Water and fertilizer integrated management system based on greenhouse internet of things

Abstract

The invention relates to the technical field of agricultural Internet of Things, in particular to a water and fertilizer integrated management system based on greenhouse Internet of Things. A water and fertilizer strategy auxiliary mechanism containing a flower feature extraction unit, a sensor feature extraction unit and an auxiliary strategy output unit is constructed to generate a feature detection demand and a task detection demand, a total detection signal number is obtained, and a monitoring frequency interval is matched to generate a parameter optimization strategy. Fine monitoring requirements of different flowers and planting tasks can be met, and limitation of a universal strategy is reduced; real-time reference spectral data of a sensor are collected through a preset automatic calibration mechanism, spectral deviation is calculated, a sampling deviation index is obtained, a sudden change sensor is marked by analyzing parameter fluctuation of the sensor in a stable state time period, then a time sequence sudden change frequency is obtained, a comprehensive abnormal value is output through calculation, and the sensor to be calibrated is accurately recognized through the comprehensive abnormal value. Manual calibration dependence is reduced, sensor data precision is improved, and influence of unreliable data on water and fertilizer management effects is avoided.

Description

Technical Field

[0001] This invention relates to the field of agricultural Internet of Things (IoT) technology, and in particular to an integrated water and fertilizer management system based on greenhouse IoT. Background Technology

[0002] As urban residents' living standards improve, the demand for flowers in the market is not only large but also trending towards high-end products. As a result, the demand for planting various types of flowers has increased significantly. At the same time, the Internet of Things (IoT) technology is gradually maturing, providing technical support for real-time monitoring and remote control of greenhouse environments. This promotes the transformation of integrated water and fertilizer management from manual and extensive methods to intelligent and precise methods, in order to solve the problems of waste of agricultural production resources and low efficiency, and to meet the needs of refined propagation of various specialty flowers.

[0003] Current integrated water and fertilizer management systems monitor soil nutrients and moisture in real time with high precision. However, existing soil sensors are susceptible to environmental factors, resulting in low measurement accuracy and frequent calibration requirements. They cannot provide reliable data in real time, and the monitoring strategies available in the system are only general-purpose. Users still need to make adaptive settings to develop personalized monitoring strategies, which requires a high level of expertise. Summary of the Invention

[0004] This invention provides a water and fertilizer integrated management system based on greenhouse Internet of Things (IoT) to solve the aforementioned technical problems existing in the prior art.

[0005] The first aspect of the present invention provides a water and fertilizer integrated management system based on greenhouse Internet of Things, including a data sensing layer, an Internet of Things communication layer, a processing and decision-making layer and a terminal execution layer.

[0006] The data perception layer uses the obtained parameter optimization strategy to collect and adjust the preset multimodal sensor group to acquire water and fertilizer multimodal data, and obtain the sensor operation data of the multimodal sensor group.

[0007] The decision-making processing layer includes a water and fertilizer strategy generation and processing module and an automatic calibration module. Based on the IoT communication layer, it acquires multi-mode water and fertilizer data and sensor operation data, and inputs the multi-mode water and fertilizer data and sensor operation data into the water and fertilizer strategy generation and processing module to obtain parameter optimization strategies. The sensor operation data is input into the automatic calibration module to obtain the sensor to be calibrated.

[0008] As a further improvement of the present invention, the water and fertilizer strategy generation and processing module specifically executes as follows: the received water and fertilizer multi-mode data and sensor operation data are input into a preset water and fertilizer strategy auxiliary mechanism, the water and fertilizer strategy auxiliary mechanism performs strategy matching analysis on the water and fertilizer multi-mode data and sensor operation data, outputs parameter optimization strategy, and sends the parameter optimization strategy to the terminal execution layer.

