A precise fertilization method based on real-time detection of soil fertility
By configuring image acquisition and light wave modules in the fertilization machine, identifying the pigment characteristics of plant leaves and planning the fertilization formula, the problem of real-time monitoring and fertilization accuracy in the existing technology is solved, real-time detection and precise fertilization of soil fertility are achieved, and the improvement of soil fertility is improved. The efficiency of fertilizer use.
Patent Information
- Application Number
- CN202211628113.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-12-16
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2042-12-16
AI Technical Summary
When monitoring and fertilizing in real time, the existing technology mainly relies on crop image recognition, resulting in low accuracy and high lag, and the inability to effectively identify soil fertility status, resulting in inaccurate use of fertilizers.
The image acquisition module, light wave generation module, light wave recovery module and fertilization control component are configured in the fertilization machine. By identifying the pigment characteristics of the plant leaves, the absorption spectrum is generated, and the light waves are emitted according to the spectrum proportion, the current pigment content of the leaves is calculated, and the soil fertility is combined. The state is planned for fertilization formula and time, and dynamically adjust the fertilization method.
It improves the accuracy of fertilizer use, avoids interference from different bands, accurately identifyes the types of fertilizers required by the leaves, reduces fertilizer waste and pollution, and realizes real-time detection of soil fertility and precise fertilization.
Smart Images

Figure CN116058144B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of smart agriculture, and in particular to a precision fertilization method based on real-time detection of soil fertility. Background Art
[0002] With the development of smart agriculture, the idea of real-time monitoring and fertilization has improved the accuracy of fertilization. However, the current real-time monitoring objects are mostly predicted by crop image recognition, which has a large lag and leads to low accuracy. It is necessary to provide a new method. Summary of the invention
[0003] The embodiment of this specification provides a precision fertilization method based on real-time detection of soil fertility, which is used to improve the accuracy of fertilizer use, specifically including:
[0004] An image acquisition module, a light wave generation module, a light wave recovery module, a data processing module and a fertilization control component are configured in the fertilizer spreader;
[0005] The image acquisition module acquires field images and identifies multiple leaves of each plant, and determines characteristic leaves according to the states of the multiple leaves. The data processing module generates an absorption spectrum according to the light wave absorption characteristics of different pigments, and the absorption spectrum contains the proportion of light wave absorption rates of different frequency bands;
[0006] The corresponding absorption spectrum is determined for each pigment, and the light wave generating module is controlled to generate and emit light waves to the characteristic blade in the order of the proportion in the absorption spectrum from high to low, and after the light wave recovery module receives the reflected light wave, the corresponding absorption spectrum is determined for the next pigment, and the light wave is generated and emitted for recovery, and the light wave recovery module generates a reflection sequence in the order of reception;
[0007] The data processing module compares the real-time generated reflection sequence with a pre-configured reference reflection sequence to calculate the current content of each pigment in the characteristic leaf;
[0008] The fertility status of the soil at the plant is obtained, and a fertilizer formula, fertilizer amount and next fertilization time are planned in combination with the current pigment content of the characteristic leaves, the target content of each pigment and the fertility status of the soil. The fertilizer control component controls the fertilizer machine to fertilize according to the fertilizer formula and fertilizer amount and triggers the next fertilization task at the next fertilization time.
[0009] Optionally, the planning of the fertilization formula, fertilization amount and next fertilization time in combination with the current content of each pigment in the characteristic leaf, the target content of each pigment and the fertility status of the soil includes:
[0010] Scan the soil at the plant site, extract features from the scanned soil image to predict the fertilizer capacity and permeability of the soil at the plant site, construct a search space based on the fertilizer formula, fertilizer amount and next fertilization time, construct a reward function based on the fertilizer capacity of the soil, the permeability of the soil and the target content of each pigment, obtain the future temperature and light sequence to construct the environmental space, search for fertilization strategies in the search space, apply temperature and light in the environmental space to the fertilization strategies, and use the reward function to calculate the reward value of the fertilization strategy after applying the temperature and light, select the fertilization strategy with the largest reward value from multiple fertilization strategies, and determine the fertilization formula and fertilizer amount for this fertilization and the corresponding next fertilization time.
