Seedling raising greenhouse environment intelligent regulation and control system based on Internet of Things

Abstract

The invention discloses an intelligent seedling greenhouse environment regulation and control system based on the Internet of Things, and belongs to the field of intelligent regulation and control of the Internet of Things, and the system comprises a heterogeneous sensor dynamic networking construction module which is used for deploying self-positioning wireless sensor nodes and obtaining a dynamic networking scheme based on the wireless sensor nodes; the environment data acquisition and preprocessing module is used for acquiring a multi-dimensional environment data set and preprocessing the multi-dimensional environment data set to obtain a complete environment data set; the growth stage self-adaptive control module is used for dynamically regulating and controlling a seedling culture strategy by adopting an environment variable coupling method based on the complete environment data set to obtain a self-adaptive regulation and control scheme; and the virtual feedback simulation module is used for performing calibration regulation and control through a virtual feedback mechanism and an actual acquisition device to obtain an optimized regulation and control scheme.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent control of the Internet of Things, and in particular relates to an intelligent control system for a seedling greenhouse environment based on the Internet of Things. Background Art

[0002] The rapid development of intelligent agriculture has brought unprecedented opportunities to modern agriculture. Through precise environmental control and data-driven decision-making, it has significantly increased crop yields and resource utilization efficiency, becoming a key direction for promoting sustainable agricultural development. However, existing intelligent agricultural solutions still face numerous shortcomings in practical application. Many systems rely on a single type of sensor, making it difficult to fully capture complex environmental information. They also employ fixed networking approaches, lacking the ability to dynamically adapt to complex farmland environments. Furthermore, control strategies are often based on static thresholds, making them difficult to adapt to the dynamic demands of different stages of the crop growth cycle, leading to resource waste or poor growth performance. Inadequate coordination between actuators and data analysis also makes it difficult for systems to achieve a balance between low energy consumption and efficient growth. Against this backdrop, core challenges in the field of intelligent agriculture are becoming increasingly apparent. Complex farmland environments require flexible sensor networking to adapt to diverse terrain and crop distributions. However, traditional fixed networks are difficult to dynamically adjust, limiting data collection coverage and accuracy. Consequently, the spatiotemporal consistency of environmental data becomes a challenge. Incomplete or abnormal sensor data further complicates analysis and impacts the accuracy of control decisions. Furthermore, crop growth is influenced by the interactions of multiple factors, such as light, temperature, and CO2. Existing methods lack in-depth analysis of the coupled relationships between these factors, making it difficult to achieve adaptive control based on growth stages. Ultimately, the dynamic matching of actual environmental data with crop growth status makes it difficult for the system to optimize control strategies using real-time feedback. Therefore, designing an intelligent agricultural system that dynamically networks to comprehensively collect environmental data, implements phased adaptive control through multi-factor coupled analysis, and utilizes a virtual feedback mechanism to calibrate actual growth status has become a key issue in promoting the development of precision agriculture. Summary of the Invention

[0003] In order to solve the above technical problems, the present invention proposes an intelligent control system for seedling greenhouse environment based on the Internet of Things to solve the problems existing in the above-mentioned prior art.

[0004] To achieve the above objectives, the present invention provides an Internet of Things-based intelligent control system for a nursery greenhouse environment, comprising:

[0005] A heterogeneous sensor dynamic networking construction module is used to deploy self-locating wireless sensor nodes and obtain a dynamic networking solution based on the wireless sensor nodes;

[0006] An environmental data collection and preprocessing module is used to collect a multidimensional environmental data set and preprocess the multidimensional environmental data set to obtain a complete environmental data set;

[0007] A growth stage adaptive control module is used to dynamically control the seedling raising strategy based on the complete environmental data set using an environmental variable coupling method to obtain an adaptive control scheme;

[0008] The virtual feedback simulation module is used to calibrate and control the actual acquisition device through the virtual feedback mechanism to obtain an optimized control solution.

[0009] Optionally, the heterogeneous sensor dynamic networking construction module includes: a sensor node arrangement unit and a network structure adjustment unit;

[0010] Wherein, the sensor node arrangement unit obtains the connection relationship between wireless sensor nodes based on LoRa+BLE dual-mode communication;

[0011] The network structure adjustment unit generates an initial mesh network based on the connection relationship between sensor nodes based on a topology reconstruction algorithm, and adjusts the connection relationship between wireless sensor nodes according to the terrain and crop distribution to obtain a dynamic networking solution.

