Intelligent control system and method for agricultural greenhouse
Through the intelligent control system and soil pH prediction model, the pH value of agricultural greenhouse soil is adjusted in real time, which solves the problem of lagging soil pH adjustment, ensures a suitable crop growth environment, and improves crop yields and economic benefits.
Patent Information
- Application Number
- CN202411494294.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-24
- Publication Date
- 2025-01-24
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
There is a lag in the regulation of soil pH in agricultural greenhouses, causing soil pH to exceed the appropriate range of crop growth, inhibiting or damaging crop growth and causing economic losses.
An intelligent control system is used to create a soil pH prediction model by acquiring historical soil pH data and change data after adjustment measures, analyze and select appropriate adjustment measures, and adjust soil pH in real time to avoid lagging problems.
The rapid and accurate adjustment of the soil pH value is achieved, ensuring that the soil pH value is always within the scope of appropriate growth of crops, avoiding the growth or damage of crop growth, and improving the output and economic benefits of agricultural crops.
Smart Images
Figure CN119356423A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of intelligent agriculture, and in particular relates to an intelligent control system and method for an agricultural greenhouse. Background Art
[0002] With the rapid development of science and technology, my country's agriculture is also constantly developing in the direction of intelligence. China has always been a big agricultural country, but it has not been able to become an agricultural power. Agriculture is the basic industry on which my country depends for survival. For a long time, traditional agricultural production has been restricted by the environmental conditions and weather conditions for the growth of crops. Under such circumstances, agricultural greenhouses came into being. In greenhouses, the corresponding production environment can be created according to the growth environment required by the crops planted therein. This makes it possible to avoid the negative impact of the natural environment on the growth of crops, thereby promoting the growth of crops.
[0003] Although current agricultural greenhouses can create a more suitable growth environment for various crops, such as providing suitable temperature, humidity and light, the current agricultural greenhouses require various fertilizers or irrigation water, which causes the pH value of the soil to change. As time goes by, the pH value of the soil becomes less and less suitable for the growth of crops. After the management personnel take measures to adjust the pH value of the soil, the pH value of the soil cannot be significantly improved in the short term like the temperature, humidity and light, but can only be improved after a certain period of time, that is, there is a lag in the adjustment of the pH value of the soil. If the pH value of the soil is obtained to be beyond the pH value range suitable for the growth of crops, and measures are taken to adjust the pH value of the soil, the soil may not be suitable for the growth of crops for a period of time, and may even inhibit or damage the growth of crops, thereby causing economic losses.
[0004] Therefore, how to improve the control system of existing agricultural greenhouses to avoid delayed soil pH regulation, and thus avoid economic losses caused by the soil pH exceeding the suitable pH range for crop growth, inhibiting or damaging the growth of crops, is a technical problem that urgently needs to be solved. Summary of the invention
[0005] The purpose of the present invention is to provide an intelligent control system and method for an agricultural greenhouse, so as to improve the existing control system of the agricultural greenhouse to avoid the lag in the pH value regulation of the soil, and further avoid the economic losses caused by the pH value of the soil exceeding the suitable pH range for the growth of crops, thereby inhibiting or damaging the growth of crops.
[0006] In order to solve the above technical problems, the technical solution adopted by the present invention is as follows: In a first aspect, an intelligent control method for an agricultural greenhouse is provided, comprising the following steps: S1: Obtain historical soil pH value data of normal fertilization and irrigation of each agricultural greenhouse, and obtain historical soil pH value change data after implementing various soil pH value adjustment measures, and pre-process the obtained historical soil pH value data and historical soil pH value change data; S2: Create a soil pH prediction model to predict the soil pH at a specified time point after normal fertilization and irrigation and the soil pH at a specified time point after the implementation of soil pH adjustment measures based on historical soil pH data and historical soil pH change data; S3: The data analysis module analyzes and obtains the implementation time node of each soil pH value adjustment measure according to the predicted soil pH value at a specified time point after normal fertilization and irrigation and the predicted soil pH value at a specified time point after the implementation of the soil pH value adjustment measure; S4: Selecting corresponding soil pH adjustment measures based on the implementation time nodes of each soil pH adjustment measure and in combination with the current growth stage of the crop, and implementing the corresponding soil pH adjustment measures through the adjustment module at the corresponding implementation time nodes to adjust the soil pH value; S5: measuring the soil pH value at a designated time node after the corresponding soil pH value adjustment measure is implemented, and judging whether the pH value meets the preset pH value range of the corresponding time node based on the measured soil pH value at the designated time node, if so, continuing to implement the soil pH value adjustment measure based on the implementation time node of the soil pH value adjustment measure in step S3, if not, executing step S6; S6: Based on the measured soil pH value at the designated time node and the preset pH value range of the corresponding time node, the implementation time node is adjusted or the soil pH value adjustment measures are adjusted and implemented, and steps S5 and S6 are iterated until the soil pH value meets the preset pH value range of the corresponding time node.
