matching, thereby improving the accuracy of state diagnosis and the reliability of tracing the source of key factors. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a flowchart of a livestock breeding constant temperature control method according to an embodiment of the present invention; Figure 2 This is a schematic diagram of a livestock breeding constant temperature control system according to an embodiment of the present invention.

[0021] In the picture: 1. Data acquisition module; 2. Trend analysis module; 3. Anomaly analysis module; 4. Constant temperature control module. Detailed Implementation

[0022] To further illustrate the various embodiments, the present invention provides accompanying drawings, which are part of the disclosure of the present invention. These drawings are mainly used to illustrate the embodiments and can be used in conjunction with the relevant descriptions in the specification to explain the operating principles of the embodiments. With reference to these drawings, those skilled in the art should be able to understand other possible implementation methods and the advantages of the present invention.

[0023] According to an embodiment of the present invention, a method and system for constant temperature control in livestock farming are provided.

[0024] The present invention will now be further described in conjunction with the accompanying drawings and specific embodiments, such as... Figure 1 As shown, the livestock breeding temperature control method according to an embodiment of the present invention includes: S1. Obtain real-time environmental data of the aquaculture environment, and preprocess the real-time environmental data to extract environmental feature data; Specifically, environmental data includes temperature, humidity, air quality parameters, ventilation status, heating / cooling equipment status, light intensity and duration, etc.

[0025] Specifically, environmental characteristic data includes temperature characteristics, temperature and humidity coupling characteristics, equipment correlation characteristics, load correlation characteristics, etc.

[0026] S2. Analyze the evolution trend of environmental characteristic data to predict the dynamic change trend of the aquaculture environment under different control conditions; S3. Based on dynamic change trends, a stability and anomaly analysis subspace is constructed by extracting stable features, key change patterns are separated and their contribution is quantified, state representation is generated by feature mapping, and key factor data affecting environmental constant temperature are identified by pattern matching. Specifically, key factor data includes environmental disturbance factor data, equipment operation factor data, aquaculture load factor data, spatial distribution factor data, etc.

[0027] S4. Based on key factor data, match and execute corresponding preset control strategies, and adjust the operating status of environmental control equipment through real-time feedback to achieve constant temperature control of the breeding environment.

[0028] Specifically, based on key factor data, the system matches corresponding equipment control schemes in a preset control strategy library, generating an instruction set containing specific parameters and execution timing. According to this instruction set, environmental control equipment such as fans, heaters, and evaporative cooling pads are controlled collaboratively, with speed, power, and start-stop timing adjusted synchronously. During execution, the system monitors environmental changes in real time through sensors, dynamically adjusting equipment output using a combination of feedforward prediction and feedback correction to achieve adaptive closed-loop constant temperature control. After reaching the target temperature zone, the system automatically switches to steady-state maintenance mode and records control process data to continuously optimize the strategy library, thereby improving control accuracy and energy efficiency under different operating conditions.

[0029] In this optional embodiment, real-time environmental data of the aquaculture environment is acquired, and the real-time environmental data is preprocessed to extract environmental feature data, including: S11. Perform noise reduction, filtering, and smoothing on duplicate data, missing values, and outliers in the real-time environmental data to obtain standardized real-time environmental data. S12. Reorganize and align the standardized real-time environmental data according to the monitoring location and time dimension to construct a multi-dimensional environmental status dataset; S13. Using each current environmental state point in the multidimensional environmental state dataset as a reference point, calculate the environmental feature correlation between the current environmental state point and other environmental state points. S14. Based on a preset feature similarity threshold, select the environmental state points with the highest similarity to the benchmark point to form a local environmental feature set. S15. Extract the most representative feature combination from the local environmental feature set, and finally fuse them to generate environmental feature data.

[0030] Specifically, this invention extracts environmental feature data through a grouping algorithm, which is a data grouping and feature extraction method based on feature similarity. In this invention, the algorithm constructs a multidimensional environmental state dataset, calculates the correlation of environmental features point by point, filters local environmental feature sets according to a preset similarity threshold, and extracts the most representative feature combinations from each set, thereby achieving dimensionality reduction and feature enhancement of the original monitoring data of the aquaculture environment, providing high-quality input for subsequent analysis.