[0012] Optionally, the environmental data acquisition and preprocessing module includes: a data acquisition unit, a first processing unit and a second processing unit;

[0013] The data acquisition unit is used to extract sensor node locations from the dynamic networking scheme and collect multi-dimensional environmental data sets using a layered sampling method;

[0014] The first processing unit is used to perform partition processing on the multidimensional environmental data set by using a K-means clustering algorithm to determine the environmental pattern of each area in the seedling greenhouse environment;

[0015] The second processing unit is used to perform spatiotemporal alignment and outlier repair on the multidimensional environmental dataset based on the environmental pattern of each region using the built-in lightweight crop growth model of the wireless sensor node to obtain a complete environmental dataset.

[0016] Optionally, the growth phase adaptive control module includes: an environmental variable coupling analysis unit, a control parameter set determination unit, and a dynamic control strategy generation unit;

[0017] The environmental variable coupling analysis unit is used to extract the coupling relationship characteristics between light, temperature, and relative air humidity using an environmental parameter coupling analysis algorithm for the complete environmental data set, establish a cross-influence matrix of the three factors of light, temperature, and relative air humidity, and quantify the weight of each factor through tensor decomposition to obtain an environmental variable coupling model;

[0018] The control parameter set determination unit is configured to obtain characteristic weights of environmental variables based on the environmental variable coupling model, and determine a control parameter set corresponding to a current growth stage by combining the characteristic weights of the environmental variables with a preset crop growth stage division rule;

[0019] The dynamic control strategy generation unit is used to generate dynamic control strategies for lighting, irrigation, and circulating fans using a fuzzy logic algorithm for a control parameter set to obtain an adaptive control solution.

[0020] Optionally, the environmental variable coupling analysis unit includes: a key coupling feature extraction subunit, an environmental variable weight distribution subunit and a partition feature analysis subunit;

[0021] The key coupling feature extraction subunit is used to normalize the complete environment data set and perform principal component analysis on the normalized complete environment data set to obtain key coupling features;

[0022] The environmental variable weight distribution subunit is used to construct a cross-influence matrix of three factors: light, temperature, and relative air humidity based on the key coupling characteristics, and process the established cross-influence matrix of three factors: light, temperature, and relative air humidity by a matrix decomposition method to obtain an environmental variable weight distribution;

[0023] The partition feature analysis subunit is used to update the normalized complete environmental data set based on the environmental variable weight distribution, divide the spatial distribution of the light, temperature, and relative air humidity coupling characteristics based on the updated normalized complete environmental data set, and construct a dynamic distribution map of the environmental variables based on the spatial distribution of the light, temperature, and relative air humidity coupling characteristics to determine the environmental variable coupling characteristics of each area in the greenhouse environment.

[0024] Optionally, the control parameter set determination unit includes: a stage prediction subunit, a growth stage classification subunit and an environmental parameter control subunit;

[0025] The stage prediction subunit is used to generate a weight distribution data set based on the weight distribution of the environmental variables, and to use a linear regression algorithm to fit the relationship between light intensity, temperature change, and relative humidity of the air and the crop growth stage based on the weight distribution data set to obtain a stage prediction model;

[0026] The growth stage classification subunit is used to determine the crop growth stage corresponding to the current environmental variable based on the stage prediction model to obtain the current growth stage classification;

[0027] The environmental parameter control subunit is used to query a preset growth stage and control parameter mapping table based on the current growth stage classification, obtain corresponding light, temperature and air relative humidity control parameter sets, and generate an environmental control parameter set.

[0028] Optionally, the virtual feedback simulation module includes: a simulation growth unit, an optimization control scheme generation unit and a user-friendly unit;

[0029] The simulated growth unit is used to run an adaptive control scheme in a preset virtual farmland environment through a virtual feedback mechanism to obtain simulated crop growth status data;

[0030] The optimized control scheme generating unit is used to extract key indicators from the simulated growth state data, compare the key indicators of the simulated growth state data with the crop growth data collected by the actual sensor, calibrate the control parameters through an error analysis algorithm, and obtain an optimized control scheme;

[0031] The user-friendly unit is used to feed back the optimization control plan to the user, and the user manages the seedling greenhouse environment based on the optimization control plan.