[0011] Optionally, it also includes:
[0012] Establishing a data table for the identification of each plant, and storing all previous fertilization formulas and fertilization amounts in the data table;
[0013] When inspecting the plants, the fertilization record prediction model is used to learn and predict the images of the plants to obtain a fertilization record sequence, scan the plant identification, search the data table of the identification, obtain the previous fertilization formulas and fertilizer amounts in the data table, calculate the similarity with the predicted fertilization record sequence, and determine whether the currently inspected plants are forged based on whether the similarity is greater than a threshold.
[0014] Optionally, it also includes:
[0015] The experimental samples were selected into groups, and different fertilization formulas and fertilization amounts were applied to different groups of experimental samples. Plant images of the samples were collected, and texture density recognition was performed on the plant images. Training labels were set according to the previous fertilization formulas and fertilization amounts of the samples, and the fertilization record prediction model was trained.
[0016] Optionally, comparing the real-time generated reflection sequence with a pre-configured reference reflection sequence to calculate the current content of each pigment of the characteristic leaf includes:
[0017] The current content of each pigment = (ab) c / b, where a is the sum of the reflected light wave intensities of each band calculated for a certain pigment, the light wave intensity calculated based on the reflection sequence generated in real time, b is the sum of the reflected light wave intensities of each band calculated for a certain pigment in the mature leaves of the experiment, and c is the pigment content of the mature leaves.
[0018] Optionally, the using of a reward function to calculate a reward value of the fertilization strategy after applying temperature and light includes:
[0019] The reward function is used to combine the association strength between different fertilizers and different pigments to calculate the reward value of the fertilization strategy after applying temperature and light.
[0020] Optionally, the fertilization control component controls the fertilizer applicator to apply fertilizer according to the fertilization formula and the fertilization amount, including:
[0021] During the fertilization process for the same plant, the fertilization method is adjusted dynamically in real time according to the fertilization performance.
[0022] Optionally, during the fertilization process for the same plant, dynamically adjusting the fertilization method in real time according to the fertilization performance includes:
[0023] After releasing water and fertilizer to the current plant, the water and fertilizer flow shape image is collected, and the flow morphology characteristics are extracted to predict the water absorption parameters of the soil at the current plant. The fertilizer concentration, water and fertilizer release rate and fertilizer spreader movement speed are planned according to the water absorption parameters. The water and fertilizer release rate is lowered for the soil with slower water absorption parameters to reduce the lateral flow ratio of water and fertilizer and increase the longitudinal flow ratio of water and fertilizer.
[0024] The various technical solutions provided in the embodiments of this specification configure an image acquisition module, a light wave generation, a recovery module and a fertilizer control component in the fertilizer spreader, collect images to identify characteristic leaves, emit light waves to the leaves in the order of the proportion of different pigments in the absorption spectrum from high to low, and then emit light waves to the next pigment after receiving the reflected light waves. The light wave recovery module generates a reflection sequence according to the receiving order, and the data processing module compares the real-time generated reflection sequence with the pre-configured reference reflection sequence to calculate the current content of each pigment in the leaves. Combined with the current content of each pigment, the target content and the current fertility status of the soil, the fertilizer formula, the amount of fertilizer and the next fertilization time are planned, and the control component fertilizes according to the plan. Light waves are emitted in sequence according to the spectrum proportion, avoiding interference between different bands, increasing the weight of light waves with a large absorption proportion, more accurately identifying the type of fertilizer required by the leaves, and improving the accuracy of fertilizer use. BRIEF DESCRIPTION OF THE DRAWINGS
[0025] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:
[0026] Figure 1 A schematic diagram of the principle of a precision fertilization method based on real-time detection of soil fertility provided in an embodiment of this specification. DETAILED DESCRIPTION
[0027] Exemplary embodiments of the present invention will now be described more fully with reference to the accompanying drawings. However, exemplary embodiments can be implemented in a variety of forms, and should not be construed as limiting the present invention to the embodiments set forth herein. On the contrary, providing these exemplary embodiments enables the present invention to be more comprehensive and complete, and is more convenient for fully conveying the inventive concept to those skilled in the art. The same reference numerals in the figures represent the same or similar elements, components or parts, and thus their repeated description will be omitted.
[0028] Under the premise of being consistent with the technical concept of the present invention, the features, structures, characteristics or other details described in a specific embodiment do not exclude that they can be combined in one or more other embodiments in a suitable manner.