[0031] In this optional embodiment, analyzing the evolution trend of environmental characteristic data and predicting the dynamic change trend of the aquaculture environment under different control conditions includes: S21. Construct an environmental state evolution network based on environmental feature data. Each network node corresponds to the aquaculture environment state under preset control conditions at a certain moment. Initially, all network node states are marked as unclassified, and the overlapping relationships between network nodes are marked as non-overlapping. S22. Calculate the ratio of environmental stability to regulation response based on the network node status, and use this ratio as an indicator to select the most representative network node status as the central network node status. Assign the remaining nodes whose status differs from the central network node status by less than a set threshold to the same cluster to form an environmental evolution pattern cluster. S23. Match the current environmental feature data with the environmental evolution pattern cluster, match the evolution pattern to which the current environmental state belongs, and extract the complete state transition change law of the evolution pattern under the same historical control conditions. S24. Based on the matching evolution pattern and the corresponding state transition change law, establish a time series prediction model to predict the dynamic change trend of the aquaculture environment under different control conditions.

[0032] Specifically, this invention uses an overlapping box covering algorithm to analyze the evolution trend of environmental feature data. The overlapping box covering algorithm is a pattern partitioning method based on network topology and similarity clustering. In this invention, the algorithm constructs and initializes a state node network based on environmental features. Then, it selects the central node by calculating the ratio of node environmental stability to regulatory response, and groups neighboring nodes into the same cluster based on a state difference threshold, thus forming environmental pattern clusters reflecting different evolution patterns. Historical state transition patterns are extracted through cluster matching, and a time-series model is established to predict the dynamic trend of the aquaculture environment under multiple regulatory conditions.

[0033] In this optional embodiment, based on the matching evolution pattern and the corresponding state transition change law, a time series prediction model is established to predict the dynamic change trend of the aquaculture environment under different control conditions, including: S241. Obtain the environmental characteristic time series corresponding to the historical time period and the target control condition covariate series corresponding to the future target time period, respectively. S242. Input the historical environmental feature sequence into the encoding network, and encode the environmental features of each moment in the sequence into a high-dimensional feature vector through the encoding network to generate an environmental evolution feature representation sequence that reflects the historical dynamic evolution law. S243. Based on the attention mechanism, the control variables at each time step in the covariate sequence of the target control conditions corresponding to the future target time period are fused with the feature representations of the corresponding time in the historical environmental evolution feature representation sequence to generate a future environmental feature sequence with fused control information. S244. Input the future environmental feature sequence into the prediction network for time series modeling and multi-step forward extrapolation, and output the predicted value of the dynamic change trend of the aquaculture environment under different control conditions in the future target period to form a visualized trend map.

[0034] Specifically, the system acquires records of temperature, humidity, and ventilation equipment operation in the farm to create a historical environmental characteristic time series. Simultaneously, it pre-determines planned fan speeds and heating switch sequences for future moments, forming a target control condition covariate sequence. An encoding network encodes the historical data into high-dimensional feature vectors moment-by-moment, generating an environmental evolution feature representation sequence. Based on an attention mechanism, the system fuses the control variables for each future moment with their corresponding historical features, forming a future environmental characteristic sequence with integrated control information. After receiving this sequence, the prediction network performs multi-step forward extrapolation using a time-series prediction model, outputting predicted dynamic changes in key environmental indicators under different future control conditions, and generating intuitive trend graphs for management reference.

[0035] Specifically, the time-series prediction model employs an encoder-decoder architecture. The encoder is a bidirectional LSTM used to encode historical environmental feature sequences into hidden state sequences. The decoder is an LSTM with an attention mechanism, whose input at each step integrates the hidden state from the previous step, the predicted output from the previous step, and the corresponding future regulatory condition covariates. The attention mechanism dynamically focuses on key historical information from the encoder's hidden states, thereby progressively generating a multi-step environmental state prediction sequence. The model is trained on a dataset constructed using a historical sliding window, employing a smooth L1 loss function and the AdamW optimizer. Training is conducted by gradually increasing the prediction step size from easy to difficult using a learning strategy. After training, the model receives real-time environmental feature sequences and the proposed future regulatory plan sequence, outputting corresponding multi-step environmental change prediction values, providing a quantitative basis for simulating the effects of different regulatory strategies and making optimization decisions.

[0036] In this optional embodiment, based on an attention mechanism, the control variables at each time step in the target control condition covariate sequence corresponding to the future target time period are fused with the feature representations at the corresponding time in the historical environmental evolution feature representation sequence to generate a future environmental feature sequence with fused control information, including: S2431. Using the sequence of historical environmental evolution characteristics as keys and values, and the sequence of covariates of future target regulation conditions as queries, calculate the feature similarity between the query vector of each future time step and the key vector of all historical time steps, and generate a cross-time period attention weight matrix. S2432. Use the attention weight matrix to perform a weighted summation on the sequence of historical environment evolution feature representations to obtain the attention-weighted environment feature representation corresponding to each future time step; S2433. The regulation condition covariate of each future time step is vector-concatenated with its corresponding attention-weighted environmental feature representation to form a preliminary fused joint feature representation of the regulation environment. S2434. Arrange the joint feature representations of the regulatory environment for all future time steps in chronological order, and perform feature transformation and normalization through a feedforward network to output a future environment feature sequence that integrates regulatory information.

