Intelligent agricultural system based on big data

Abstract

The invention discloses a smart agriculture system based on big data, and relates to the technical field of smart agriculture, a data acquisition unit is used for dynamically partitioning a field to be irrigated, the dynamic partitioning is based on solar illumination energy and effective illumination duration, and the field is divided into a plurality of irrigation partitions with uniform internal illumination conditions; by means of a dynamic partitioning method based on solar illumination energy, a field is divided into a plurality of irrigation partitions with uniform inner irrigation part illumination conditions. The system can recognize and respond to field microenvironment differences, so that partition independent control and on-demand irrigation in a real sense are realized, and the irrigation uniformity and the water resource utilization efficiency are greatly improved.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of intelligent agriculture, and particularly relates to an intelligent agricultural system based on big data. BACKGROUND

[0002] Traditional agricultural irrigation systems mostly use fixed mode of timing and quantity for irrigation control, or only start and stop according to a single soil humidity sensor. Such methods have the following significant shortcomings: 1. Ignoring the differences in field micro-environment: Due to factors such as terrain undulations, crop shading, and sunlight angles, there are significant differences in the intensity and duration of sunlight received by different areas in the same field, resulting in different soil water evaporation rates and crop water requirements. The traditional fixed zoning method according to area or terrain cannot respond to such dynamic changes, causing uneven irrigation, with some areas having excess water and others experiencing water stress.

[0003] 2. Soil humidity monitoring is not comprehensive: Most systems rely on a single root humidity sensor. However, crop water conditions are the result of both surface evaporation and root absorption. Monitoring only the root humidity cannot effectively reflect the trend of rapid surface water evaporation, leading to delayed irrigation decisions, especially in high-temperature and windy weather, which can cause the system to respond slowly and make crops face short-term drought stress.

[0004] 3. Lack of forward-looking decision-making ability: Existing systems are mostly "reactive" irrigation, i.e., irrigation is started only when the soil humidity is below a threshold. This method relies entirely on the current state and cannot predict the impact of future weather conditions (such as temperature, wind speed, and light) on water loss. In the case of upcoming high-temperature sunny weather, water is not replenished in advance, causing crops to face water shortages during the peak water demand period; or ineffective irrigation is performed before the rain arrives, resulting in waste of water resources.

[0005] Therefore, a solution is provided. SUMMARY

[0006] The present application aims to at least solve one of the technical problems existing in the prior art; for this purpose, the present application proposes an intelligent agricultural system based on big data.

[0007] The intelligent agricultural system based on big data comprises: A data acquisition unit for dynamically zoning the field to be irrigated, the dynamic zoning being based on solar radiation energy and effective illumination time to divide the field into a plurality of irrigation zones with uniform internal illumination conditions; A humidity monitoring unit for comprehensive humidity monitoring of each irrigation zone, the humidity monitoring using a dual-sensor system arranged on the ground and the root, and calculating the comprehensive humidity of each zone through a weighted fusion algorithm; a data fusion prediction unit configured to predict the average moisture loss rate of each subzone in a future period of time based on historical environmental data and future weather forecast through a machine learning model; a decision unit configured to determine whether irrigation is needed according to the current comprehensive humidity and the predicted moisture loss rate, and generate corresponding irrigation instructions.

[0008] Further, the data acquisition unit performs the following steps: deploying a solar radiation sensor above the field to continuously collect light data of at least one complete sunny day; dividing the field into a micro grid, calculating the cumulative light energy and effective light duration of each grid in a standard day; based on the set light energy threshold and light duration threshold, using a clustering algorithm to merge adjacent grids with similar light conditions into an irrigation subzone.

[0009] Further, the calculation formula of the cumulative light energy is: E = Σ (light intensity I x time interval At).

[0010] Further, the humidity monitoring unit arranges a ground humidity sensor and a root humidity sensor in each irrigation subzone, and the root humidity sensor is buried 15-20 cm deep underground.

[0011] Further, the calculation formula of the comprehensive humidity is: H = a x H surface + b x H root , wherein a and b are weight coefficients, and a+b=1.

[0012] Further, the weight coefficients a and b are dynamically adjusted according to the growth stage of the crop and the current air temperature.

[0013] Further, the data fusion prediction unit uses a multiple linear regression or neural network model, and the input variables include the current temperature, the future predicted temperature, the wind speed, the light intensity and the current comprehensive humidity, and the output is the average moisture loss rate in the next 24 hours.

[0014] Further, the decision unit determines whether to irrigate according to the comparison result of the predicted humidity H forecast and the preset minimum humidity threshold Hmin. If H forecast < H min , irrigation is started.