[0032] Compared with the prior art, the present invention has the following advantages and technical effects:

[0033] The present invention discloses an intelligent agricultural environmental control system. By deploying self-positioning wireless sensor nodes to form a dynamic mesh network, multi-dimensional environmental data is collected. A lightweight crop growth model is used to align the data in time and space and repair anomalies, extracting the coupling relationship characteristics of factors such as light, temperature, and relative humidity. In combination with the crop growth stage, a fuzzy logic algorithm is used to generate an adaptive control strategy. The control effect is simulated through a virtual feedback mechanism, and the calibration parameters are compared with actual growth data. Ultimately, continuous optimization and control of the agricultural environment is achieved, improving crop growth efficiency. The present invention solves the problems of low accuracy and fixed control strategies in traditional agricultural environmental monitoring, and realizes intelligent and precise control based on multi-dimensional environmental data. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] The accompanying drawings, which constitute part of this application, are intended to provide a further understanding of this application. The exemplary embodiments and descriptions of this application are intended to explain this application and do not constitute an improper limitation on this application. In the accompanying drawings:

[0035] Figure 1 This is a structural diagram of the intelligent control system for the seedling greenhouse environment based on the Internet of Things according to an embodiment of the present invention. DETAILED DESCRIPTION

[0036] It should be noted that, in the absence of conflict, the embodiments and features of the embodiments in this application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0037] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0038] Example 1

[0039] like Figure 1 As shown, this embodiment provides an intelligent control system for seedling greenhouse environment based on the Internet of Things, including: a heterogeneous sensor dynamic networking construction module, an environmental data acquisition and preprocessing module, a growth stage adaptive control module, and a virtual feedback simulation module.

[0040] The heterogeneous sensor dynamic networking construction module is used to deploy self-locating wireless sensor nodes and generate dynamic networking solutions based on these wireless sensor nodes. This module includes a sensor node placement unit and a network structure adjustment unit. The sensor node placement unit uses LoRa+BLE dual-mode communication to obtain the connection relationships between sensor nodes. The network structure adjustment unit uses a topology reconstruction algorithm to generate an initial mesh network based on the connection relationships between sensor nodes. It then adjusts the sensor node connections based on terrain and crop distribution to generate a dynamic networking solution.

[0041] Furthermore, as a specific implementation of this embodiment, the process for the heterogeneous sensor dynamic networking construction module to obtain a dynamic networking solution includes: collecting air relative humidity, light intensity, temperature and humidity, and soil conductivity data through sensor nodes, and using self-positioning functions to determine the spatial location of wireless sensor nodes to obtain a wireless sensor node distribution dataset. A topology analysis algorithm is used to process the wireless sensor node distribution dataset, calculate the distance and signal strength between wireless sensor nodes, and generate an initial mesh network structure. Real-time communication quality data between nodes is obtained through LoRa and BLE communication, and a topology reconstruction algorithm is used to adjust node connectivity to obtain an optimized mesh network topology. Based on terrain feature data, whether there is obstruction between nodes is determined. If so, the topology reconstruction algorithm is used to reallocate communication paths to obtain a terrain-adapted network connection solution. Crop distribution data is obtained, and the degree of overlap between the crop distribution and the coverage of the wireless sensor nodes is analyzed. If the overlap is below a preset threshold, the wireless sensor node locations are adjusted or the number of nodes is increased to obtain an optimized wireless sensor node distribution solution. The optimized wireless sensor node distribution solution is verified using a topology analysis algorithm, and network coverage and communication stability are calculated to obtain the final dynamic networking solution.

[0042] Furthermore, wireless sensor nodes deployed in the nursery greenhouse can collect data on light intensity, temperature and humidity, and soil conductivity every 10 minutes. Temperature and humidity affect crop growth, while soil conductivity indicates soil fertility. This data is acquired by the node's built-in sensor modules, with an accuracy of ±5% for light intensity, ±0.5°C / ±3% for temperature and humidity, and ±10μS / cm for soil conductivity. After collection, the node uses self-positioning capabilities, such as those based on GPS or ultra-wideband technology, to determine its coordinates in three-dimensional space with an accuracy of up to ±0.5m, thus forming a node-distributed dataset.

[0043] The topology analysis algorithm generates an initial mesh network structure based on inter-node distance and signal strength. Nodes A and B are 50 meters apart, with an RSSI of -60dBm, indicating good signal quality and a direct connection. Nodes C and D, however, are 200 meters apart, with an RSSI of -90dBm, indicating a weak signal and requiring a relay node to connect. This forms the initial mesh network, achieving an initial coverage rate of 85%. Wireless sensor nodes transmit environmental data via LoRa and exchange topology adjustment commands via BLE. Communication quality data, such as packet loss rate (less than 5% is considered optimal), drives the topology reconstruction algorithm, dynamically adjusting connection relationships.