[0037] In this optional embodiment, based on dynamic change trends, a stability and anomaly analysis subspace is constructed through stable feature extraction, key change patterns are separated and their contribution is quantified, state representations are generated using feature mapping, and key factors affecting environmental constant temperature are identified through pattern matching, including: S31. Analyze the dynamic change trend and calculate the characteristic components corresponding to the unchanging direction as the basic characteristics of the steady state. S32. Construct a feature set based on the basic features of the stable state, and establish a trend stability analysis matrix based on the feature set. Obtain the structural composition of the stability subspace and the anomaly subspace through feature decomposition. S33. Construct a trend component separation matrix, extract stable change patterns and abnormal fluctuation patterns through feature decomposition, and calculate the contribution of stable change patterns and abnormal fluctuation patterns to trend changes respectively. S34. Select the stable change pattern and the abnormal fluctuation pattern with the greatest contribution, construct feature mapping rules, and convert the dynamic change trend into a feature representation that distinguishes between stable and abnormal states through the feature mapping rules. S35. By analyzing the matching degree between the feature representation and the known state pattern features, the feature pattern with the highest correlation to the current dynamic change trend is identified as the key factor data affecting the constant temperature of the breeding environment.

[0038] Specifically, this invention identifies key factors affecting environmental constant temperature using the I-VLDNS algorithm. The I-VLDNS algorithm is an improved linear discriminant subspace pattern recognition algorithm. Its improvements lie in separating steady-state and disturbance components through variational mode decomposition preprocessing to enhance noise resistance, establishing a local-global dual discrimination criterion to improve robustness to mixed gradual and abrupt modes, and introducing elastic net regularization to achieve sparse feature selection to improve the interpretability of the results. In this invention, the algorithm extracts steady-state features from dynamic trends to construct an analysis subspace, separates stable change modes and abnormal fluctuation modes through feature decomposition and quantifies their contribution, selects key modes to construct feature mapping rules to generate feature representations that can distinguish states, and accurately identifies key factors affecting environmental constant temperature through pattern matching.

[0039] In this optional embodiment, the stable change pattern and the abnormal fluctuation pattern with the greatest contribution are selected, feature mapping rules are constructed, and the dynamic change trend is converted into a feature representation that distinguishes between stable and abnormal states through the feature mapping rules, including: S341. Construct a key pattern basis vector set for state discrimination by using the stable change pattern and the abnormal fluctuation pattern that contribute the most. S342. Based on the key pattern basis vector set, construct a linear projection mapping rule from the original dynamic change trend to the low-dimensional feature space; S343. Project the dynamic trend data using linear projection mapping rules to generate feature representations that can clearly distinguish between stable and abnormal states.

[0040] In this optional embodiment, the dynamic trend data is projected and transformed using a linear projection mapping rule to generate a feature representation that can clearly distinguish between stable and abnormal states, including: S3431. Standardize the dynamic change trend and spatially align it with the key mode basis vector group to ensure the consistency of projection calculation. S3432. According to the linear projection mapping rule, calculate the projection coefficients of the standardized dynamic change trend on each key mode basis vector to form the preliminary projection feature vector. S3433. Normalize and enhance the projected feature vectors to generate low-dimensional feature representations, so as to clearly distinguish the feature representations of stable and abnormal states.

[0041] In this optional embodiment, the projection coefficients of the standardized dynamic change trend on each key mode basis vector are calculated according to the linear projection mapping rule to form a preliminary projection feature vector, including: S34321. Calculate the inner product of the standardized dynamic change trend and each key mode basis vector respectively to obtain the projection component values ​​of the standardized dynamic change trend in each basis vector direction. S34322. Arrange all projection component values ​​in the order of their corresponding basis vectors to form a projection coefficient vector that represents the distribution of the trend in the key mode space. S34323. Confirm the dimensions and standardize the format of the projection coefficient vector to form a preliminary projection feature vector.

[0042] According to another embodiment of the invention, such as Figure 2 As shown, a livestock breeding temperature control system is also provided, which includes: Data acquisition module 1 is used to acquire real-time environmental data of the aquaculture environment and preprocess the real-time environmental data to extract environmental feature data; Trend Analysis Module 2 is used to analyze the evolution trend of environmental characteristic data and predict the dynamic change trend of the aquaculture environment under different control conditions. Anomaly analysis module 3 is used to construct a stability and anomaly analysis subspace based on dynamic change trends through stable feature extraction, separate key change patterns and quantify their contribution, generate state representations using feature mapping, and identify key factor data affecting environmental constant temperature through pattern matching. The constant temperature control module 4 is used to match and execute corresponding preset control strategies based on key factor data, and adjust the operating status of environmental control equipment through real-time feedback to achieve constant temperature control of the breeding environment.

