A crop yield prediction device and method based on in-situ monitoring

Abstract

A crop yield prediction device and method based on in-situ monitoring, relating to the field of crop yield prediction technology, includes a local data processing terminal and a cloud data center. The local data processing terminal is used for wireless data transmission with in-situ monitoring sensors and field weather stations, and for preprocessing data collected from these sensors and stations before uploading it to the cloud data center. The cloud data center embeds a crop yield prediction model. This invention employs core mechanisms such as multi-depth monitoring, water-fertilizer coupling, and nutrient weight allocation, demonstrating good predictive performance in practical applications.

Description

Technical Field

[0001] This invention relates to the field of crop yield prediction technology, and in particular to a crop yield prediction device and method based on in-situ monitoring. Background Technology

[0002] Crop yield forecasting is a crucial basis for agricultural production management, food security assessment, and agricultural policy formulation. With the development of information technology, data-driven yield forecasting methods have become a research hotspot, and can be mainly summarized into the following technical approaches: The first category is prediction methods based on meteorological data and statistical / machine learning models. These methods analyze the relationship between historical meteorological data (such as temperature, precipitation, and sunshine) and yield, constructing regression or neural network models for prediction. For example, Chinese patent application CN108665107A proposes a yield prediction method based on meteorological data and a VGG neural network model; Chinese patent application CN121503788A further integrates vegetation indices extracted from remote sensing images and micro-meteorological data, and introduces an attention mechanism to construct a deep learning model. The prediction accuracy of these methods highly depends on the complexity of the model and the quality of the training data. Furthermore, their essence is to establish a statistical correlation between historical data and yield, making it difficult to reveal the physiological and ecological mechanisms of crop yield formation. Especially under climate change or extreme weather conditions, the predictive stability of the model faces challenges.

[0003] The second category is yield estimation methods based on remote sensing imagery and vegetation indices. These methods utilize multispectral and hyperspectral sensors mounted on satellites or drones to acquire information about large areas of crop canopy, and indirectly estimate biomass and yield by retrieving vegetation indices (such as NDVI, EVI, GNDVI, and leaf area index LAI). For example, Chinese patent application CN114612380A estimates yield by establishing a regression equation based on the leaf area index of a tea garden; Chinese patent application CN121350491A combines binocular vision with multi-angle images from drones to extract the three-dimensional structural features of fruit trees for yield prediction. However, remote sensing technology is greatly affected by weather conditions (such as cloud cover), and it acquires instantaneous, population-level information about the crop canopy, making it difficult to continuously and dynamically capture real-time changes in the soil-crop system during the critical growth period of crop yield formation.

[0004] The third category is prediction methods based on multi-source data fusion and deep learning. To overcome the limitations of a single data source, existing technologies attempt to fuse multiple data sources. For example, Chinese patent application CN118887546A fuses RGB color features and infrared temperature features from multispectral images; Chinese patent application CN120873468A fuses solar-induced chlorophyll fluorescence (SIF) data, vegetation indices, and meteorological data, and uses deep neural networks (DNNs) for maize yield prediction, showing particularly high sensitivity in hot and dry years. While these methods improve the comprehensiveness of predictions through data fusion, their core still relies on complex deep learning models (such as DNNs, LSTMs, 3D-CNNs, and generative adversarial networks (GANs) to uncover nonlinear relationships between data. The interpretability of these models is poor, and they place high demands on computational resources and the size of training samples.

[0005] In summary, existing technologies mainly follow the path of "multi-source data + complex models," focusing on mining statistical patterns among data through advanced algorithms. However, these methods all share a common technological gap: the lack of a technical solution that can directly and effectively utilize real-time in-situ field monitoring data (such as soil moisture, nutrients, crop growth, and meteorological indicators) to dynamically and accurately predict crop yield without relying on any black-box models or complex algorithms. Most existing in-situ monitoring technologies remain at the level of data acquisition and environmental perception, with applications primarily concentrated on real-time management decisions such as soil moisture early warning and automatic irrigation control. How to correlate this high temporal resolution in-situ data with crop yield formation mechanisms to construct a yield prediction method with a clear mechanism, simple computation, and reliable results remains a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0006] In view of the above-mentioned shortcomings and deficiencies of the prior art, the present invention provides a crop yield prediction device and method based on in-situ monitoring.