[0009] The water and fertilizer strategy support mechanism includes a flower feature extraction unit, a sensor feature extraction unit, and an auxiliary strategy output unit; Among them, the flower feature extraction unit obtains feature detection requirements and task detection requirements based on water and fertilizer multi-modal data. Specifically, it obtains the current flower category based on water and fertilizer multi-modal data, and crawls the flower physiological feature data of the corresponding category from the flower knowledge base in the database based on the current flower category. The flower physiological feature data includes ideal environmental parameters, physiological response dynamics characteristics and planting task level. Based on the physiological response dynamics, the corresponding soil moisture response time, nutrient absorption rate, and water evapotranspiration are obtained to obtain the maximum fluctuation rate of soil water and fertilizer and the environmental recovery response time. The ideal environmental tolerance of the corresponding flowers is obtained based on the database. Under the condition of equal changes in environmental parameters, the maximum fluctuation rate of soil water and fertilizer is the maximum rate of decrease of the index corresponding to each ideal environmental parameter of the flower. The environmental recovery response time is the time it takes for the index of the corresponding environmental parameter to recover to the ideal environmental parameter from the time when the actuator starts to perform water and fertilizer management. The maximum fluctuation rate of soil water and fertilizer is divided into three maximum fluctuation rate intervals based on a preset fluctuation rate interval. Fluctuation rate levels are generated by arranging the rates within each maximum fluctuation rate interval in descending order: high fluctuation level, normal fluctuation level, and low fluctuation level. Similarly, the environmental recovery response time is divided into three environmental recovery response time intervals and arranged in descending order to generate environmental recovery levels: high recovery level, normal recovery level, and low recovery level. Feature indices are assigned to the fluctuation rate levels, with p3, p2, and p1 corresponding to high, normal, and low fluctuation levels, respectively. Simultaneously, feature indices are assigned to the environmental recovery levels, with p1, p2, and p3 corresponding to high, normal, and low recovery levels, respectively (0 < p1 < p2 < p3). The feature indices for each flower category's fluctuation rate level and environmental recovery level are summed to obtain a total feature index. Feature detection requirements are generated based on the magnitude of the total feature index. Feature detection requirements are positively correlated with the value of the total feature index, i.e., feature detection requirements include high, medium, and low requirements. Based on the importance of each planting task level, corresponding task detection requirements are generated. The task detection requirements are positively correlated with the importance of the task level. That is, when the planting task level is scientific research level, the task detection requirements are high; when the planting task level is premium production level, the task detection requirements are medium; and when the planting task level is ordinary production level, the task detection requirements are low.

[0010] The sensor feature extraction unit obtains the monitoring frequency range based on sensor operating data, feature detection requirements, and task detection requirements, specifically as follows: The auxiliary strategy output unit obtains the monitoring frequency range corresponding to each environmental parameter of each flower category, outputs the corresponding parameter monitoring frequency strategy based on the monitoring frequency range of each environmental parameter, and combines the parameter monitoring frequency strategies of each sensor to obtain the parameter optimization strategy.

[0011] Based on sensor operating data, the sensor's monitoring capability range, i.e., the adjustable range of monitoring frequency, is obtained. The monitoring frequency capability range is divided into multiple monitoring frequency ranges based on the pre-designed monitoring frequency intervals, namely, high-frequency monitoring range, medium-frequency monitoring range, and low-frequency monitoring range. The feature detection requirements and task detection requirements of each environmental parameter corresponding to the current flower category are obtained. The feature detection requirements and task detection requirements are quantified through a preset requirement quantification mechanism to obtain the number of feature detection signals and the number of task detection signals, respectively. The total number of detected signals is calculated by substituting the number of feature detection signals and the number of task detection signals into the preset two-factor coupling calculation formula. The preset signal number intervals corresponding to the high-frequency monitoring interval, the medium-frequency monitoring interval, and the low-frequency monitoring interval are obtained as the high-frequency signal number interval, the medium-frequency signal number interval, and the low-frequency signal number interval, respectively. The total number of detected signals corresponding to the current environmental parameters of flowers is matched with each signal number interval to obtain the corresponding monitoring frequency interval.

[0012] As a further improvement of the present invention, the automatic calibration module specifically performs the following steps: inputting the received sensor operating data into a preset automatic calibration mechanism, and the automatic calibration mechanism performs calibration analysis on the sensor operating data to obtain the sensor to be calibrated.

[0013] Furthermore, the specific execution of the automatic calibration mechanism is as follows: Based on the sensor operation data corresponding to each sensor, a preset number of real-time reference spectral values ​​are obtained according to the sensor operation data. The real-time reference spectral data is collected through a preset spectral extraction mechanism. The real-time reference spectral data is compared with the preset reference standard spectral values, and the difference between the real-time reference spectral data and the reference standard spectral values ​​is calculated to obtain the spectral deviation value. When the spectral deviation value exceeds the preset spectral deviation threshold, the value of the spectral deviation value exceeding the spectral deviation threshold is recorded as the sampling deviation index. Based on the preset analysis time period and sensor operation data, the various environmental parameters corresponding to the preset number of flowers in the current analysis time period are obtained. The various environmental parameters are identified. When the environmental parameter is within the preset parameter fluctuation range, the corresponding environmental parameter is marked as a steady-state parameter. The number of steady-state parameters is counted to obtain the total number of environmental stable parameters. The ratio of the number of environmental stable parameters to the total number of environmental parameters is calculated to obtain the environmental stability ratio. When the environmental stability ratio exceeds the preset stability upper limit ratio, the corresponding analysis time period is marked as a steady-state period. The system acquires the parameter values ​​of similar sensors for each environmental parameter during a steady-state period, and obtains the numerical fluctuation of each parameter value. When the numerical fluctuation exceeds a preset fluctuation threshold, the corresponding sensor is marked as a sudden change sensor. Using the sudden change sensor as the center, the system acquires similar sensors within a preset distance radius corresponding to the sudden change sensor. When none of the similar sensors within the nearby range are marked as sudden change sensors, the system generates the temporal sudden change frequency of the sudden change sensor, which corresponds to the number of times the numerical fluctuation exceeds the fluctuation threshold during the analysis period. Conversely, when a sudden change sensor appears in a similar sensor within the nearby range, the system monitors the numerical fluctuation of each sudden change sensor in real time to obtain the duration of the fluctuation. When the duration of the fluctuation exceeds a preset upper limit, the sudden change sensor marking of each sensor is removed; otherwise, the temporal sudden change frequency of each sudden change sensor is obtained.