[0043] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A livestock breeding thermostatic control method, characterized by, include: S1. Obtain real-time environmental data of the aquaculture environment, and preprocess the real-time environmental data to extract environmental feature data; S2. Analyze the evolution trend of environmental characteristic data to predict the dynamic change trend of the aquaculture environment under different control conditions; S3. Based on dynamic change trends, a stability and anomaly analysis subspace is constructed by extracting stable features, key change patterns are separated and their contribution is quantified, state representation is generated by feature mapping, and key factor data affecting environmental constant temperature are identified by pattern matching. S4. Based on key factor data, match and execute corresponding preset control strategies, and adjust the operating status of environmental control equipment through real-time feedback to achieve constant temperature control of the breeding environment.

2. A livestock breeding thermostatic control method according to claim 1, wherein, The process of acquiring real-time environmental data of the aquaculture environment and preprocessing the real-time environmental data to extract environmental feature data includes: S11. Perform noise reduction, filtering, and smoothing on duplicate data, missing values, and outliers in the real-time environmental data to obtain standardized real-time environmental data. S12. Reorganize and align the standardized real-time environmental data according to the monitoring location and time dimension to construct a multi-dimensional environmental status dataset; S13. Using each current environmental state point in the multidimensional environmental state dataset as a reference point, calculate the environmental feature correlation between the current environmental state point and other environmental state points. S14. Based on a preset feature similarity threshold, select the environmental state points with the highest similarity to the benchmark point to form a local environmental feature set. S15. Extract the most representative feature combination from the local environmental feature set, and finally fuse them to generate environmental feature data.

3. The livestock breeding thermostatic control method according to claim 1, wherein, The analysis of the evolution trend of environmental characteristic data to predict the dynamic change trend of the aquaculture environment under different control conditions includes: S21. Construct an environmental state evolution network based on environmental feature data. Each network node corresponds to the aquaculture environment state under preset control conditions at a certain moment. Initially, all network node states are marked as unclassified, and the overlapping relationships between network nodes are marked as non-overlapping. S22. Calculate the ratio of environmental stability to regulation response based on the network node status, and use this ratio as an indicator to select the most representative network node status as the central network node status. Assign the remaining nodes whose status differs from the central network node status by less than a set threshold to the same cluster to form an environmental evolution pattern cluster. S23. Match the current environmental feature data with the environmental evolution pattern cluster, match the evolution pattern to which the current environmental state belongs, and extract the complete state transition change law of the evolution pattern under the same historical control conditions. S24. Based on the matching evolution pattern and the corresponding state transition change law, establish a time series prediction model to predict the dynamic change trend of the aquaculture environment under different control conditions.

4. A livestock breeding thermostatic control method according to claim 3, wherein, The aforementioned time-series prediction model, based on the matching evolution pattern and corresponding state transition change law, predicts the dynamic change trend of the aquaculture environment under different control conditions, including: S241. Obtain the environmental characteristic time series corresponding to the historical time period and the target control condition covariate series corresponding to the future target time period, respectively. S242. Input the historical environmental feature sequence into the encoding network, and encode the environmental features of each moment in the sequence into a high-dimensional feature vector through the encoding network to generate an environmental evolution feature representation sequence that reflects the historical dynamic evolution law. S243. Based on the attention mechanism, the control variables at each time step in the covariate sequence of the target control conditions corresponding to the future target time period are fused with the feature representations of the corresponding time in the historical environmental evolution feature representation sequence to generate a future environmental feature sequence with fused control information. S244. Input the future environmental feature sequence into the prediction network for time series modeling and multi-step forward extrapolation, and output the predicted value of the dynamic change trend of the aquaculture environment under different control conditions in the future target period to form a visualized trend map.