[0007] To achieve the above objectives, the main technical solutions adopted by the present invention include: A crop yield prediction device based on in-situ monitoring includes a local data processing terminal and a cloud data center. The local data processing terminal is used for wireless data transmission with in-situ monitoring sensors and field weather stations, and for preprocessing the data collected from the in-situ monitoring sensors and field weather stations before uploading it to the cloud data center. The cloud data center embeds a crop yield prediction model as follows: ; Where Y is the target yield; Y0 is the crop potential yield constant; CC is the crop canopy coverage; θ is the soil volumetric water content; C xl This is data on soil nutrient content.l =1~4, l =1 represents the soil nitrate nitrogen content. l =2 represents the soil ammonium nitrogen content. l =3 represents the available phosphorus content in the soil. l =4 represents the available potassium content in the soil; ω j Soil nutrient weighting coefficients, j=1~4, j=1 represents soil nitrate nitrogen content, j=2 represents soil ammonium nitrogen content, j=3 represents soil available phosphorus content, and j=4 represents soil available potassium content; α represents soil water and fertilizer coupling weighting coefficient; Li represents light intensity; GDD represents accumulated temperature; i (1~n) represents daily data within each growth period; t n t0 is the last day of each reproductive period; t0 is the first day of each reproductive period.

[0008] Furthermore, cloud data centers communicate with user clients through access.

[0009] Furthermore, the data collected by in-situ monitoring sensors and field weather stations include soil volumetric water content, soil nutrient content, crop canopy coverage, meteorological temperature, light intensity, and accumulated temperature.

[0010] Furthermore, the local data processing terminal preprocesses the data collected by the in-situ monitoring sensors and field weather stations, including removing outliers and missing values, and then using the average of adjacent time periods to complete the data.

[0011] Furthermore, the cloud data center stores the soil nutrient weighting coefficient ω. j The data shows that the weighting coefficient for nitrate nitrogen content in soils of grain crops is 0.35, for soils of cash crops it is 0.32, and for soils of fruit and vegetable crops it is 0.30; the weighting coefficient for ammonium nitrogen content in soils of grain crops is 0.25, for soils of cash crops it is 0.23, and for soils of fruit and vegetable crops it is 0.22; the weighting coefficient for available phosphorus content in soils of grain crops is 0.20, for soils of cash crops it is 0.22, and for soils of fruit and vegetable crops it is 0.25; and the weighting coefficient for potassium ion content in soils of grain crops is 0.20, for soils of cash crops it is 0.23, and for soils of fruit and vegetable crops it is 0.23.

[0012] Furthermore, the cloud data center stores crop potential yield constant Y0 data, in kilograms per mu (approximately 0.067 hectares). The potential yield constants for various crops are as follows: Liaoning maize: 1350; Northeast spring maize: 833; Huang-Huai-Hai summer maize: 833; Southern hilly maize: 833; Early indica rice: 700; Medium and late indica rice: 780; Northeast japonica rice: 850; Northern winter wheat: 787; Spring wheat: 700; Northeast soybean: 203; and Southern soybean: [data missing]. The potential yield constants for various crops are as follows: peanut (169), rapeseed (450), Xinjiang cotton (550), inland cotton (350), potato (3500), sorghum (600), millet (400), mung bean (250), greenhouse tomato (12000), open-field cabbage (6000), dwarf dense-planted apple (4500), citrus (2500), and grape (169). The potential yield constants for various fruits and vegetables are as follows: banana (2000), pear (3500), peach (2500), strawberry (3000), blueberry (1000), watermelon (4000), cantaloupe (3000), greenhouse cucumber (10000), chili pepper (4000), eggplant (5000), Chinese cabbage (8000), kale (6000), radish (5000), and carrot (5000). The potential yield constants are as follows: 4000 for potatoes, 3500 for garlic, 2000 for onions, 5000 for scallions, 4000 for green beans, 2500 for asparagus, 1500 for dried tea leaves, 300 for sugarcane, 8000 for sugar beets, 5000 for peanuts, 450 for rapeseed, 250 for sunflowers, 300 for dried tobacco leaves, and 250 for dried tobacco leaves.