[0014] The sampling deviation index and the temporal mutation frequency are both normalized to map their values ​​to the [0,1] interval, and the normalized values ​​are denoted as the sampling deviation analysis value and the temporal mutation analysis value, namely Cy and Sx, respectively. The deviation index membership degree and the temporal mutation membership degree are calculated by using the S-shaped fuzzy membership function. The sampling quantity corresponding to the two analysis values ​​is taken as the fuzzy entropy, and two weight coefficients are calculated by using the preset weight calculation formula for the fuzzy entropy and membership degree of the two analysis values. The deviation index membership degree and the temporal mutation membership degree and the corresponding weight coefficients are calculated by using the comprehensive calculation formula to obtain the comprehensive outlier value. When the comprehensive outlier value is greater than the preset threshold, the corresponding sensor is marked as a sensor to be calibrated.

[0015] The terminal execution layer includes a preset smart gateway controller and an actuator. Based on the IoT communication layer, it receives parameter optimization strategies and sensors to be calibrated, feeds back the parameter optimization strategies to the data sensing layer, and executes a preset calibration procedure on the sensors to be calibrated. The IoT communication layer provides the network communication foundation for the data sensing layer, processing and decision-making layer, and terminal execution layer through deployed IoT communication technologies. Specifically, the data sensing layer sends multi-mode water and fertilizer data and sensor operation data to the processing and decision-making layer through the connection of the IoT communication layer; it receives parameter optimization strategies and sensors to be calibrated, and sends the parameter optimization strategies and sensors to be calibrated to the terminal execution layer respectively; it also obtains operation adjustment instructions and feeds the operation adjustment instructions back to the data sensing layer. As a further improvement of the present invention, the demand quantification mechanism is as follows: obtain the demand level corresponding to the feature detection demand, assign the number of detection signals according to the order of the demand level, the number of detection signals increases as the demand level increases, and the corresponding number of detection signals is recorded as the feature detection signal number; similarly, obtain the demand level corresponding to the task detection demand, and then assign the number of detection signals accordingly, and the corresponding number of detection signals is recorded as the task detection signal number.

[0016] As a further improvement of the present invention, the spectral extraction mechanism is as follows: the sample liquid acquired by the probe is pumped into a pre-set reference cell by a micro-pump built into the sensor detection probe, and the sample liquid in the reference cell is detected in real time by a micro-spectrometer to obtain real-time reference spectral data; the reference standard spectral value is the standard spectral value corresponding to various environmental parameters obtained by experiments under standard laboratory conditions.

[0017] The technical solution provided by this invention has the following advantages compared with the prior art: 1. This invention generates feature detection requirements and task detection requirements by constructing a water and fertilizer strategy auxiliary mechanism containing flower feature extraction, sensor feature extraction and auxiliary strategy output units, and obtains the total number of detection signals. It then matches the monitoring frequency range to generate parameter optimization strategies, which can adapt to the refined monitoring requirements of different flowers and planting tasks and reduce the limitations of general strategies. 2. This invention collects real-time reference spectral data from sensors through a preset automatic calibration mechanism to calculate the spectral deviation and obtain the sampling deviation index. It also analyzes the fluctuations in sensor parameters during the steady-state period to mark abruptly changing sensors and obtain the temporal abrupt change frequency. The invention calculates and outputs a comprehensive outlier value, which accurately identifies the sensor to be calibrated, reduces reliance on manual calibration, improves the accuracy of sensor data, and avoids unreliable data affecting the water and fertilizer management effect. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the principle of the present invention; Figure 2 This is a block diagram illustrating the principle of the processing decision layer of the present invention. Detailed Implementation

[0019] The technical solution of the present invention will now be clearly and completely described in conjunction with the accompanying drawings and specific embodiments.