[0007] Preferably, the soil pH value prediction model in step S2 is a convolutional neural network model, including an input layer, a feature extraction layer, a feature fusion layer, a fully connected layer and an output layer; The input layer includes two independent input layer units, each of which inputs soil pH value data and soil pH value change data respectively; Each input layer is linked to a feature extraction layer, wherein the feature extraction layer includes a plurality of convolutional layers, activation layers and pooling layers, and each feature extraction layer independently extracts features of input soil pH value data and soil pH value change data; The feature fusion layer fuses the data features of the input soil pH value data and the soil pH value change data extracted by the feature extraction layer in a specified manner, and inputs the fused features into the fully connected layer; The fully connected layer is used to map the fused features to the final output space, and predict the soil pH value at a specified time point after normal fertilization and irrigation and the soil pH value at a specified time point after the implementation of soil pH adjustment measures.
[0008] Preferably, the feature fusion layer fuses the data features of the input soil pH value data and the soil pH value change data extracted by the feature extraction layer in a specified manner by outer product feature fusion, and the specific process is as follows: S21: converting the soil pH value data and the soil pH value change data into feature vectors at different spatial positions, and calculating the outer product of the feature vectors at different spatial positions between the soil pH value data and the soil pH value change data; S22: performing average aggregation on the outer products of the different spatial positions, and integrating the characteristic relationship between the soil pH value data and the soil pH value change data into the global feature; S23: Using a gradient descent algorithm, the soil pH prediction model is optimized based on the soil pH data, the soil pH change data, the predicted soil pH at a specified time point, and the soil pH at a specified time point after the soil pH adjustment measures are implemented.
[0009] Preferably, the formula for calculating the outer product of the characteristic vectors of the soil pH value data and the soil pH value change data at different spatial positions in step S21 is: Vector M = (a1, a2, a3···an); Vector N = (b1, b2, b3···bn); Then the outer product between vector M and vector N is: M×N= (a1×b1, a2×b2, a3×b3···an×bn); Among them, a1, a2, a3···an are the n characteristic components after the soil pH value data is converted into a vector, and b1, b2, b3···bn are the n characteristic components after the soil pH value change data is converted into a vector.
[0010] Preferably, the specific process of step S3 is as follows: S31: The data analysis module extracts the time nodes at which the soil pH value does not meet the preset pH value range at the specified time point according to the predicted soil pH value at the specified time point after normal fertilization and irrigation, that is, the time nodes at which the pH value needs to be adjusted; S32: Calculate the implementation time node of each soil pH value adjustment measure according to the predicted soil pH value at a designated time point after the implementation of the soil pH value adjustment measure and the time node when the pH value adjustment is required.
[0011] Preferably, the specific process of calculating the implementation time node of each soil pH value adjustment measure in step S32 is as follows: Let the time node for pH value adjustment be A, the target soil pH value to be adjusted at time node A be P1, the soil pH value at the current time node is P0, and the time required to adjust the soil pH value from P0 to P1 by adopting the first soil pH value adjustment measure is t, then the first implementation time node of the implementation time of the first soil pH value adjustment measure is the time node of time t reversed from time node A, that is, the first implementation time node = At; The calculation process of the implementation time nodes of other soil pH adjustment measures is the same as the calculation process of the first soil pH adjustment measure.
[0012] Preferably, the soil pH adjustment measures include increasing the pH value for acidic soil and decreasing the pH value for alkaline soil. The pH adjustment measures include adding soil acidifiers, spreading quicklime and applying decomposed farmyard manure to the irrigation water, and the pH adjustment measures include adding acidic substances to the irrigation water and using acidic fertilizers.
[0013] Preferably, the soil acidifier is made of agricultural water retaining agent and natural peat as raw materials and added with biologically active ingredients, and the acidic substances added to the irrigation water include ferrous sulfate, aluminum sulfate, acetic acid and citric acid.