[0044] The analysis of crop distribution data involves the degree of overlap between the coverage of wireless sensor nodes and the crop area. Assume that the coverage radius of wireless sensor nodes is 100m and the crop area is 2hm. 2If the overlap rate falls below 80%, adjustments are required. After adding two new nodes, the overlap rate increased to 95%, enabling more comprehensive crop growth monitoring. The adjusted wireless sensor node distribution was verified through topology analysis, achieving 98% network coverage and significantly improving communication stability (bit error rate below 1%).

[0045] The environmental data acquisition and preprocessing module is used to collect multidimensional environmental data sets and preprocess the multidimensional environmental data sets to obtain a complete environmental data set. The environmental data acquisition and preprocessing module includes: a data acquisition unit, a first processing unit, and a second processing unit. The data acquisition unit is used to extract sensor node locations from the dynamic networking solution and use a stratified sampling method to collect the multidimensional environmental data set. The first processing unit is used to partition the multidimensional environmental data set using the K-means clustering algorithm to determine the environmental patterns of each area in the seedling greenhouse environment. The second processing unit is used to perform spatiotemporal alignment and outlier repair on the multidimensional environmental data set based on the environmental patterns of each area using the built-in lightweight crop growth model of the wireless sensor node to obtain a complete environmental data set.

[0046] Furthermore, as a specific implementation method of this embodiment, the process of the environmental data acquisition module collecting a multidimensional environmental dataset includes: obtaining wireless sensor coordinate data from a dynamic networking solution, using a stratified sampling strategy to collect light, temperature, and relative humidity variables, and generating a multidimensional environmental dataset. The multidimensional environmental dataset is partitioned using a K-means clustering algorithm, and the environmental pattern of each area is determined based on the sensor coordinates and environmental variable characteristics. The multidimensional environmental dataset is subjected to spatiotemporal alignment, and a lightweight long-short-term memory model is used to correct timestamp deviations to generate an aligned environmental dataset. If the aligned environmental dataset contains missing values ​​or outliers, the data is repaired using an interpolation algorithm to obtain a complete environmental dataset.

[0047] Furthermore, the K-means clustering algorithm can partition a multidimensional environmental dataset into three regions. Assuming the mean light intensity in region A is 800 lx, in region B is 600 lx, and in region C is 400 lx, combined with the coordinate data, region A is determined to be in a high-light mode, region B is in a medium-light mode, and region C is in a low-light mode. Clustering clearly delineates environmental characteristics. If the daily variation in light intensity in region A exceeds the 100 lx threshold, five additional sampling points can be generated in that region using the Kriging interpolation algorithm. These newly added sampling points collect light data at 850 lx, 830 lx, and so on, providing a more comprehensive dataset for region A.

[0048] Spatiotemporal alignment requires addressing sensor timestamp deviation. Assuming that some sensors have a 3-second timestamp deviation due to network latency, a lightweight long-short-term memory model can be used to predict and correct this deviation by learning the time series patterns of sensor data. When the model analyzes the data sequence over the past hour and identifies a node with a 3-second timestamp deviation, it automatically adjusts its timestamp to align all data to a unified timeline, generating an aligned environmental dataset. If the aligned environmental dataset contains missing or outliers, such as a node failing to record humidity data due to a malfunction or recording an abnormal temperature of 50°C, these can be corrected using an interpolation algorithm.

[0049] The growth stage adaptive control module is used to dynamically control the seedling raising strategy based on the complete environmental data set using the environmental variable coupling method to obtain an adaptive control scheme. The growth stage adaptive control module includes: an environmental variable coupling analysis unit, a control parameter set determination unit, and a dynamic control strategy generation unit. Among them, the environmental variable coupling analysis unit is used to extract the coupling relationship characteristics between light, temperature, and relative air humidity using the environmental parameter coupling analysis algorithm for the complete environmental data set, establish the cross-influence matrix of the three factors of light-temperature-relative air humidity, and quantify the weights of each factor through tensor decomposition to obtain an environmental variable coupling model; the control parameter set determination unit is used to obtain the characteristic weights of environmental variables based on the environmental variable coupling model, and the characteristic weights of environmental variables are combined with the preset crop growth stage division rules to determine the control parameter set corresponding to the current growth stage; the dynamic control strategy generation unit is used to generate dynamic control strategies for light, irrigation, and circulating fans for the control parameter set using a fuzzy logic algorithm to obtain an adaptive control scheme.