[0020] For ease of understanding, the specific process of the embodiments of the present invention will be described below. For example... Figure 1-2 As shown, one embodiment of the integrated water and fertilizer management system based on the Internet of Things for greenhouses in this invention includes: The data sensing layer, based on the obtained parameter optimization strategy, collects and adjusts the pre-set multimodal sensor group to acquire multimodal water and fertilizer data, and obtains the sensor operation data of the multimodal sensor group. The multimodal sensor group includes, but is not limited to, a collection of sensors for soil temperature and humidity, EC value, and ambient temperature, which can simultaneously collect multi-dimensional water and fertilizer related data. It should be noted that: EC value is soil electrical conductivity, an indicator for measuring water-soluble salts in soil.

[0021] The decision-making processing layer includes a water and fertilizer strategy generation and processing module and an automatic calibration module. Based on the IoT communication layer, it acquires multi-mode water and fertilizer data and sensor operation data, and inputs the multi-mode water and fertilizer data and sensor operation data into the water and fertilizer strategy generation and processing module to obtain parameter optimization strategies. The sensor operation data is input into the automatic calibration module to obtain the sensor to be calibrated.

[0022] The water and fertilizer strategy generation and processing module inputs the received water and fertilizer multi-mode data and sensor operation data into the preset water and fertilizer strategy auxiliary mechanism. The water and fertilizer strategy auxiliary mechanism performs strategy matching analysis on the water and fertilizer multi-mode data and sensor operation data, outputs parameter optimization strategy, and sends the parameter optimization strategy to the terminal execution layer.

[0023] The specific implementation details of the water and fertilizer strategy support mechanism are as follows: The water and fertilizer strategy support mechanism includes a flower feature extraction unit, a sensor feature extraction unit, and an auxiliary strategy output unit.

[0024] Furthermore, the flower feature extraction unit obtains the current flower category based on water and fertilizer multi-model data, and crawls the corresponding flower physiological feature data from the flower knowledge base in the database based on the current flower category. The flower physiological feature data includes ideal environmental parameters, physiological response dynamics characteristics, and planting task level; ideal environmental parameters include but are not limited to the target range of soil temperature and humidity, the target range of pH, and the optimal diurnal temperature range; physiological response dynamics characteristics include but are not limited to soil moisture response time, nutrient absorption rate, and water evapotranspiration; planting task level includes research level, premium production level, and ordinary production level based on the importance of the planting task.

[0025] Based on the physiological response dynamics, the corresponding soil moisture response time, nutrient absorption rate, and water evaporation are obtained to obtain the maximum fluctuation rate of soil water and fertilizer and the environmental recovery response time. Furthermore, based on the database, the ideal environmental tolerance (i.e., the parameter fluctuation range for healthy growth of each type of flower) is obtained. Under the condition of equal changes in environmental parameters, the maximum fluctuation rate of soil water and fertilizer is the maximum rate of decrease of the index corresponding to each ideal environmental parameter of the flower. The environmental recovery response time is the time it takes for the index of the corresponding environmental parameter to recover to the ideal environmental parameter when the actuator starts water and fertilizer management. For example, the maximum fluctuation rate of soil water and fertilizer includes the maximum rate of decrease of soil moisture of flowers under strong light or high temperature evaporation environment. Therefore, the environmental recovery response time is the time it takes for the soil moisture index corresponding to the moment when the irrigation actuator starts irrigation to recover to the ideal environmental moisture. The maximum fluctuation rate of soil water and fertilizer is divided into three maximum fluctuation rate intervals based on a preset fluctuation rate interval. Fluctuation rate levels are generated by arranging the rates within each maximum fluctuation rate interval in descending order of numerical value: high fluctuation level, normal fluctuation level, and low fluctuation level. Similarly, the environmental recovery response time is divided into three environmental recovery response time intervals and arranged in descending order to generate environmental recovery levels: high recovery level, normal recovery level, and low recovery level. Characteristic indices are assigned to the fluctuation rate levels, with p3, p2, and p1 corresponding to the high, normal, and low fluctuation levels, respectively. Simultaneously, characteristic indices are assigned to the environmental recovery levels, resulting in high, normal, and low recovery levels. The corresponding feature indices are p1, p2, and p3 (0 < p1 < p2 < p3). The feature indices corresponding to the fluctuation rate level and environmental recovery level of each flower category are summed to obtain the total feature index. Feature detection requirements are generated based on the magnitude of the total feature index. The feature detection requirements are positively correlated with the value of the total feature index. That is, the feature detection requirements include high feature requirements (p3+p3), medium feature requirements (p3+p2, p2+p2), and low feature requirements (p1+p2, p1+p1). For example, the fluctuation rate level and environmental recovery level of a certain flower category are high fluctuation level and low recovery level, respectively. Therefore, the total feature index is p3+p3. Thus, its corresponding feature detection requirement is a high feature requirement.