[0014] In a second aspect, an intelligent control system for an agricultural greenhouse is provided, which is used to implement any one of the intelligent control methods for an agricultural greenhouse, including a data acquisition module, a data processing module, a model creation module, a soil pH value prediction model, a data analysis module, an adjustment module and a pH value determination module; the data acquisition module is connected to the data processing module, the data processing module is connected to the soil pH value prediction model, the model creation module is connected to the soil pH value prediction model, the soil pH value prediction model is connected to the data analysis module, the data analysis module is connected to the adjustment module, and the data analysis module is connected to the pH value determination module; The data acquisition module is used to obtain the historical soil pH value data of each agricultural greenhouse during normal fertilization and irrigation, and to obtain the historical soil pH value change data after various soil pH value adjustment measures are implemented; The data processing module is used to pre-process the acquired historical soil pH value data and historical soil pH value change data; The model creation module is used to create a soil pH value prediction model; The soil pH value prediction model is used to predict the soil pH value at a specified time point after normal fertilization and irrigation and the soil pH value at a specified time point after the implementation of soil pH value adjustment measures based on historical soil pH value data and historical soil pH value change data; The data analysis module is used to analyze and obtain the implementation time node of each soil pH value adjustment measure based on the predicted soil pH value at a specified time point after normal fertilization and irrigation and the predicted soil pH value at a specified time point after the implementation of the soil pH value adjustment measure, and select the corresponding soil pH value adjustment measure based on the implementation time node of each soil pH value adjustment measure and in combination with the current growth stage of the crop; The regulating module is used to regulate the soil pH value by implementing corresponding soil pH regulating measures through the regulating module at corresponding implementation time nodes; The pH value determination module is used to measure the soil pH value at a specified time point after the corresponding soil pH value adjustment measures are implemented and transmit the measurement results to the data analysis module.
[0015] Preferably, it also includes a data acquisition module, which includes a temperature acquisition module, a humidity acquisition module, a light acquisition module and a carbon dioxide acquisition module, and the data acquisition module is connected to the data processing module; the data acquisition module is used to collect temperature, humidity, light intensity and carbon dioxide concentration data in the crop growth environment, and adjust the temperature, humidity, light intensity and carbon dioxide concentration through the adjustment module.
[0016] The beneficial effects of the present invention include: The intelligent control system and method for agricultural greenhouses provided by the present invention obtain soil pH value data of normal fertilization and irrigation of each agricultural greenhouse and soil pH value change data after implementing soil pH value adjustment measures, create a soil pH value prediction model to predict the soil pH value at a specified time point after normal fertilization and irrigation and the soil pH value at a specified time point after implementing soil pH value adjustment measures; obtain the implementation time node of each soil pH value adjustment measure; select and implement the corresponding soil pH value adjustment measure; measure the soil pH value at the specified time node after implementing the corresponding soil pH value adjustment measure, and judge whether the pH value meets the preset pH value range of the corresponding time node. If so, continue to implement the soil pH value adjustment measure; if not, adjust the implementation time node or adjust the soil pH value adjustment measure and implement it.
[0017] Firstly, by creating a soil pH prediction model, an input layer, a feature extraction layer, a feature fusion layer, a fully connected layer and an output layer are set; two independent input layer units are set in the input layer to input soil pH data and soil pH change data respectively; the feature extraction layer sets multiple convolutional layers, activation layers and pooling layers to independently extract the features of the input soil pH data and soil pH change data; the input soil pH data and soil pH change data extracted by the feature extraction layer are feature fused through the feature fusion layer, thereby achieving a better characterization of the different input data; the fully connected layer predicts the soil pH value at a specified time point after normal fertilization and irrigation and the soil pH value at a specified time point after the implementation of soil pH adjustment measures, so that the predicted corresponding data has high accuracy and reliability.
[0018] Secondly, by converting soil pH data and soil pH change data into feature vectors at different spatial positions and calculating the outer product between the vectors; the outer products are averaged and aggregated, the characteristic relationship between soil pH data and soil pH change data is integrated into the global features, and the soil pH prediction model is optimized using the gradient descent algorithm. The feature dimension and computational complexity are reduced through outer product feature fusion, which can better characterize the complex relationship between the two modes and improve the performance and generalization ability of the soil pH prediction model.
[0019] Again, by setting the time node at which pH value adjustment is required to be A, the target soil pH value that needs to be adjusted to time node A to be P1, the soil pH value at the current time node to be P0, and the time required for the first soil pH value adjustment measure to adjust the soil pH value from P0 to P1 to be t, the first implementation time node of the implementation time of the first soil pH value adjustment measure is used as the process of reversing the time node of time t from time node A, combined with the prediction of the soil pH value prediction model, the lag problem of soil adjustment is solved, and the accurate adjustment of the pH value at the corresponding time node is achieved.