[0050] Furthermore, the environmental variable coupling analysis unit includes: a key coupling feature extraction subunit, an environmental variable weight distribution subunit and a partition feature analysis subunit; wherein the key coupling feature extraction subunit is used to normalize the complete environmental data set, and perform principal component analysis on the normalized complete environmental data set to obtain key coupling features; the environmental variable weight distribution subunit is used to construct a cross-influence matrix of three factors of light, temperature and relative air humidity based on the key coupling features, and process the established cross-influence matrix of three factors of light, temperature and relative air humidity through a matrix decomposition method to obtain the weight distribution of environmental variables; the partition feature analysis subunit is used to update the normalized complete environmental data set based on the environmental variable weight distribution, divide the spatial distribution of the coupling features of light, temperature and relative air humidity based on the updated normalized complete environmental data set, and construct a dynamic distribution map of environmental variables based on the spatial distribution of the coupling features of light, temperature and relative air humidity to determine the coupling characteristics of environmental variables in each area of ​​the greenhouse environment.

[0051] As a specific implementation method of this embodiment, the complete environmental data set is normalized to obtain a normalized environmental variable data set. Based on the normalized environmental variable data set, a principal component analysis algorithm is used to extract key coupling features between light, temperature, and relative air humidity to obtain a reduced-dimensional feature data set. A cross-influence matrix of light, temperature, and relative air humidity is constructed from the reduced-dimensional feature data set. The weight coefficient of each environmental variable is calculated using a matrix decomposition method to obtain a variable weight distribution. If the weight coefficient of any environmental variable in the variable weight distribution is lower than a preset threshold, higher-resolution environmental variable data is obtained by increasing the data sampling frequency of the sensor node, and the normalized environmental variable data set is updated. Based on the updated normalized environmental variable data set, a K-means clustering algorithm is used to partition the spatial distribution of the light, temperature, and relative air humidity coupling features to obtain a spatial partitioning model of the environmental variables. From the spatial partitioning model of the environmental variables, the mean of the light, temperature, and relative air humidity coupling features of each partition is extracted to generate a partitioned feature data set. Based on the partitioned feature data set, a dynamic distribution map of the environmental variables is constructed to determine the environmental variable coupling characteristics of each region.

[0052] In one possible implementation, when constructing the cross-influence matrix for light, temperature, and relative humidity, the Pearson correlation coefficient is used to calculate the correlation between variables. The correlation coefficient between light and temperature is 0.8, indicating a strong correlation; the correlation coefficient between light and relative humidity is 0.3, indicating a weak correlation. Matrix decomposition methods such as singular value decomposition can further extract weight coefficients. Assume that the weight for light is 0.5, for temperature 0.4, and for relative humidity 0.1. If the relative humidity weight falls below a preset threshold of 0.2, the sensor sampling frequency is increased from once per hour to every 30 minutes to obtain higher-resolution data.

[0053] Using the K-means clustering algorithm to delineate the spatial distribution of coupled light, temperature, and relative humidity characteristics, farmland can be divided into three zones: a high-light, high-temperature zone, a moderate-light, moderate-temperature zone, and a low-light, low-temperature zone. Assuming the high-light, high-temperature zone has an average light intensity of 30,000 lux, an average temperature of 28 degrees Celsius, and an average relative humidity of 75%, clustering generates a spatial partitioning model of environmental variables, clarifying the differences in characteristics across regions.

[0054] The characteristic mean values ​​of high-intensity and high-temperature areas can be used to generate dynamic distribution maps, showing the spatial trends of light, temperature, and relative humidity. The dynamic distribution maps are presented as heat maps, with red representing high-value areas and blue representing low-value areas, intuitively reflecting the coupled characteristics of farmland environmental variables.

[0055] Furthermore, the control parameter set determination unit includes: a stage prediction subunit, a growth stage classification subunit and an environmental parameter control subunit; wherein the stage prediction subunit is used to generate a weight distribution data set based on the weight distribution of environmental variables, and use a linear regression algorithm to fit the relationship between light intensity, temperature change and relative air humidity and crop growth stage based on the weight distribution data set to obtain a stage prediction model; the growth stage classification subunit is used to judge the crop growth stage corresponding to the current environmental variable based on the stage prediction model, and obtain the current growth stage classification; the environmental parameter control subunit is used to query the preset growth stage and control parameter mapping table based on the current growth stage classification, obtain the corresponding light, temperature and relative air humidity control parameter sets, and generate an environmental control parameter set.