[0026] Based on the importance of each planting task level, corresponding task detection requirements are generated. The task detection requirements are positively correlated with the importance of the task level. That is, when the planting task level is scientific research level, the task detection requirements are high; when the planting task level is premium production level, the task detection requirements are medium; and when the planting task level is ordinary production level, the task detection requirements are low.

[0027] Furthermore, the sensor feature extraction unit obtains the sensor's monitoring capability range, i.e., the adjustable range of monitoring frequency, based on the sensor's operating data. The monitoring frequency capability range is divided into multiple monitoring frequency ranges based on the pre-designed monitoring frequency intervals, namely, high-frequency monitoring range, medium-frequency monitoring range, and low-frequency monitoring range. The feature detection requirements and task detection requirements corresponding to each environmental parameter of the current flower category are obtained. The feature detection requirements and task detection requirements are quantified through a preset requirement quantification mechanism to obtain the number of feature detection signals and the number of task detection signals, respectively. The demand quantification mechanism is as follows: Obtain the demand level (high demand, medium demand, low demand) corresponding to the feature detection demand. Assign a number of detection signals based on the order of demand level, with the number of detection signals increasing as the demand level increases. This number of detection signals is recorded as the feature detection signal count. Similarly, obtain the demand level (high demand, medium demand, low demand) corresponding to the task detection demand, and assign a number of detection signals accordingly. This number of detection signals is recorded as the task detection signal count. For example, the number of detection signals assigned to high demand, medium demand, and low demand are 5, 3, and 1, respectively. Substitute the number of feature detection signals and the number of task detection signals into the preset two-factor coupling calculation formula. The total number of detected signals Rz is calculated; where Tz and Rw are the number of feature-detected signals and the number of task-detected signals, respectively. The preset synergistic weighting coefficients are generated through experiments based on flower varieties and growth stages. The minimum value is preset to avoid the denominator being zero; the preset signal number intervals corresponding to the high frequency monitoring interval, the medium frequency monitoring interval, and the low frequency monitoring interval are obtained as the high frequency signal number interval, the medium frequency signal number interval, and the low frequency signal number interval, respectively. The total number of detected signals corresponding to various environmental parameters of the current flower is matched with each signal number interval to obtain the corresponding monitoring frequency interval.

[0028] Furthermore, the auxiliary strategy output unit obtains the monitoring frequency range corresponding to each environmental parameter of each flower category, outputs the corresponding parameter monitoring frequency strategy based on the monitoring frequency range of each environmental parameter, and combines the parameter monitoring frequency strategies of each sensor to obtain the parameter optimization strategy.

[0029] The automatic calibration module inputs the received sensor operating data into a preset automatic calibration mechanism, which then performs calibration analysis on the sensor operating data to obtain the sensor to be calibrated.

[0030] The specific execution of the automatic calibration mechanism is as follows: Based on the sensor operation data corresponding to each sensor, a preset number of real-time reference spectral values ​​are obtained according to the sensor operation data. The real-time reference spectral data is collected through a preset spectral extraction mechanism. The real-time reference spectral data is compared with the preset reference standard spectral values, and the difference between the real-time reference spectral data and the reference standard spectral values ​​is calculated to obtain the spectral deviation value. When the spectral deviation value exceeds the preset spectral deviation threshold, the value of the spectral deviation value exceeding the spectral deviation threshold is recorded as the sampling deviation index. The spectral extraction mechanism uses a built-in micro-pump in the sensor detection probe to pump the sample liquid acquired by the probe into a pre-set reference cell. The micro-spectrometer then performs real-time detection on the sample liquid in the reference cell to obtain real-time reference spectral data. The reference standard spectral values ​​are obtained by experiments under standard laboratory conditions, corresponding to various environmental parameters.

[0031] Based on the preset analysis time period and sensor operation data, the various environmental parameters corresponding to the preset number of flowers in the current analysis time period are obtained. The various environmental parameters are identified. When the environmental parameter is within the preset parameter fluctuation range, the corresponding environmental parameter is marked as a steady-state parameter. The number of steady-state parameters is counted to obtain the total number of environmental stable parameters. The ratio of the number of environmental stable parameters to the total number of environmental parameters is calculated to obtain the environmental stability ratio. When the environmental stability ratio exceeds the preset stability upper limit ratio, the corresponding analysis time period is marked as a steady-state period. The system acquires the parameter values ​​of similar sensors for each environmental parameter during a steady-state period, and obtains the numerical fluctuation of each parameter value. When the numerical fluctuation exceeds a preset fluctuation threshold, the corresponding sensor is marked as a sudden change sensor. Using the sudden change sensor as the center, the system acquires similar sensors within a preset distance radius corresponding to the sudden change sensor. When none of the similar sensors within the nearby range are marked as sudden change sensors, the system generates the temporal sudden change frequency of the sudden change sensor, which corresponds to the number of times the numerical fluctuation exceeds the fluctuation threshold during the analysis period. Conversely, when a sudden change sensor appears in a similar sensor within the nearby range, the system monitors the numerical fluctuation of each sudden change sensor in real time to obtain the duration of the fluctuation. When the duration of the fluctuation exceeds a preset upper limit, the sudden change sensor marking of each sensor is removed; otherwise, the temporal sudden change frequency of each sudden change sensor is obtained.