[0015] Further, the decision unit also calculates the required irrigation amount, and the calculation formula is: W need = (H t -H forecast ) x SoV x Cr, wherein H t is the target humidity, SoV is the soil volume, and Cr is the crop coefficient.

[0016] Compared with the prior art, the present application has the following advantages: 1. Realize accurate zoning and on-demand irrigation: through the dynamic block method based on solar light energy, the field is divided into multiple internal irrigation zoning with uniform illumination conditions. The system can identify and respond to the microenvironment differences in the field, thus realizing truly independent control and on-demand irrigation, greatly improving the uniformity of irrigation and water resource utilization efficiency.

[0017] 2. Improve the accuracy and reliability of humidity monitoring: a dual-sensor monitoring system is used, combining surface and root sensors, and a weighted fusion algorithm is used to calculate the comprehensive humidity, which can more comprehensively and truly reflect the actual water environment of crops. Both the effective moisture of the root layer and the influence of surface evaporation are considered, providing a more accurate data basis for irrigation decision-making.

[0018] 3. Realize forward-looking and adaptive irrigation decision-making: by integrating machine learning models and weather forecast data, the system can predict the humidity loss rate in the future irrigation period, so as to actively trigger irrigation before the crop actually suffers from water stress. This predictive irrigation mode breaks the lag of traditional reactive irrigation, effectively preventing the occurrence of crop water stress and ensuring the healthy growth of crops. BRIEF DESCRIPTION OF DRAWINGS

[0019] Figure 1 The system block diagram of the present application. DETAILED DESCRIPTION

[0020] The technical solutions of the present application will be described below in conjunction with the embodiments. Obviously, the described embodiments are only a part of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0021] Please refer to Figure 1 , the present application provides a big data-based intelligent agricultural system.

[0022] Embodiment one, specifically includes: A data acquisition unit for dynamically zoning the field to be irrigated, the dynamic zoning is based on solar light energy and effective illumination time, and the field is divided into several internal irrigation zoning with uniform illumination conditions; A humidity monitoring unit for monitoring the comprehensive humidity of each irrigation zoning, the humidity monitoring uses a dual-sensor system arranged on the surface and the root, and a weighted fusion algorithm is used to calculate the comprehensive humidity of each zoning; a data fusion prediction unit configured to predict, based on historical environmental data and future weather forecasts, average humidity loss rates of each subzone in a future period of time by using a machine learning model; a decision unit configured to determine whether irrigation is needed according to the current comprehensive humidity and the predicted humidity loss rate, and generate a corresponding irrigation instruction.

[0023] Embodiment Two, This embodiment details a precision irrigation method for open fields based on sunlight intensity and duration, which uses double-sensor comprehensive humidity monitoring and dynamic humidity loss prediction; The dynamic sub-blocking method based on sunlight of the field is one of the core innovations of the present application, which breaks the traditional fixed mode of area or terrain partitioning and adopts a dynamic partitioning method based on sunlight energy. The scheme provided by the present application specifically includes: a data acquisition unit, which, for ease of description, is marked as a target object throughout the text, is used to acquire and divide the target object to be irrigated, and the specific processing process is: S1: Deploy a solar radiation sensor above the field to be irrigated to monitor the light intensity (unit: W / m2) in real time, combined with the geographical position (latitude and longitude) and historical meteorological data, to obtain the theoretical daily maximum sunshine duration, which is a prior art and will not be described in detail; The system continuously acquires light data for at least one complete sunny day; S2: Calculate the light energy integral value of each position; The system divides the field into initial small grids (e.g., 1mx1m); for each grid, the system calculates the cumulative light energy it receives in a standard day, and the calculation formula can be simplified as: E=Σ(light intensityIx time intervalΔt); Here, Δt is the time interval of data acquisition, and I is the average light intensity in each time interval; At the same time, record the effective light duration of each grid, where the effective light duration is the time when the light intensity exceeds the compensation point of crop photosynthesis; S3: Cluster analysis and subzone merging; The system sets a light energy threshold and a light duration threshold; Using a clustering algorithm (such as K-means or DBSCAN), adjacent grids with similar cumulative light energy and effective light duration are aggregated into an irrigation subzone; in other words, if the illumination and light duration of several grid regions are highly consistent in combination, they are divided into the same block; if the difference is significant, they are divided into several independent blocks according to different light energy levels.