[0056] As a specific implementation method of this embodiment, a weight distribution dataset is generated based on the weight distribution of environmental variables; based on the weight distribution dataset, a linear regression algorithm is used to fit the relationship between light intensity, temperature change and relative air humidity and the crop growth stage to obtain a stage prediction model; if the prediction error of the stage prediction model exceeds a preset threshold, higher resolution environmental variable data is obtained by adjusting the sensor sampling frequency, and the weight distribution dataset is updated; based on the updated weight distribution dataset, a linear regression algorithm is re-used to fit the relationship between light intensity, temperature change and relative air humidity and the crop growth stage to obtain an optimized stage prediction model; from the optimized stage prediction model, combined with a preset crop growth stage division rule, the crop growth stage corresponding to the current environmental variable is determined to obtain the current growth stage classification; based on the current growth stage classification, a preset growth stage and control parameter mapping table is queried to obtain the corresponding light, temperature and relative air humidity control parameter set, and an environmental control parameter set is generated; through the environmental control parameter set, dynamic adjustment instructions for light, temperature and relative air humidity are generated, and a control instruction dataset is output.

[0057] In greenhouse seedling operations for nightshade vegetables (such as tomatoes and peppers), light is weighted at 0.45, temperature at 0.35, and relative humidity at 0.20, forming a weighted distribution dataset. This weighted distribution reflects the dominant role of light in photosynthesis, followed by temperature, and a minor influence of relative humidity on respiration. When using a linear regression algorithm to fit the relationship between light intensity, temperature fluctuations, and relative humidity and crop growth stages, the weighted distribution dataset can be combined with actual crop growth stage data. Tomato seedlings require a light intensity of 8,000 lux, a temperature of 22 degrees Celsius, and a relative humidity of 85%, while flowering requires 10,000 lux, a temperature of 25 degrees Celsius, and a relative humidity of 75%. Using linear regression, the mapping between environmental variables and stages such as seedling and flowering is fitted to form a stage prediction model. If the prediction error of the stage prediction model exceeds a preset threshold, such as if the predicted flowering time is off by more than three days, the sensor sampling frequency can be increased from hourly to every 30 minutes to obtain higher-resolution data. The optimized stage prediction model is combined with the crop growth stage division rules to determine the growth stage corresponding to the current environmental variables.

[0058] When querying the growth stage and control parameter mapping table, you can categorize by flowering stage to obtain control parameter sets, such as adjusting the light intensity to 10,000 lux, maintaining the temperature at 25 degrees Celsius, and increasing the relative humidity to 75%. Control instruction data sets include: starting the fill light to 10,000 lux, turning on the sprinkler system or circulating fan to maintain 25 degrees Celsius and 75% relative humidity. Instructions must take into account device response time, such as requiring the fill light to reach the target light intensity within 5 minutes, to ensure smooth control execution.

[0059] The virtual feedback simulation module is used to calibrate and control the virtual feedback mechanism with the actual acquisition device to obtain an optimized control solution. The virtual feedback simulation module includes: a simulated growth unit, an optimized control solution generation unit, and a user-friendly unit;

[0060] As a specific implementation method of this embodiment, a fuzzy logic algorithm is used to generate a preliminary control strategy for light, irrigation, and relative air humidity based on the environmental variable data set and preset crop demand rules. From the execution parameter set, combined with real-time monitoring data, it is determined whether the light, irrigation, and relative air humidity meet the crop needs to obtain a control status assessment. Based on the control status assessment, the preset feedback rules are used to adjust the input weights of the fuzzy logic algorithm and optimize the adaptive control scheme.

[0061] Furthermore, as a specific implementation of this embodiment, the process of obtaining an optimized control scheme based on a virtual feedback simulation module includes: a simulated growth unit, an optimized control scheme generation unit, and a user-friendly unit. The simulated growth unit is used to execute an adaptive control scheme in a preset virtual farmland environment through a virtual feedback mechanism to obtain simulated crop growth state data; the optimized control scheme generation unit is used to extract key indicators from the simulated growth state data, compare the key indicators of the simulated growth state data with the crop growth data collected by actual sensors, calibrate the control parameters through an error analysis algorithm, and obtain an optimized control scheme; the user-friendly unit is used to feed back the optimized control scheme to the user, who then manages the seedling greenhouse environment based on the optimized control scheme.