[0032] The sampling deviation index and the temporal mutation frequency are both normalized to map their values ​​to the [0,1] interval, and the normalized values ​​are denoted as the sampling deviation analysis value and the temporal mutation analysis value, namely Cy and Sx, respectively. The deviation index membership degree and the temporal mutation membership degree are calculated by using the S-shaped fuzzy membership function. The sampling quantity corresponding to the two analysis values ​​is taken as the fuzzy entropy, and two weight coefficients are calculated by using the preset weight calculation formula for the fuzzy entropy and membership degree of the two analysis values. The deviation index membership degree and the temporal mutation membership degree and the corresponding weight coefficients are calculated by using the comprehensive calculation formula to obtain the comprehensive outlier value. When the comprehensive outlier value is greater than the preset threshold, the corresponding sensor is marked as a sensor to be calibrated.

[0033] The formula used in the above analysis is as follows: S-shaped fuzzy membership function: , ; Weight calculation formula: , ; Comprehensive calculation formula: ; Among them, L C L S These are the membership degree of the deviation index and the membership degree of the temporal abrupt change, respectively. These are the preset membership weight coefficients. These are the preset normal threshold values ​​corresponding to the sampling deviation analysis value and the time series abrupt change analysis value, respectively; M S M C These are the fuzzy entropies corresponding to the two analytical values. These are the weighting coefficients corresponding to the analyzed values; A CS This is represented as a composite outlier.

[0034] The terminal execution layer includes a preset smart gateway controller and actuators. Based on the IoT communication layer, it receives parameter optimization strategies and sensors to be calibrated, and feeds back the parameter optimization strategies to the data perception layer. The parameter optimization strategies are sampling detection frequency adjustment commands. It executes preset calibration procedures on the sensors to be calibrated, such as sending early warning information and starting automatic calibration mechanisms.

[0035] The IoT communication layer provides a network communication foundation for the data perception layer, processing and decision-making layer, and terminal execution layer through the deployed IoT communication technology. Specifically, the data perception layer sends water and fertilizer multi-mode data and sensor operation data to the processing and decision-making layer through the connection of the IoT communication layer; receives parameter optimization strategies and sensors to be calibrated, sends the parameter optimization strategies and sensors to be calibrated to the terminal execution layer respectively, obtains operation adjustment instructions, and feeds back the operation adjustment instructions to the data perception layer.

[0036] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit them; those skilled in the art can modify or make equivalent substitutions to the technical solutions described in the foregoing embodiments; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A water and fertilizer integrated management system based on greenhouse Internet of Things, characterized in that, include: The data sensing layer collects and adjusts the preset multimodal sensor group based on the obtained parameter optimization strategy to acquire water and fertilizer multimodal data, and obtains the sensor operation data of the multimodal sensor group. The decision-making layer includes a water and fertilizer strategy generation and processing module and an automatic calibration module; The system acquires multi-mode water and fertilizer data and sensor operation data based on the IoT communication layer, and inputs the multi-mode water and fertilizer data and sensor operation data into the water and fertilizer strategy generation and processing module to obtain parameter optimization strategies. The sensor's operating data is input into the automatic calibration module to obtain the sensor to be calibrated; The terminal execution layer includes a preset smart gateway controller and actuators. Based on the IoT communication layer, it receives parameter optimization strategies and sensors to be calibrated, feeds back the parameter optimization strategies to the data sensing layer, and executes preset calibration procedures on the sensors to be calibrated. The IoT communication layer provides the network communication foundation for the data sensing layer, processing and decision-making layer, and terminal execution layer through the deployment of IoT communication technologies.

2. The integrated water and fertilizer management system based on greenhouse IoT as described in claim 1, characterized in that, The specific execution of the water and fertilizer strategy generation and processing module is as follows: inputting the received water and fertilizer multi-mode data and sensor operation data into a preset water and fertilizer strategy auxiliary mechanism, performing strategy matching analysis on the water and fertilizer multi-mode data and sensor operation data through the water and fertilizer strategy auxiliary mechanism, outputting parameter optimization strategies, and sending the parameter optimization strategies to the terminal execution layer.

3. The integrated water and fertilizer management system based on greenhouse IoT as described in claim 2, characterized in that, The automatic calibration module specifically performs the following steps: it inputs the received sensor operating data into a preset automatic calibration mechanism, and the automatic calibration mechanism performs calibration analysis on the sensor operating data to obtain the sensor to be calibrated.