[0024] S4: Subzone identification; Ultimately, the field was divided into n independent irrigation zones, labeled Zi, i=1, ..., n; the light conditions within each zone were highly uniform, while there were significant differences between the zones; this zoning result will serve as the basis for subsequent independent control. Several irrigation zones Zi were obtained; The data acquisition unit transmits data from several irrigation zones to the humidity monitoring unit. The humidity monitoring unit monitors the overall humidity of different irrigation zones. The specific monitoring method is as follows: SS1. Within each irrigation zone Zi, deploy a unique dual-sensor system to more accurately reflect the crop's water environment. The sensor deployment is as follows: Surface humidity sensors are installed on the soil surface to monitor the effects of surface evaporation rate and near-surface air humidity. Root humidity sensors are installed in the main root activity layer of the crop. Generally, the root humidity sensors are placed 15-20 cm deep underground. Of course, the administrator can adjust the depth according to the specific situation. They are used to monitor the effective water that the crop can directly absorb. SS2. Then, a comprehensive humidity calculation is performed on the humidity detected by the two humidity sensors. The specific calculation method is as follows: This method does not simply take the average of the two sensor readings; instead, a weighted fusion algorithm is used to calculate the overall humidity. The weighted fusion algorithm is as follows: H = α × H surface +β×H root ; In the formula, H surface H represents the reading from the surface humidity sensor. root The readings are from the root humidity sensor; α and β are weighting coefficients, and α+β=1; Thus, the comprehensive humidity Hi of each irrigation zone Zi is obtained, where i=1,...,n, and Hi and Zi have a one-to-one correspondence. It also includes a data fusion and prediction unit to perform dynamic humidity loss prediction based on weather forecasts; by predicting future humidity loss, it enables proactive irrigation, and its decision-making method is as follows: Step 1: Establish a humidity loss rate model: The system backend stores historical big data, which includes the natural rate of decrease in soil moisture when not irrigated under different temperature, humidity, wind speed and light conditions. Historical data is analyzed using machine learning algorithms, such as multiple linear regression or neural networks, to train a model for predicting the rate of humidity loss. The specific process of model training is as follows: Data collection: Collect historical environmental data, including temperature, humidity, wind speed, light intensity, soil moisture, etc., and record the rate of change of soil moisture within the corresponding time period as label data; Feature engineering: Select input features, including current temperature, future predicted temperature, future wind speed, future light intensity, and current overall humidity; future meteorological data (future predicted temperature, future wind speed, and future light intensity) are obtained by accessing meteorological service interfaces (such as the National Meteorological Administration or commercial weather forecast APIs), and the system periodically calls these interfaces to obtain forecast data for the next 24 hours; Model training: When using multiple linear regression, the relationship between features and humidity loss rate is fitted using the least squares method; when using neural networks, a multilayer perceptron (MLP) model is constructed, which includes an input layer, a hidden layer, and an output layer, and trained using a backpropagation algorithm to minimize prediction error; Model validation: Use cross-validation or reserved datasets to evaluate model performance and ensure that prediction accuracy meets requirements; The model's input variables include: current temperature, predicted future temperature, future wind speed, future light intensity, and current overall humidity; the output is the average humidity loss rate V over a future period. loss (Unit: % volumetric moisture content / hour); Here, the future period is generally used as 24 hours, which is the predicted average humidity loss rate V over 24 hours. loss ; It also includes a decision-making unit, which receives the average humidity loss rate transmitted by the data fusion and prediction unit and the comprehensive humidity Hi corresponding to each irrigation zone Zi transmitted by the humidity monitoring unit, and makes irrigation decisions, specifically as follows: Get the comprehensive humidity Hi corresponding to each irrigation zone Zi, and get the weather forecast for the next 48 hours from the meteorological service interface, and automatically get the rainfall for the next 48 hours. Automatically calculate the predicted humidity H 24 hours later forecast The calculation formula is: H forecast = H current -V loss x24; Obtain the minimum humidity threshold H preset by the administrator for each zone and each crop growth stage. min (Ensure crops are not under stress) and within the optimal humidity range; If the calculated H forecast Below H min If so, the system determines that irrigation is needed.

[0025] The irrigation amount was determined based on the median value of the overall humidity level, which was found to be within the optimal humidity range.

[0026] Example 3 is an implementation based on Example 2, with the following differences: When calculating the overall humidity, the weighting coefficients α and β in the humidity monitoring unit are not fixed, but dynamically adjusted according to the crop growth stage and the current temperature. The specific adjustment method is as follows: The current growth cycle of the crop is obtained. In the early stage of crop growth, the root system is shallow and more dependent on surface water. Therefore, the α value can be set relatively high, such as 0.45, and the β value is 0.55. In the middle and late stages of crop growth, the root system is well developed and water is mainly absorbed from the deep soil. Therefore, the α value decreases and can be set to 0.3, while the β value increases and can be set to 0.7.

[0027] In hot weather, surface evaporation is intense, and the readings of surface humidity sensors are highly volatile. At this time, the system will automatically reduce the weight of α and rely more on the stability of root humidity.