5. A livestock breeding thermostatic control method according to claim 4, wherein, The attention-based mechanism involves fusing the control variables at each time step in the target control condition covariate sequence corresponding to the future target time period with the feature representations at the corresponding time in the historical environmental evolution feature representation sequence to generate a future environmental feature sequence with fused control information, including: S2431. Using the sequence of historical environmental evolution characteristics as keys and values, and the sequence of covariates of future target regulation conditions as queries, calculate the feature similarity between the query vector of each future time step and the key vector of all historical time steps, and generate a cross-time period attention weight matrix. S2432. Use the attention weight matrix to perform a weighted summation on the sequence of historical environment evolution feature representations to obtain the attention-weighted environment feature representation corresponding to each future time step; S2433. The regulation condition covariate of each future time step is vector-concatenated with its corresponding attention-weighted environmental feature representation to form a preliminary fused joint feature representation of the regulation environment. S2434. Arrange the joint feature representations of the regulatory environment for all future time steps in chronological order, and perform feature transformation and normalization through a feedforward network to output a future environment feature sequence that integrates regulatory information.

6. The livestock breeding thermostatic control method according to claim 1, wherein, Based on dynamic change trends, a stability and anomaly analysis subspace is constructed through stable feature extraction, key change patterns are separated and their contribution is quantified, state representations are generated using feature mapping, and key factors affecting environmental constant temperature are identified through pattern matching, including: S31. Analyze the dynamic change trend and calculate the characteristic components corresponding to the unchanging direction as the basic characteristics of the steady state. S32. Construct a feature set based on the basic features of the stable state, and establish a trend stability analysis matrix based on the feature set. Obtain the structural composition of the stability subspace and the anomaly subspace through feature decomposition. S33. Construct a trend component separation matrix, extract stable change patterns and abnormal fluctuation patterns through feature decomposition, and calculate the contribution of stable change patterns and abnormal fluctuation patterns to trend changes respectively. S34. Select the stable change pattern and the abnormal fluctuation pattern with the greatest contribution, construct feature mapping rules, and convert the dynamic change trend into a feature representation that distinguishes between stable and abnormal states through the feature mapping rules. S35. By analyzing the matching degree between the feature representation and the known state pattern features, the feature pattern with the highest correlation to the current dynamic change trend is identified as the key factor data affecting the constant temperature of the breeding environment.

7. A livestock breeding thermostatic control method according to claim 6, wherein, The process of selecting the stable change pattern and the abnormal fluctuation pattern with the greatest contribution, constructing feature mapping rules, and converting the dynamic change trend into a feature representation that distinguishes between stable and abnormal states through the feature mapping rules includes: S341. Construct a key pattern basis vector set for state discrimination by using the stable change pattern and the abnormal fluctuation pattern that contribute the most. S342. Based on the key pattern basis vector set, construct a linear projection mapping rule from the original dynamic change trend to the low-dimensional feature space; S343. Project the dynamic trend data using linear projection mapping rules to generate feature representations that can clearly distinguish between stable and abnormal states.

8. A livestock breeding thermostatic control method according to claim 7, characterized in that, The step of projecting and transforming dynamically changing trend data using linear projection mapping rules to generate feature representations that can clearly distinguish between stable and abnormal states includes: S3431. Standardize the dynamic change trend and spatially align it with the key mode basis vector group to ensure the consistency of projection calculation. S3432. According to the linear projection mapping rule, calculate the projection coefficients of the standardized dynamic change trend on each key mode basis vector to form the preliminary projection feature vector. S3433. Normalize and enhance the projected feature vectors to generate low-dimensional feature representations, so as to clearly distinguish the feature representations of stable and abnormal states.

9. A livestock breeding thermostatic control method according to claim 8, wherein, The step of calculating the projection coefficients of the standardized dynamic change trend on each key mode basis vector according to the linear projection mapping rule to form the preliminary projection feature vector includes: S34321. Calculate the inner product of the standardized dynamic change trend and each key mode basis vector respectively to obtain the projection component values ​​of the standardized dynamic change trend in each basis vector direction. S34322. Arrange all projection component values ​​in the order of their corresponding basis vectors to form a projection coefficient vector that represents the distribution of the trend in the key mode space. S34323. Confirm the dimensions and standardize the format of the projection coefficient vector to form a preliminary projection feature vector.

10. A livestock breeding thermostatic control system for implementing the livestock breeding thermostatic control method according to any one of claims 1 to 9, characterized by, The system includes: The data acquisition module is used to acquire real-time environmental data of the aquaculture environment and preprocess the real-time environmental data to extract environmental feature data. The trend analysis module is used to analyze the evolution trend of environmental characteristic data and predict the dynamic change trend of the aquaculture environment under different control conditions. The anomaly analysis module is used to construct a stability and anomaly analysis subspace based on dynamic change trends through stable feature extraction, separate key change patterns and quantify their contribution, generate state representations using feature mapping, and identify key factor data affecting environmental constant temperature through pattern matching. The constant temperature control module is used to match and execute corresponding preset control strategies based on key factor data, and adjust the operating status of environmental control equipment through real-time feedback to achieve constant temperature control of the breeding environment.

Images