[0013] The crop yield prediction method using the aforementioned prediction device includes: deploying in-situ soil moisture monitoring equipment in the field to collect soil volumetric water content data; deploying in-situ soil nutrient monitoring equipment in the field to collect soil nutrient content data; deploying in-situ crop growth monitoring equipment in the field to collect crop canopy coverage data; collecting temperature and light intensity data through field weather stations to generate accumulated temperature data; wirelessly transmitting the collected data to a local data processing terminal for data processing; uploading the processed data to a cloud data center; the cloud data center outputting the target yield based on the collected data, the crop yield prediction model, and model parameter data; storing the predicted yield results in the cloud data center; and allowing clients to access the cloud data center data.

[0014] Furthermore, the soil moisture in-situ monitoring equipment collects a set of data every hour, and the daily average is used as the data for that day; Furthermore, the sensors of the soil moisture in-situ monitoring equipment and the soil nutrient in-situ monitoring equipment are deployed at depths of 10cm, 20cm, 30cm, and 40cm below the soil surface layer, respectively.

[0015] Furthermore, the in-situ crop growth monitoring equipment is set to collect a set of crop canopy coverage data daily; the field weather station is set to acquire a set of data every hour, and the average value is taken as the daily data.

[0016] The beneficial effects of this invention are as follows: The crop yield prediction device and method based on in-situ monitoring of this invention, employing core mechanisms such as multi-depth monitoring, water-fertilizer coupling, and nutrient weight allocation, demonstrates excellent predictive performance in practical applications. This invention can operate stably in actual farmland environments, with excellent coordination between data acquisition, transmission, and processing, resulting in high prediction accuracy. Attached Figure Description

[0017] Figure 1 This is an in-situ image monitoring diagram of Embodiment 1 of the present invention. Detailed Implementation

[0018] To better explain and facilitate understanding of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0019] This invention provides a crop yield prediction device based on in-situ monitoring, comprising a local data processing terminal and a cloud data center. The local data processing terminal is used for wireless data transmission with in-situ monitoring sensors and field weather stations, and for preprocessing the data collected from these sensors and stations before uploading it to the cloud data center. Specifically, the data collected by the in-situ monitoring sensors and field weather stations includes soil volumetric water content, soil nutrient content, crop canopy coverage, meteorological temperature, light intensity, and accumulated temperature. The preprocessing of the data collected by the local data processing terminal includes removing outliers and missing values ​​to ensure the validity of the input data. Outliers are defined as monitoring data exceeding the physiological threshold of the parameters, and are supplemented using the average of data from adjacent time periods. The cloud data center communicates with the user client via access. The electronic components in the device can be connected to power supply devices such as solar power devices to provide power to the electronic components.

[0020] The cloud data center has an embedded crop yield prediction model as follows: ; Where Y is the target yield (kg / mu); Y0 is the crop potential yield constant (kg / mu); CC is the crop canopy coverage (%,) and θ is the soil volumetric water content (%,). The data at 10, 20, 30, and 40 cm after daily averaging are then averaged again based on depth; C xl This is data on soil nutrient content. l =1~4, l =1 represents the soil nitrate nitrogen content. l =2 represents the soil ammonium nitrogen content. l =3 represents the available phosphorus content in the soil. l =4 represents the available potassium content in the soil. The average value of daily data collected at 10, 20, 30, and 40 cm depths was calculated. ω j α represents the soil nutrient weighting coefficient, j=1~4, where j=1 represents soil nitrate nitrogen content, j=2 represents soil ammonium nitrogen content, j=3 represents soil available phosphorus content, and j=4 represents soil available potassium content, with a total of 1. This coefficient is set according to crop nutrient requirements to reflect the differences in the contribution of different nutrients to yield. α is the soil water-fertilizer coupling weighting coefficient, 0.58, a constant term. Li represents light intensity in Lux. GDD represents accumulated temperature in °C. i (1~n) represents daily data within each growth period. t n t0 is the last day of each reproductive period; t0 is the first day of each reproductive period.

[0021] The model of this invention has the following characteristics: 1. Explicit expression of water and fertilizer coupling: By introducing the α coefficient, the allocation of α and (1-α) in the algorithm reflects the synergistic contribution of water and fertilizer to yield, breaking the assumption of independent variables.

[0022] 2. Nutrient Differentiation Weighting: Establish a crop-specific ωⱼ database and introduce nutrient weighting coefficients ωⱼ to reflect the differentiated contributions of nitrogen, phosphorus, and potassium to yield. However, current technologies all use equal-weighted summation or principal component dimensionality reduction, which obscures the physiological significance.