[0020] Finally, a data acquisition module is further set up to collect data on temperature, humidity, light intensity and carbon dioxide concentration in the crop growth environment, and the temperature, humidity, light intensity and carbon dioxide concentration are adjusted through the adjustment module to achieve accurate control of the temperature, humidity, light intensity and carbon dioxide concentration of crops, so that various parameters in the agricultural greenhouse are suitable for the growth of crops, thereby increasing crop yields and improving economic benefits. BRIEF DESCRIPTION OF THE DRAWINGS
[0021] Figure 1 It is a schematic diagram of the flow of the intelligent control method of the agricultural greenhouse of the present invention.
[0022] Figure 2It is a schematic diagram of the flow of outer product feature fusion in the intelligent control method of the agricultural greenhouse of the present invention.
[0023] Figure 3 It is a schematic diagram of the architecture of the intelligent control system of the agricultural greenhouse of the present invention. DETAILED DESCRIPTION
[0024] The following is combined with Figure 1~Figure 3 The present invention is further described in detail: Example 1 See attached Figure 1 As shown, an intelligent control method for an agricultural greenhouse comprises the following steps: S1: Obtain the historical soil pH value data of each agricultural greenhouse under normal fertilization and irrigation, and obtain the historical soil pH value change data after implementing various soil pH value adjustment measures, and preprocess the acquired historical soil pH value data and historical soil pH value change data. The preprocessing process here includes data cleaning, processing erroneous data, duplicate data and abnormal data in the acquired data, and integrating data from multiple data sources into one data set, resolving conflicts and redundancies between data sources, and then converting the data into a format or form suitable for analysis, such as unifying data formats and types. Finally, linear normalization is performed to eliminate the differences in dimensions and value ranges between the data, so that different indicators are comparable.
[0025] S2: Create a soil pH prediction model to predict the soil pH at a specified time point after normal fertilization and irrigation and the soil pH at a specified time point after the implementation of soil pH adjustment measures based on historical soil pH data and historical soil pH change data. The soil pH is predicted by creating a soil pH prediction model with input layer, feature extraction layer, feature fusion layer, fully connected layer and output layer, and the predicted corresponding data has high accuracy and reliability.
[0026] S3: The data analysis module analyzes and obtains the implementation time nodes of each soil pH value adjustment measure based on the predicted soil pH value at a specified time point after normal fertilization and irrigation and the predicted soil pH value at a specified time point after the implementation of the soil pH value adjustment measure. The pH value at the corresponding time node is adjusted by respectively calculating the implementation time of different soil pH value adjustment measures, and combined with the prediction of the soil pH value prediction model, the pH value at the corresponding time node is accurately adjusted, so that the lag problem of soil adjustment is solved.
[0027] S4: Based on the implementation time nodes of each soil pH adjustment measure and in combination with the current growth stage of the crop, the corresponding soil pH adjustment measure is selected, and the soil pH is adjusted by the adjustment module at the corresponding implementation time node. By selecting the appropriate soil pH adjustment measure at the corresponding growth stage of the crop, for example, in the growth stage of the crop that needs fertilization, the corresponding acid fertilizer can be applied, and in the growth stage that needs irrigation, soil acidifiers or acidic substances can be added to the irrigation water. For crops that are not affected by quicklime, quicklime can be spread. That is, different soil pH adjustment measures are selected for different crops and responses to different soil pH adjustment measures to promote the growth of crops.
[0028] S5: measuring the soil pH value at a designated time node after the corresponding soil pH value adjustment measure is implemented, and judging whether the pH value meets the preset pH value range of the corresponding time node based on the measured soil pH value at the designated time node, if so, continuing to implement the soil pH value adjustment measure based on the implementation time node of the soil pH value adjustment measure in step S3, if not, executing step S6; S6: Based on the measured soil pH value at the designated time node and the preset pH value range at the corresponding time node, the implementation time node is adjusted or the soil pH value adjustment measures are adjusted and implemented, and steps S5 and S6 are iteratively executed until the soil pH value meets the preset pH value range at the corresponding time node. Through the iterative adjustment process of the soil pH value in steps S5 and S6, the pH value of the soil in the entire life cycle of the crop can be within the preset pH value range, promoting the normal growth of the crop.