4. The integrated water and fertilizer management system based on greenhouse IoT as described in claim 3, characterized in that, The specific implementation details of the water and fertilizer strategy support mechanism are as follows: The water and fertilizer strategy support mechanism includes a flower feature extraction unit, a sensor feature extraction unit, and an auxiliary strategy output unit; The flower feature extraction unit obtains feature detection requirements and task detection requirements based on multi-modal water and fertilizer data; The sensor feature extraction unit obtains the monitoring frequency range based on sensor operating data, feature detection requirements, and task detection requirements; The auxiliary strategy output unit obtains the monitoring frequency range corresponding to each environmental parameter of each flower category, outputs the corresponding parameter monitoring frequency strategy based on the monitoring frequency range of each environmental parameter, and combines the parameter monitoring frequency strategies of each sensor to obtain the parameter optimization strategy.

5. The integrated water and fertilizer management system based on greenhouse Internet of Things as described in claim 4, characterized in that, The flower feature extraction unit obtains feature detection requirements and task detection requirements based on water and fertilizer multi-modal data, specifically as follows: Based on water and fertilizer multi-model data, the current flower category is obtained. Based on the current flower category, the physiological characteristic data of the corresponding flower category is crawled from the flower knowledge base in the database. The physiological characteristic data of the flower includes ideal environmental parameters, physiological response dynamics characteristics and planting task level. Based on the physiological response dynamics, the corresponding soil moisture response time, nutrient absorption rate, and water evapotranspiration were obtained to obtain the maximum fluctuation rate of soil water and fertilizer and the environmental recovery response time. Furthermore, the ideal environmental tolerance for the corresponding flowers was obtained based on the database. Under the condition of equal changes in environmental parameters, the maximum fluctuation rate of soil water and fertilizer is the maximum rate of decrease of the index corresponding to each ideal environmental parameter of the flower. The environmental recovery response time is the time from when the index of the corresponding environmental parameter starts water and fertilizer management to when it recovers to the ideal environmental parameter. The maximum fluctuation rate of soil water and fertilizer was divided into three maximum fluctuation rate intervals based on a preset fluctuation rate interval. Fluctuation rate levels were generated by arranging the rates within each maximum fluctuation rate interval in descending order: high fluctuation level, normal fluctuation level, and low fluctuation level. Similarly, the environmental recovery response time was divided into three environmental recovery response time intervals and arranged in descending order to generate environmental recovery levels: high recovery level, normal recovery level, and low recovery level. Characteristic indices were assigned to the fluctuation rate levels: p3, p2, and p1 for high fluctuation, normal fluctuation, and low fluctuation levels, respectively. Simultaneously, characteristic indices were assigned to the environmental recovery levels: p1, p2, and p3 for high recovery, normal recovery, and low recovery levels, respectively. The total characteristic index was obtained by summing the characteristic indices for the fluctuation rate level and environmental recovery level corresponding to each flower category. Feature detection requirements are generated based on the magnitude of the total feature index. The feature detection requirements are positively correlated with the value of the total feature index, that is, feature detection requirements include high feature requirements, medium feature requirements, and low feature requirements. Based on the importance of each planting task level, corresponding task detection requirements are generated. The task detection requirements are positively correlated with the importance of the task level. That is, when the planting task level is scientific research level, the task detection requirements are high; when the planting task level is premium production level, the task detection requirements are medium; and when the planting task level is ordinary production level, the task detection requirements are low.

6. The integrated water and fertilizer management system based on greenhouse Internet of Things as described in claim 5, characterized in that, The sensor feature extraction unit obtains the monitoring frequency range based on sensor operating data, feature detection requirements, and task detection requirements, specifically as follows: Based on sensor operating data, the sensor's monitoring capability range, i.e., the adjustable range of monitoring frequency, is obtained. The monitoring frequency capability range is divided into multiple monitoring frequency ranges based on a pre-given monitoring frequency interval, namely, high-frequency monitoring range, medium-frequency monitoring range, and low-frequency monitoring range. The feature detection requirements and task detection requirements corresponding to each environmental parameter of the current flower category are obtained. The feature detection requirements and task detection requirements are quantified through a preset requirement quantification mechanism to obtain the number of feature detection signals and the number of task detection signals, respectively. The total number of detected signals is calculated by substituting the number of feature detection signals and the number of task detection signals into the preset two-factor coupling calculation formula. The preset signal number intervals corresponding to the high-frequency monitoring interval, the medium-frequency monitoring interval, and the low-frequency monitoring interval are obtained as the high-frequency signal number interval, the medium-frequency signal number interval, and the low-frequency signal number interval, respectively. The total number of detected signals corresponding to the current environmental parameters of flowers is matched with each signal number interval to obtain the corresponding monitoring frequency interval.