[0023] 3. Cumulative effect over time: Light and accumulated temperature are added together over the growing season, which is consistent with the yield formation pattern. However, existing technologies mostly use instantaneous or average values ​​in terms of time, and do not divide the growing season, which does not conform to the characteristics of crop growth.

[0024] After calculation, the predicted yield results will be automatically stored in the cloud data center and can be viewed in real time by users through the platform. This invention constructs a complete data link of in-situ monitoring, wireless transmission, and automatic cleaning, and a complete outlier identification and completion mechanism.

[0025] The cloud data center stores the soil nutrient weight coefficient ω j The data is shown in Table 1.

[0026] Table 1 is a comparison table of soil nutrient weight coefficient ωⱼ parameters: .

[0027] The cloud data center stores crop potential yield constant Y0 data in kilograms per acre, as shown in Table 2: Table 2 is a comparison table of crop potential yield constants: ; This invention also provides a method for predicting crop yield using the aforementioned predictive device, comprising deploying in-situ soil moisture monitoring equipment in the field to collect soil volumetric water content data. Specifically, it can be implemented using the method described in patent application number 201610244211.9, entitled "A Non-Contact Water-Salt Sensor Based on Soil Texture and Its Testing Method." By deploying in-situ monitoring equipment in the field, in-situ soil volumetric water content (θ) is measured using a high-frequency dielectric method, with data in percentage (%), one set of data per hour, and the daily average as the daily data.

[0028] Soil nutrient content data is collected by deploying in-situ soil nutrient monitoring equipment in the field. This can be achieved using the methods described in patent applications No. 201611209179.7 ("A Rapid Method for In-situ Measurement of Soil Nutrients Based on Dielectric Spectroscopy") and No. 202510857128.8 ("A Rapid Device and Method for In-situ Measurement of Soil Exchangeable Nutrients").

[0029] By deploying in-situ monitoring equipment in the field, the dielectric spectral frequency division method was used to monitor the content of nitrate nitrogen (Cx1), ammonium nitrogen (Cx2), available phosphorus (Cx3), and available potassium (Cx4) in the soil in situ, with the unit being mg / kg, and one set of data per day.

[0030] To reflect the characteristics of crop root stratification in water and nutrient utilization, sensors for in-situ soil moisture monitoring equipment and in-situ soil nutrient monitoring equipment are deployed at depths of 10cm, 20cm, 30cm, and 40cm, respectively. This invention introduces yield prediction through stratified monitoring to realize the root response mechanism.

[0031] The existing technology involves deploying in-situ crop growth monitoring equipment in the field to collect crop canopy coverage data, which will not be elaborated upon here.

[0032] Temperature (T, unit °C) and light intensity (Li, unit Lux) are simultaneously acquired through farmland microclimate weather stations, generating hourly data sets of accumulated temperature (GDD, unit °C), and the average value is taken as the data for the day.

[0033] The collected data is wirelessly transmitted to a local data processing terminal for processing. The processed data is then uploaded to a cloud data center. Based on the collected data, crop yield prediction model, and model parameter data, the cloud data center outputs the target yield. The predicted yield results are stored in the cloud data center, and the client can access the cloud data center to obtain the data.

[0034] Example: Application of maize crop yield prediction in Changtu County, Liaohe Plain: The plot in Changtu County, located at 123°39′ longitude, 43°21′ latitude, and 162m altitude, is planted with maize. The growing season spans from May to October each year, including six growth stages: sowing, seedling, jointing, booting, grain-filling, and maturity. Yield calculations are based on five of these stages: seedling, jointing, booting, grain-filling, and maturity. First, in-situ soil moisture, nitrate nitrogen content, ammonium nitrogen content, available phosphorus content, available potassium content, crop canopy coverage, temperature, and light data at the start time of each growth stage are obtained (see Table 3). In-situ image monitoring data is available in [reference needed]. Figure 1 .

[0035] Table 3 shows the in-situ data obtained: ; All in-situ monitoring data are uploaded to the local data processing terminal in real time via a 4G wireless transmission module. The local data processing terminal automatically identifies and removes outliers based on preset physiological thresholds. The criteria for determining outliers are as follows: (1) Soil moisture θ < 5% or > 45% (exceeding the physiological tolerance range of crops); (2) The soil nutrient content exceeds the historical maximum value of the region by 1.5 times; (3) Canopy coverage (CC) < 0% or > 100%; (4) Illumination intensity Li < 0 or > 120000 Lux.