[0029] Example 2 On the basis of Example 1, the soil pH prediction model in step S2 is a convolutional neural network model, including an input layer, a feature extraction layer, a feature fusion layer, a fully connected layer and an output layer. The input layer includes two independent input layer units, each of which inputs soil pH data and soil pH change data respectively. Each input layer is linked to a feature extraction layer, and the feature extraction layer includes multiple convolutional layers, activation layers and pooling layers, and each feature extraction layer independently extracts the features of the input soil pH data and the soil pH change data. The feature fusion layer fuses the data features of the input soil pH data and the soil pH change data extracted by the feature extraction layer in a specified manner, and inputs the fused features into the fully connected layer. The fully connected layer is used to map the fused features to the final output space, and predict the soil pH value at a specified time point after normal fertilization and irrigation and the soil pH value at a specified time point after the implementation of soil pH adjustment measures.
[0030] In this embodiment, the feature fusion layer fuses the data features of the input soil pH value data and the soil pH value change data extracted by the feature extraction layer in a specified manner, which is outer product feature fusion. The specific process is as follows: S21: converting the soil pH value data and the soil pH value change data into feature vectors at different spatial positions, and calculating the outer product of the feature vectors at different spatial positions between the soil pH value data and the soil pH value change data; S22: performing average aggregation on the outer products of the different spatial positions, and integrating the characteristic relationship between the soil pH value data and the soil pH value change data into the global feature; S23: Using a gradient descent algorithm, the soil pH prediction model is optimized based on the soil pH data, the soil pH change data, the predicted soil pH at a specified time point, and the soil pH at a specified time point after the soil pH adjustment measures are implemented.
[0031] Through the above, the soil pH value data and the soil pH value change data are converted into feature vectors at different spatial positions and the outer product between the vectors is calculated; the outer products are averaged and aggregated, the feature relationship between the soil pH value data and the soil pH value change data is integrated into the global features, and the soil pH value prediction model is optimized using the gradient descent algorithm. The feature dimension and computational complexity are reduced through outer product feature fusion, so as to better characterize the complex relationship between the two modes and improve the performance and generalization ability of the soil pH value prediction model.
[0032] The formula for calculating the outer product of the characteristic vectors of the soil pH value data and the soil pH value change data at different spatial positions in step S21 of the above process is: Vector M = (a1, a2, a3···an); Vector N = (b1, b2, b3···bn); Then the outer product between vector M and vector N is: M×N= (a1×b1, a2×b2, a3×b3···an×bn); Among them, a1, a2, a3···an are the n characteristic components after the soil pH value data is converted into a vector, and b1, b2, b3···bn are the n characteristic components after the soil pH value change data is converted into a vector.
[0033] In this embodiment, a soil pH prediction model is created, and an input layer, a feature extraction layer, a feature fusion layer, a fully connected layer and an output layer are set; two independent input layer units are set in the input layer to input soil pH value data and soil pH value change data respectively; the feature extraction layer sets multiple convolution layers, activation layers and pooling layers to independently extract the features of the input soil pH value data and the soil pH value change data; the input soil pH value data and the soil pH value change data extracted by the feature extraction layer are feature fused through the feature fusion layer, thereby achieving better characterization of different input data; the fully connected layer predicts the soil pH value at a specified time point after normal fertilization and irrigation and the soil pH value at a specified time point after the implementation of soil pH adjustment measures, so that the predicted corresponding data has high accuracy and reliability.
[0034] Example 3 Based on Example 1 or Example 2, the specific process of step S3 is as follows: S31: The data analysis module extracts the time nodes at which the soil pH value does not meet the preset pH value range at the specified time point according to the predicted soil pH value at the specified time point after normal fertilization and irrigation, that is, the time nodes at which the pH value needs to be adjusted; S32: Calculate the implementation time node of each soil pH value adjustment measure according to the predicted soil pH value at a designated time point after the implementation of the soil pH value adjustment measure and the time node when the pH value adjustment is required.
[0035] In this embodiment, the specific process of calculating the implementation time node of each soil pH value adjustment measure in step S32 is as follows: Let the time node that needs to adjust the pH value be A, the target soil pH value that needs to be adjusted to at time node A be P1, the soil pH value at the current time node be P0, and the time required to adjust the soil pH value from P0 to P1 by adopting the first soil pH value adjustment measure be t, then the first implementation time node of the implementation time of the first soil pH value adjustment measure is the time node of time t reversed from time node A, that is, the first implementation time node = At. Combined with the prediction of the soil pH value prediction model and the calculation of the implementation time of different soil pH value adjustment measures, the lag problem of soil adjustment is solved, and the accurate adjustment of the pH value at the corresponding time node is achieved.