[0028] Example 4 is an implementation based on Example 3, except that the decision-making unit in this example also needs to calculate the irrigation amount: If irrigation is required, the system will ensure that the humidity does not fall below H for the next 24 hours. min And calculate the required irrigation amount based on the principle of getting as close as possible to the optimal range; The target humidity Ht is set to be slightly higher than the median of the upper and lower limits of the optimal range for Hmin; Water demand calculation: W need =(H t -H forecast )×SoV×Cr; In the formula, SoV is the total volume of the root zone soil in that zone, and Cr is the crop coefficient, which is determined according to different crops and growth stages and is preset by the administrator.

[0029] The system also considers the probability of precipitation in the weather forecast; if there is a high probability (e.g., >60%) of effective rainfall in the next 12 hours, the irrigation plan will be reduced or canceled accordingly based on the predicted rainfall.

[0030] Taking partition Z1 as an example: 1. The system reads data from the dual sensors in the Z1 partition and dynamically calculates the current overall humidity to be 65%.

[0031] 2. The system obtains the weather forecast, showing that the weather will be sunny for the next 24 hours, with the temperature remaining between 30-35℃ and the wind speed at level 3.

[0032] 3. Based on these conditions, the humidity loss model predicts that the average loss rate of zone Z1 is 2% / hour.

[0033] 4. The system calculates that after 24 hours, the predicted humidity of zone Z1 will be 65% - (2% / h × 24h) = 17%, which is far below the minimum threshold of 40% set for the current growth stage of the crop.

[0034] 5. The system thus generates an irrigation decision; the target humidity is set at 45%, and based on the soil volume and crop coefficient, it is calculated that 15 cubic meters of water need to be injected into zone Z1.

[0035] 6. During the preset low-temperature evaporation period, such as 4 a.m., the central controller only sends instructions to the high-precision valves in zone Z1 to perform precise irrigation, while other zones with sufficient humidity or low loss rate do not operate.

[0036] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. A big data based smart agriculture system, characterized in that, The application relates to a dynamic irrigation system, comprising: a data acquisition unit for dynamically partitioning a field to be irrigated, the dynamic partitioning being based on solar radiation energy and effective illumination time, and the field being divided into a plurality of irrigation partitions with uniform internal illumination conditions; a humidity monitoring unit for comprehensively monitoring the humidity of each irrigation partition, the humidity monitoring adopting a double-sensor system arranged on the ground surface and the root part, and calculating the comprehensive humidity of each partition through a weighted fusion algorithm; a data fusion prediction unit for predicting the average humidity loss rate of each partition in a future period of time through a machine learning model based on historical environmental data and future weather forecasts; a decision unit for judging whether irrigation is needed according to the current comprehensive humidity and the predicted humidity loss rate, and generating corresponding irrigation instructions.

2. The big data based smart agriculture system as claimed in claim 1, wherein, The data acquisition unit performs the following steps: deploying a solar radiation sensor above the field to continuously collect illumination data of at least one complete sunny day; dividing the field into a micro grid to calculate the cumulative illumination energy and effective illumination time of each grid in a standard day; based on the set illumination energy threshold and illumination time threshold, using a clustering algorithm to combine adjacent grids with similar illumination conditions into an irrigation partition.

3. The big data based smart agriculture system as claimed in claim 1, wherein, The calculation formula of the cumulative illumination energy is E = Sigma (illumination intensity I x time interval Delta t).

4. The big data based smart agriculture system as claimed in claim 1, wherein, The humidity monitoring unit arranges a ground humidity sensor and a root humidity sensor in each irrigation partition, and the root humidity sensor is buried at a depth of 15-20 cm underground.

5. The big data based smart agriculture system as claimed in claim 4, wherein, The formula for calculating the comprehensive humidity is: H = a x H surface + b x H root where a and b are weight coefficients, and a + b = 1.

6. The big data based smart agriculture system as claimed in claim 5, wherein, The weight coefficients alpha and beta are dynamically adjusted according to the growth stage of crops and the current air temperature.

7. The big data based smart agriculture system as claimed in claim 1, wherein, The data fusion prediction unit uses a multiple linear regression or neural network model, the input variables include the current temperature, the future predicted temperature, the wind speed, the illumination intensity and the current comprehensive humidity, and the output is the average humidity loss rate in the next 24 hours.

8. The big data based smart agriculture system as claimed in claim 1, wherein, The decision unit determines whether to irrigate based on a comparison result of the predicted humidity H forecast with a preset minimum humidity threshold H min . If H forecast < H min , irrigation is started.

9. The big data based smart agriculture system as claimed in claim 1, wherein, The decision unit also calculates the required irrigation amount, using the formula: W need = (H t - H forecast ) x SoV x Cr, where H t is the target humidity, SoV is the soil volume, and Cr is the crop coefficient.

Images