[0036] After removing outlier data, the mean of valid data from adjacent time periods is used to complete the data sequence, ensuring its continuity and integrity.

[0037] Parameter configuration 1. Selection of soil nutrient weighting coefficients based on maize crop: ω1 (nitrate nitrogen weight) = 0.35 ω2 (weight of ammonium nitrogen) = 0.25 ω3 (weight of available phosphorus) = 0.20 ω4 (weight of available potassium) = 0.20 The total is 1.00, which is consistent with the fertilizer requirements of corn.

[0038] 2. Other model constants Potential crop yield Y0: Based on regional trial data of this variety, it is set at 1350 kg / mu (already embedded in the cloud data center). The water-fertilizer coupling weighting coefficient α is 0.58.

[0039] The cloud data center calculated Y to be 1031.4 kg / mu based on the model. Simultaneously, actual yield measurements were obtained by field sampling and yield determination in the experimental plots at maturity (September 25th). Sampling method: Five-point sampling method, collecting all ears of fruit within a 10㎡ area at each point.

[0040] Yield measurement results: After threshing, drying, weighing, and conversion, the actual corn yield of this plot in that year was 1025.35 kg / mu.

[0041] Accuracy is evaluated using a relative error formula: Relative error = |Predicted value - Measured value|Measured value × 100%; Relative error = Measured value | Predicted value - Measured value | × 100%; Relative error = |1031.4−1025.35|1025.35×100%=6.051025.35×100%=0.59%; Relative error = 1025.35 | 1031.4 − 1025.35 | × 100% = 1025.35 / 6.05 × 100% = 0.59%.

[0042] In this embodiment, the yield prediction model outputs 1031.4 kg / mu, which is 6.05 kg / mu in absolute error and only 0.59% in relative error compared with the actual field yield of 1025.35 kg / mu. This is far below the 10% error threshold usually accepted in agricultural production.

[0043] The results show that the technical solution of the present invention can operate stably in actual farmland environment, and the data collection, transmission and processing links are well connected.

[0044] High prediction accuracy: The relative error of 0.59% verifies the model's ability to fuse multi-source heterogeneous data and the rationality of the parameter settings.

[0045] Technological advantages are evident: core mechanisms such as multi-depth monitoring, water-fertilizer coupling, and nutrient weight allocation demonstrate good predictive performance in practical applications.

[0046] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Any modifications, alterations, substitutions, and variations made by those skilled in the art to the above embodiments are within the scope of the present invention.

Claims

1. A crop yield prediction device based on in-situ monitoring, characterized in that, This includes a local data processing terminal and a cloud data center. The local data processing terminal is used for wireless data transmission with in-situ monitoring sensors and field weather stations, and for preprocessing the data collected from the in-situ monitoring sensors and field weather stations before uploading it to the cloud data center. The cloud data center embeds a crop yield prediction model as follows: ； Where Y is the target yield; Y0 is the crop potential yield constant; CC is the crop canopy coverage; θ is the soil volumetric water content; C xl This is data on soil nutrient content. l =1~4, l =1 represents the soil nitrate nitrogen content. l =2 represents the soil ammonium nitrogen content. l =3 represents the available phosphorus content in the soil. l =4 represents the available potassium content in the soil; ω j Soil nutrient weighting coefficients, j=1~4, j=1 represents soil nitrate nitrogen content, j=2 represents soil ammonium nitrogen content, j=3 represents soil available phosphorus content, and j=4 represents soil available potassium content; α represents soil water and fertilizer coupling weighting coefficient; Li represents light intensity; GDD represents accumulated temperature; i (1~n) represents daily data within each growth period; t n t0 is the last day of each reproductive period; t0 is the first day of each reproductive period.

2. The crop yield prediction device based on in-situ monitoring according to claim 1, characterized in that: Cloud data centers communicate with user clients through access.

3. The crop yield prediction device based on in-situ monitoring according to claim 1, characterized in that: Data collected by in-situ monitoring sensors and field weather stations include soil volumetric water content, soil nutrient content, crop canopy coverage, meteorological temperature, light intensity, and accumulated temperature.