[0062] Furthermore, as a specific implementation of this embodiment, an adaptive control scheme is executed in a preset virtual farmland environment through a virtual feedback mechanism to obtain simulated crop growth status data. Environmental variable features are extracted from the simulated crop growth status data, and growth stages are divided using a k-means clustering algorithm to obtain a stage division result. Based on the stage division result and in combination with preset crop requirement rules, preliminary control parameters for light, irrigation, and relative air humidity are generated, and a control parameter set is determined. If the deviation between the control parameter set and the real-time environmental variables in the virtual farmland environment exceeds a preset threshold, the sampling frequency of the virtual feedback mechanism is adjusted to update the environmental variable features and obtain updated simulated growth status data. Key growth indicators are extracted from the updated simulated growth status data, and a linear regression algorithm is used to predict the growth trend for the next stage, obtaining a growth trend forecast. Based on the growth trend forecast, the control parameters for light, irrigation, and relative air humidity are optimized, dynamic adjustment instructions are generated, and an execution instruction set is determined. By executing the instruction set, the optimized control scheme is executed in the virtual farmland environment to obtain new simulated crop growth status data, determine whether the control status meets the crop requirements, and obtain a status assessment result.

[0063] Environmental variable features are extracted from the simulated growth state data. A decision tree algorithm is used to divide crop growth stages and determine the growth stage division results. Based on the growth stage division results and in combination with preset crop requirement rules, preliminary lighting, irrigation, and fertilization control parameters are generated to determine the initial control parameter set. If the deviation between the initial control parameter set and the real-time environmental variables exceeds a preset threshold, the data sampling frequency is adjusted to update the environmental variable features and obtain updated simulated growth state data. Key growth indicators are extracted from the updated simulated growth state data. A mean square error algorithm is used to calculate the deviation from the actual growth data collected by the sensors and determine the deviation calibration results. Based on the deviation calibration results, the lighting, irrigation, and fertilization control parameters are adjusted, dynamic adjustment instructions are generated, and the optimized control parameter set is determined. Based on the optimized control parameter set, execution instructions for the virtual farmland environment are generated, new simulated growth state data is obtained, and whether the control state meets the crop requirement rules is determined to determine the state evaluation results. Environmental variable features that do not meet the crop requirement rules are extracted from the state evaluation results, and the training data for the decision tree algorithm is updated to determine the new growth stage division results.

[0064] Environmental response characteristics are extracted from the actuator's operating status data. A time series analysis algorithm is used to identify operating status trends and determine an operating status classification result. Based on the operating status classification results and in combination with pre-set networking rules, preliminary network topology adjustment parameters are generated to determine the initial networking configuration. If the deviation between the initial networking configuration and the real-time data stream exceeds a preset threshold, the data collection frequency is adjusted to update the environmental response characteristics and obtain updated operating status data. Key response indicators are extracted from the updated operating status data, and the deviation from the pre-set networking rules is calculated using a mean square error algorithm to determine the deviation calibration result. Based on the deviation calibration result, the network topology parameters are adjusted, dynamic networking adjustment instructions are generated, and the optimized networking configuration is determined. Based on the optimized networking configuration, operating instructions for the actuator are generated, and new operating status data is obtained. The network status is then assessed to determine whether it meets the operating rules of the intelligent agricultural system and a status assessment result is determined. Environmental response characteristics that do not meet the operating rules are extracted from the status assessment result, and the training data for the time series analysis algorithm is updated to determine a new operating status classification result.

[0065] The above are merely preferred embodiments of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.

Claims

1. An intelligent control system for seedling greenhouse environment based on the Internet of Things, characterized in that: include: A heterogeneous sensor dynamic networking construction module is used to deploy self-locating wireless sensor nodes and obtain a dynamic networking solution based on the wireless sensor nodes; An environmental data collection and preprocessing module is used to collect a multidimensional environmental data set and preprocess the multidimensional environmental data set to obtain a complete environmental data set; A growth stage adaptive control module is used to dynamically control the seedling raising strategy based on the complete environmental data set using an environmental variable coupling method to obtain an adaptive control scheme; The virtual feedback simulation module is used to calibrate and control the actual acquisition device through the virtual feedback mechanism to obtain an optimized control solution.

2. The intelligent control system for seedling greenhouse environment based on the Internet of Things according to claim 1 is characterized in that: The heterogeneous sensor dynamic networking construction module includes: a sensor node arrangement unit and a network structure adjustment unit; Wherein, the sensor node arrangement unit obtains the connection relationship between wireless sensor nodes based on LoRa+BLE dual-mode communication; The network structure adjustment unit generates an initial mesh network based on the connection relationship between sensor nodes based on a topology reconstruction algorithm, and adjusts the connection relationship between wireless sensor nodes according to the terrain and crop distribution to obtain a dynamic networking solution.