[0036] The calculation of the implementation time nodes of other soil pH adjustment measures is the same as the calculation process of the first soil pH adjustment measure. Soil pH adjustment measures include increasing the pH value for acidic soils and decreasing the pH value for alkaline soils. The pH increase measures include adding soil acidifiers, spreading quicklime and applying decomposed farmyard manure to irrigation water, and the pH decrease measures include adding acidic substances to irrigation water and using acidic fertilizers. The soil acidifier is made of agricultural water retaining agents and natural peat as raw materials and added with biologically active ingredients. The acidic substances added to irrigation water include ferrous sulfate, aluminum sulfate, acetic acid and citric acid.
[0037] An intelligent control system for an agricultural greenhouse, used to implement any one of the intelligent control methods for an agricultural greenhouse, comprising a data acquisition module, a data processing module, a model creation module, a soil pH value prediction model, a data analysis module, an adjustment module and a pH value determination module; the data acquisition module is connected to the data processing module, the data processing module is connected to the soil pH value prediction model, the model creation module is connected to the soil pH value prediction model, the soil pH value prediction model is connected to the data analysis module, the data analysis module is connected to the adjustment module, and the data analysis module is connected to the pH value determination module.
[0038] The data acquisition module is used to obtain the historical soil pH value data of normal fertilization and irrigation of each agricultural greenhouse, and to obtain the historical soil pH value change data after implementing each soil pH value adjustment measure. The data processing module is used to pre-process the acquired historical soil pH value data and historical soil pH value change data. The model creation module is used to create a soil pH value prediction model. The soil pH value prediction model is used to predict the soil pH value at a specified time point after normal fertilization and irrigation and the soil pH value at a specified time point after the implementation of the soil pH value adjustment measure based on the historical soil pH value data and the historical soil pH value change data. The data analysis module is used to analyze and obtain the implementation time node of each soil pH value adjustment measure based on the predicted soil pH value at a specified time point after normal fertilization and irrigation and the predicted soil pH value at a specified time point after the implementation of the soil pH value adjustment measure, and select the corresponding soil pH value adjustment measure based on the implementation time node of each soil pH value adjustment measure and in combination with the current growth stage of the crop. The adjustment module is used to adjust the soil pH value by implementing the corresponding soil pH value adjustment measure through the adjustment module at the corresponding implementation time node. The pH value determination module is used to measure the soil pH value at a specified time point after the corresponding soil pH value adjustment measures are implemented and transmit the measurement results to the data analysis module.
[0039] The intelligent control system of the agricultural greenhouse also includes a data acquisition module, which includes a temperature acquisition module, a humidity acquisition module, a light acquisition module and a carbon dioxide acquisition module, and the data acquisition module is connected to the data processing module; the data acquisition module is used to collect the temperature, humidity, light intensity and carbon dioxide concentration data in the crop growth environment, and adjust the temperature, humidity, light intensity and carbon dioxide concentration through the adjustment module. The data acquisition module is set to collect the temperature, humidity, light intensity and carbon dioxide concentration data in the crop growth environment, and adjust the temperature, humidity, light intensity and carbon dioxide concentration through the adjustment module to achieve accurate control of the temperature, humidity, light intensity and carbon dioxide concentration of the crops, so that the various parameters in the agricultural greenhouse are suitable for the growth of crops, thereby increasing crop yields and improving economic benefits.
[0040] In summary, the intelligent control system and method for agricultural greenhouses provided by the present invention create a soil pH prediction model, set an input layer, a feature extraction layer, a feature fusion layer, a fully connected layer and an output layer; set two independent input layer units in the input layer to input soil pH value data and soil pH value change data respectively; the feature extraction layer sets multiple convolution layers, activation layers and pooling layers to independently extract the features of the input soil pH value data and the soil pH value change data; the input soil pH value data and the soil pH value change data extracted by the feature extraction layer are feature fused through the feature fusion layer, thereby achieving better characterization of different input data; the fully connected layer predicts the soil pH value at a specified time point after normal fertilization and irrigation and the soil pH value at a specified time point after the implementation of soil pH adjustment measures, so that the predicted corresponding data has high accuracy and reliability.