4. The crop yield prediction device based on in-situ monitoring according to claim 1, characterized in that: The local data processing terminal preprocesses the data collected by in-situ monitoring sensors and field weather stations, including removing outliers and missing values, and then using the average of data from adjacent time periods to complete the data.

5. A crop yield prediction device based on in-situ monitoring according to claim 1, characterized in that: The cloud data center stores the soil nutrient weight coefficient ω j The data shows that the weighting coefficient for nitrate nitrogen content in soils of grain crops is 0.35, for soils of cash crops it is 0.32, and for soils of fruit and vegetable crops it is 0.30; the weighting coefficient for ammonium nitrogen content in soils of grain crops is 0.25, for soils of cash crops it is 0.23, and for soils of fruit and vegetable crops it is 0.22; the weighting coefficient for available phosphorus content in soils of grain crops is 0.20, for soils of cash crops it is 0.22, and for soils of fruit and vegetable crops it is 0.25; and the weighting coefficient for potassium ion content in soils of grain crops is 0.20, for soils of cash crops it is 0.23, and for soils of fruit and vegetable crops it is 0.

23.

6. A crop yield prediction device based on in-situ monitoring according to claim 1, characterized in that: The cloud data center stores crop potential yield constant Y0 data, in kilograms per mu (approximately 0.067 hectares). The potential yield constants are as follows: Liaoning maize: 1350; Northeast spring maize: 833; Huang-Huai-Hai summer maize: 833; Southern hilly maize: 833; Early indica rice: 700; Medium and late indica rice: 780; Northeast japonica rice: 850; Northern winter wheat: 787; Spring wheat: 700; Northeast soybean: 203; and Southern soybean: [data missing]. The potential yield constants for various crops are as follows: peanut (169), rapeseed (450), Xinjiang cotton (550), inland cotton (350), potato (3500), sorghum (600), millet (400), mung bean (250), greenhouse tomato (12000), open-field cabbage (6000), dwarf dense-planted apple (4500), citrus (2500), and grape (…). 2000, Banana potential yield constant is 3500, Pear potential yield constant is 3000, Peach potential yield constant is 2500, Strawberry potential yield constant is 3000, Blueberry potential yield constant is 1000, Watermelon potential yield constant is 4000, Cantaloupe potential yield constant is 3000, Greenhouse cucumber potential yield constant is 10000, Chili pepper potential yield constant is 4000, Eggplant potential yield constant is 5000, Chinese cabbage potential yield constant is 8000, Cabbage potential yield constant is 6000, Radish potential yield constant is 5000, Carrot potential yield constant is 5000, The potential yield constants are as follows: potato potential yield constant 3500, garlic potential yield constant 2000, onion potential yield constant 5000, scallion potential yield constant 4000, green bean potential yield constant 2500, asparagus potential yield constant 1500, dried tea potential yield constant 300, sugarcane potential yield constant 8000, sugar beet potential yield constant 5000, peanut potential yield constant 450, rapeseed potential yield constant 250, sunflower potential yield constant 300, and dried tobacco potential yield constant 250.

7. A crop yield prediction method using the prediction device according to any one of claims 1 to 6, characterized in that, This includes deploying in-situ soil moisture monitoring equipment in the field to collect soil volumetric water content data; deploying in-situ soil nutrient monitoring equipment in the field to collect soil nutrient content data; deploying in-situ crop growth monitoring equipment in the field to collect crop canopy coverage data; collecting temperature and light intensity data through field weather stations to generate accumulated temperature data; wirelessly transmitting the collected data to a local data processing terminal for data processing; uploading the processed data to a cloud data center; the cloud data center outputting the target yield based on the collected data, crop yield prediction model, and model parameter data; storing the predicted yield results in the cloud data center; and clients accessing the cloud data center data.

8. The crop yield prediction method according to claim 7, characterized in that, The soil moisture in-situ monitoring equipment collects a set of data every hour, and the daily average is used as the data for that day.

9. The crop yield prediction method according to claim 7, characterized in that, The sensors of the soil moisture in-situ monitoring equipment and the soil nutrient in-situ monitoring equipment are deployed at depths of 10cm, 20cm, 30cm and 40cm below the soil surface layer, respectively.

10. The crop yield prediction method according to claim 7, characterized in that, The in-situ crop growth monitoring equipment is set to collect a set of crop canopy coverage data every day; the field weather station is set to acquire a set of data every hour, and the average value is taken as the data for the day.

Images