7. The integrated water and fertilizer management system based on greenhouse IoT as described in claim 6, characterized in that, The specific execution of the automatic calibration mechanism is as follows: Based on the sensor operation data corresponding to each sensor, a preset number of real-time reference spectral values ​​are obtained according to the sensor operation data. The real-time reference spectral data is collected through a preset spectral extraction mechanism. The real-time reference spectral data is compared with the preset reference standard spectral values, and the difference between the real-time reference spectral data and the reference standard spectral values ​​is calculated to obtain the spectral deviation value. When the spectral deviation value exceeds the preset spectral deviation threshold, the value of the spectral deviation value exceeding the spectral deviation threshold is recorded as the sampling deviation index. Based on the preset analysis time period and sensor operation data, the various environmental parameters corresponding to the preset number of flowers in the current analysis time period are obtained. The various environmental parameters are identified. When the environmental parameter is within the preset parameter fluctuation range, the corresponding environmental parameter is marked as a steady-state parameter. The number of steady-state parameters is counted to obtain the total number of environmental stable parameters. The ratio of the number of environmental stable parameters to the total number of environmental parameters is calculated to obtain the environmental stability ratio. When the environmental stability ratio exceeds the preset stability upper limit ratio, the corresponding analysis time period is marked as a steady-state period. The system acquires the parameter values ​​of various environmental parameters from similar sensors during the steady-state period, and obtains the numerical fluctuation of each parameter value. When the numerical fluctuation exceeds a preset fluctuation threshold, the corresponding sensor is marked as a mutation sensor. Using the mutation sensor as the center, the system acquires similar sensors within a preset distance radius corresponding to the mutation sensor. When none of the similar sensors within the nearby range are marked as mutation sensors, the system generates the temporal mutation frequency of the mutation sensor. The temporal mutation frequency corresponds to the number of times the numerical fluctuation exceeds the fluctuation threshold during the analysis period. Conversely, when a sudden sensor appears in a nearby sensor of the same type, the value fluctuation of each sudden sensor is monitored in real time to obtain the duration of the fluctuation. When the duration of the fluctuation exceeds the preset duration limit, the sudden sensor mark of each sensor is removed; otherwise, the temporal sudden frequency of each sudden sensor is obtained. The sampling deviation index and the temporal mutation frequency are both normalized to the [0,1] interval, and the normalized values ​​are recorded as the sampling deviation analysis value and the temporal mutation analysis value, respectively. The deviation index membership degree and the temporal mutation membership degree are calculated by the S-shaped fuzzy membership function. The sampling quantity corresponding to the two analysis values ​​is taken as the fuzzy entropy. The fuzzy entropy and membership degree of the two analysis values ​​are calculated by the preset weight calculation formula to obtain two weight coefficients. The deviation index membership degree and the temporal mutation membership degree and the corresponding weight coefficients are calculated by the comprehensive calculation formula to obtain the comprehensive outlier value. When the comprehensive outlier value is greater than the preset threshold, the corresponding sensor is marked as a sensor to be calibrated.

8. The integrated water and fertilizer management system based on greenhouse IoT according to claim 7, characterized in that, The specific demand quantification mechanism is as follows: obtain the demand level corresponding to the feature detection demand, assign a number of detection signals according to the order of the demand level, and the number of detection signals increases with the increase of the demand level. The corresponding number of detection signals is recorded as the feature detection signal count. Similarly, obtain the demand level corresponding to the task detection demand, and then assign a number of detection signals accordingly. The corresponding number of detection signals is recorded as the task detection signal count.

9. The integrated water and fertilizer management system based on greenhouse Internet of Things as described in claim 8, characterized in that, The spectral extraction mechanism is as follows: the sample liquid acquired by the probe is pumped into a pre-set reference cell by a micro-pump built into the sensor detection probe, and real-time reference spectral data is obtained by real-time detection of the sample liquid in the reference cell by a micro-spectrometer. The reference standard spectral values ​​are the standard spectral values ​​corresponding to various environmental parameters obtained experimentally under standard laboratory conditions.

10. The integrated water and fertilizer management system based on greenhouse Internet of Things as described in claim 9, characterized in that, The IoT communication layer provides a network communication foundation for the data perception layer, processing and decision-making layer and terminal execution layer through the deployed IoT communication technology. Specifically, the data perception layer sends water and fertilizer multi-mode data and sensor operation data to the processing and decision-making layer through the connection of the IoT communication layer. The system receives parameter optimization strategies and sensors to be calibrated, sends the parameter optimization strategies and sensors to be calibrated to the terminal execution layer, obtains operation adjustment instructions, and feeds back the operation adjustment instructions to the data perception layer.

Images