[0041] The soil pH value data and soil pH value change data are converted into feature vectors at different spatial positions and the outer product between the vectors is calculated; the outer product is averaged and aggregated, the feature relationship between the soil pH value data and the soil pH value change data is integrated into the global feature, and the soil pH value prediction model is optimized using the gradient descent algorithm. The feature dimension and computational complexity are reduced through the outer product feature fusion, and the complex relationship between the two modes is better characterized, thereby improving the performance and generalization ability of the soil pH value prediction model. By setting the time node for pH value adjustment as A, the target soil pH value to be adjusted to at time node A as P1, the soil pH value at the current time node as P0, and the time required for the first soil pH value adjustment measure to adjust the soil pH value from P0 to P1 as t, the first implementation time node of the implementation time of the first soil pH value adjustment measure is used as the process of reversing the time node of time t from time node A, combined with the prediction of the soil pH value prediction model, the lag problem of soil adjustment is solved, and the accurate adjustment of the pH value at the corresponding time node is achieved. By setting up a data acquisition module to collect data on temperature, humidity, light intensity and carbon dioxide concentration in the crop growth environment, and adjusting the temperature, humidity, light intensity and carbon dioxide concentration through an adjustment module, accurate control of the temperature, humidity, light intensity and carbon dioxide concentration of crops can be achieved, so that various parameters in the agricultural greenhouse are suitable for the growth of crops, thereby increasing crop yields and improving economic benefits.
Claims
1. An intelligent control method for an agricultural greenhouse, characterized in that: The following steps are involved: S1: Obtain historical soil pH value data of normal fertilization and irrigation of each agricultural greenhouse, and obtain historical soil pH value change data after implementing various soil pH value adjustment measures, and pre-process the obtained historical soil pH value data and historical soil pH value change data; S2: Create a soil pH prediction model to predict the soil pH at a specified time point after normal fertilization and irrigation and the soil pH at a specified time point after the implementation of soil pH adjustment measures based on historical soil pH data and historical soil pH change data; S3: The data analysis module analyzes and obtains the implementation time node of each soil pH value adjustment measure according to the predicted soil pH value at a specified time point after normal fertilization and irrigation and the predicted soil pH value at a specified time point after the implementation of the soil pH value adjustment measure; S4: Selecting corresponding soil pH adjustment measures based on the implementation time nodes of each soil pH adjustment measure and in combination with the current growth stage of the crop, and implementing the corresponding soil pH adjustment measures through the adjustment module at the corresponding implementation time nodes to adjust the soil pH value; S5: measuring the soil pH value at a designated time node after the corresponding soil pH value adjustment measure is implemented, and judging whether the pH value meets the preset pH value range of the corresponding time node based on the measured soil pH value at the designated time node, if so, continuing to implement the soil pH value adjustment measure based on the implementation time node of the soil pH value adjustment measure in step S3, if not, executing step S6; S6: Based on the measured soil pH value at the designated time node and the preset pH value range of the corresponding time node, the implementation time node is adjusted or the soil pH value adjustment measures are adjusted and implemented, and steps S5 and S6 are iterated until the soil pH value meets the preset pH value range of the corresponding time node.
2. The intelligent control method for an agricultural greenhouse according to claim 1, characterized in that: The soil pH value prediction model in step S2 is a convolutional neural network model, including an input layer, a feature extraction layer, a feature fusion layer, a fully connected layer and an output layer; The input layer includes two independent input layer units, each of which inputs soil pH value data and soil pH value change data respectively; Each input layer is linked to a feature extraction layer, wherein the feature extraction layer includes a plurality of convolutional layers, activation layers and pooling layers, and each feature extraction layer independently extracts features of input soil pH value data and soil pH value change data; The feature fusion layer fuses the data features of the input soil pH value data and the soil pH value change data extracted by the feature extraction layer in a specified manner, and inputs the fused features into the fully connected layer; The fully connected layer is used to map the fused features to the final output space, and predict the soil pH value at a specified time point after normal fertilization and irrigation and the soil pH value at a specified time point after the implementation of soil pH adjustment measures.
3. The intelligent control method for an agricultural greenhouse according to claim 2, characterized in that: The feature fusion layer fuses the data features of the input soil pH value data and the soil pH value change data extracted by the feature extraction layer in a specified manner, which is outer product feature fusion. The specific process is as follows: S21: converting the soil pH value data and the soil pH value change data into feature vectors at different spatial positions, and calculating the outer product of the feature vectors at different spatial positions between the soil pH value data and the soil pH value change data; S22: performing average aggregation on the outer products of the different spatial positions, and integrating the characteristic relationship between the soil pH value data and the soil pH value change data into the global feature; S23: Using a gradient descent algorithm, the soil pH prediction model is optimized based on the soil pH data, the soil pH change data, the predicted soil pH at a specified time point, and the soil pH at a specified time point after the soil pH adjustment measures are implemented.