3. The intelligent control system for seedling greenhouse environment based on Internet of Things according to claim 1 is characterized in that: The environmental data acquisition and preprocessing module includes: a data acquisition unit, a first processing unit and a second processing unit; The data acquisition unit is used to extract sensor node locations from the dynamic networking scheme and collect multi-dimensional environmental data sets using a layered sampling method; The first processing unit is used to perform partition processing on the multidimensional environmental data set by using a K-means clustering algorithm to determine the environmental pattern of each area in the seedling greenhouse environment; The second processing unit is used to perform spatiotemporal alignment and outlier repair on the multidimensional environmental dataset based on the environmental pattern of each region using the built-in lightweight crop growth model of the wireless sensor node to obtain a complete environmental dataset.

4. The intelligent control system for seedling greenhouse environment based on the Internet of Things according to claim 1 is characterized in that: The growth phase adaptive control module includes: an environmental variable coupling analysis unit, a control parameter set determination unit, and a dynamic control strategy generation unit; The environmental variable coupling analysis unit is used to extract the coupling relationship characteristics between light, temperature, and relative air humidity using an environmental parameter coupling analysis algorithm for the complete environmental data set, establish a cross-influence matrix of the three factors of light, temperature, and relative air humidity, and quantify the weight of each factor through tensor decomposition to obtain an environmental variable coupling model; The control parameter set determination unit is configured to obtain characteristic weights of environmental variables based on the environmental variable coupling model, and determine a control parameter set corresponding to a current growth stage by combining the characteristic weights of the environmental variables with a preset crop growth stage division rule; The dynamic control strategy generation unit is used to generate dynamic control strategies for lighting, irrigation, and circulating fans using a fuzzy logic algorithm for a control parameter set to obtain an adaptive control solution.

5. The intelligent control system for seedling greenhouse environment based on Internet of Things according to claim 4 is characterized in that: The environmental variable coupling analysis unit includes: a key coupling feature extraction subunit, an environmental variable weight distribution subunit and a partition feature analysis subunit; The key coupling feature extraction subunit is used to normalize the complete environment data set and perform principal component analysis on the normalized complete environment data set to obtain key coupling features; The environmental variable weight distribution subunit is used to construct a cross-influence matrix of three factors: light, temperature, and relative air humidity based on the key coupling characteristics, and process the established cross-influence matrix of three factors: light, temperature, and relative air humidity by a matrix decomposition method to obtain an environmental variable weight distribution; The partition feature analysis subunit is used to update the normalized complete environmental data set based on the environmental variable weight distribution, divide the spatial distribution of the light, temperature, and relative air humidity coupling characteristics based on the updated normalized complete environmental data set, and construct a dynamic distribution map of the environmental variables based on the spatial distribution of the light, temperature, and relative air humidity coupling characteristics to determine the environmental variable coupling characteristics of each area in the greenhouse environment.

6. The intelligent control system for seedling greenhouse environment based on Internet of Things according to claim 5 is characterized in that: The control parameter set determination unit includes: a stage prediction subunit, a growth stage classification subunit and an environmental parameter control subunit; The stage prediction subunit is used to generate a weight distribution data set based on the weight distribution of the environmental variables, and to use a linear regression algorithm to fit the relationship between light intensity, temperature change, and relative humidity of the air and the crop growth stage based on the weight distribution data set to obtain a stage prediction model; The growth stage classification subunit is used to determine the crop growth stage corresponding to the current environmental variable based on the stage prediction model to obtain the current growth stage classification; The environmental parameter control subunit is used to query a preset growth stage and control parameter mapping table based on the current growth stage classification, obtain corresponding light, temperature and air relative humidity control parameter sets, and generate an environmental control parameter set.

7. The intelligent control system for seedling greenhouse environment based on Internet of Things according to claim 1 is characterized in that: The virtual feedback simulation module includes: a simulation growth unit, an optimization control scheme generation unit and a user-friendly unit; The simulated growth unit is used to run an adaptive control scheme in a preset virtual farmland environment through a virtual feedback mechanism to obtain simulated crop growth status data; The optimized control scheme generating unit is used to extract key indicators from the simulated growth state data, compare the key indicators of the simulated growth state data with the crop growth data collected by the actual sensor, calibrate the control parameters through an error analysis algorithm, and obtain an optimized control scheme; The user-friendly unit is used to feed back the optimization control plan to the user, and the user manages the seedling greenhouse environment based on the optimization control plan.

Images