4. The intelligent control method for an agricultural greenhouse according to claim 3, characterized in that: The formula for calculating the outer product of the characteristic vectors of the soil pH value data and the soil pH value change data at different spatial positions in step S21 is: Vector M = (a1, a2, a3···an); Vector N = (b1, b2, b3···bn); Then the outer product between vector M and vector N is: M×N= (a1×b1, a2×b2, a3×b3···an×bn); Among them, a1, a2, a3···an are the n characteristic components after the soil pH value data is converted into a vector, and b1, b2, b3···bn are the n characteristic components after the soil pH value change data is converted into a vector.
5. The intelligent control method for an agricultural greenhouse according to claim 1, characterized in that: The specific process of step S3 is as follows: S31: The data analysis module extracts the time nodes at which the soil pH value does not meet the preset pH value range at the specified time point according to the predicted soil pH value at the specified time point after normal fertilization and irrigation, that is, the time nodes at which the pH value needs to be adjusted; S32: Calculate the implementation time node of each soil pH value adjustment measure according to the predicted soil pH value at a designated time point after the implementation of the soil pH value adjustment measure and the time node when the pH value adjustment is required.
6. The intelligent control method for agricultural greenhouse according to claim 5, characterized in that: The specific process of calculating the implementation time node of each soil pH adjustment measure in step S32 is as follows: Let the time node for pH value adjustment be A, the target soil pH value to be adjusted at time node A be P1, the soil pH value at the current time node is P0, and the time required to adjust the soil pH value from P0 to P1 by adopting the first soil pH value adjustment measure is t, then the first implementation time node of the implementation time of the first soil pH value adjustment measure is the time node of time t reversed from time node A, that is, the first implementation time node = At; The calculation process of the implementation time nodes of other soil pH adjustment measures is the same as the calculation process of the first soil pH adjustment measure.
7. The intelligent control method for an agricultural greenhouse according to claim 1, characterized in that: The soil pH adjustment measures include increasing the pH value for acidic soil and decreasing the pH value for alkaline soil. The pH adjustment measures include adding soil acidifiers, spreading quicklime and applying decomposed farmyard manure to irrigation water, and the pH adjustment measures include adding acidic substances to irrigation water and using acidic fertilizers.
8. The intelligent control method for agricultural greenhouse according to claim 7, characterized in that: The soil acidifier is made of agricultural water retaining agent and natural peat as raw materials and added with biological active ingredients. Acidic substances including ferrous sulfate, aluminum sulfate, acetic acid and citric acid are added to irrigation water.
9. An intelligent control system for an agricultural greenhouse, used to implement an intelligent control method for an agricultural greenhouse according to any one of claims 1 to 8, characterized in that: It includes a data acquisition module, a data processing module, a model creation module, a soil pH value prediction model, a data analysis module, an adjustment module and a pH value determination module; the data acquisition module is connected to the data processing module, the data processing module is connected to the soil pH value prediction model, the model creation module is connected to the soil pH value prediction model, the soil pH value prediction model is connected to the data analysis module, the data analysis module is connected to the adjustment module, and the data analysis module is connected to the pH value determination module; The data acquisition module is used to obtain the historical soil pH value data of each agricultural greenhouse during normal fertilization and irrigation, and to obtain the historical soil pH value change data after various soil pH value adjustment measures are implemented; The data processing module is used to pre-process the acquired historical soil pH value data and historical soil pH value change data; The model creation module is used to create a soil pH value prediction model; The soil pH value prediction model is used to predict the soil pH value at a specified time point after normal fertilization and irrigation and the soil pH value at a specified time point after the implementation of soil pH value adjustment measures based on historical soil pH value data and historical soil pH value change data; The data analysis module is used to analyze and obtain the implementation time node of each soil pH value adjustment measure based on the predicted soil pH value at a specified time point after normal fertilization and irrigation and the predicted soil pH value at a specified time point after the implementation of the soil pH value adjustment measure, and select the corresponding soil pH value adjustment measure based on the implementation time node of each soil pH value adjustment measure and in combination with the current growth stage of the crop; The regulating module is used to regulate the soil pH value by implementing corresponding soil pH regulating measures through the regulating module at corresponding implementation time nodes; The pH value determination module is used to measure the soil pH value at a specified time point after the corresponding soil pH value adjustment measures are implemented and transmit the measurement results to the data analysis module.
10. The intelligent control system for agricultural greenhouses according to claim 9, characterized in that: It also includes a data acquisition module, which includes a temperature acquisition module, a humidity acquisition module, a light acquisition module and a carbon dioxide acquisition module, and the data acquisition module is connected to the data processing module; the data acquisition module is used to collect temperature, humidity, light intensity and carbon dioxide concentration data in the crop growth environment, and adjust the temperature, humidity, light intensity and carbon dioxide concentration through the adjustment module.