[0029] In the description of specific embodiments, the features, structures, characteristics or other details described in the present invention are intended to enable those skilled in the art to fully understand the embodiments. However, it does not exclude that those skilled in the art can practice the technical solutions of the present invention without one or more of the specific features, structures, characteristics or other details.
[0030] The flowcharts shown in the accompanying drawings are only exemplary and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps can be decomposed, and some operations / steps can be combined or partially combined, so the actual execution order may change according to actual conditions.
[0031] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities may be implemented in software form, or in one or more hardware modules or integrated circuits, or in different networks and / or processor devices and / or microcontroller devices.
[0032] The term "and / or" or "and / or" includes all combinations of any one or more of the associated listed items.
[0033] Figure 1 A schematic diagram of the principle of a precision fertilization method based on real-time detection of soil fertility provided in an embodiment of this specification, the method may include:
[0034] S101: configuring an image acquisition module, a light wave generating module, a light wave recycling module, a data processing module and a fertilization control component in a fertilizer spreading machine.
[0035] The image acquisition module may be a visible light acquisition module, an infrared acquisition module, or a combination module of multi-band acquisition.
[0036] The light wave generating module can generate light waves according to the required waveband, and the light wave recovery module can sense the intensity of the recovered light waves to generate data.
[0037] S102: The image acquisition module acquires field images and identifies multiple leaves of each plant, and determines characteristic leaves according to the states of the multiple leaves. The data processing module generates an absorption spectrum according to the light wave absorption characteristics of different pigments, and the absorption spectrum has a proportion of light wave absorption rates in different frequency bands.
[0038] A band with a high proportion indicates that the pigment has a better absorption rate in this band, so the reflectivity is lower.
[0039] Characteristic leaves can be mature leaves or new leaves.
[0040] S103: Determine the corresponding absorption spectrum for each pigment, and control the light wave generating module to generate and emit light waves to the characteristic blade in the order of the proportion in the absorption spectrum from high to low, and after the light wave recovery module receives the reflected light wave, determine the corresponding absorption spectrum for the next pigment, generate light waves, and emit and recover them, and the light wave recovery module generates a reflection sequence in the order of reception.
[0041] The order from high to low can accurately reflect the absorption effect of the leaves, thereby accurately characterizing the content of each pigment.
[0042] By emitting light waves in sequence, the coherence effect caused by different pigments absorbing different light waves is avoided, thereby improving accuracy.
[0043] For each pigment, multiple light waves are emitted, but the order in which the light waves are emitted is different.
[0044] The intensity of reflected light waves in different bands can be recorded in the reflection sequence.
[0045] S104: The data processing module compares the reflection sequence generated in real time with a pre-configured reference reflection sequence to calculate the current content of each pigment in the characteristic leaf.
[0046] The current pigment content of the characteristic leaf can be calculated based on the relationship between the pigment content and the reflectivity. The specific relationship between the pigment content and the reflectivity can be obtained through experiments, queries, or machine learning training models without limitation.
[0047] S105: Acquire the fertility status of the soil at the plant, and plan the fertilizer formula, fertilizer amount and next fertilization time in combination with the current pigment content of the characteristic leaves, the target content of each pigment and the fertility status of the soil. The fertilizer control component controls the fertilizer machine to fertilize according to the fertilizer formula and fertilizer amount and triggers the next fertilization task at the next fertilization time.
[0048] Fertility status can characterize the fertilizer balance, permeability and water absorption (water permeability) of the soil.
[0049] The method configures an image acquisition module, a light wave generation and recovery module, and a fertilizer control component in a fertilizer spreader, collects images to identify characteristic leaves, emits light waves to the leaves in the order of the proportions in the absorption spectrum of different pigments from high to low, and then emits light waves to the next pigment after receiving the reflected light waves. The light wave recovery module generates a reflection sequence in the order of reception, and the data processing module compares the real-time generated reflection sequence with the pre-configured reference reflection sequence to calculate the current content of each pigment in the leaves. Combined with the current content of each pigment, the target content, and the current fertility status of the soil, the fertilizer formula, the amount of fertilizer, and the next fertilization time are planned, and the control component fertilizes according to the plan. Light waves are emitted in sequence according to the spectrum proportion, avoiding interference between different bands, increasing the weight of light waves with a large absorption proportion, more accurately identifying the type of fertilizer required by the leaves, and improving the accuracy of fertilizer use.
[0050] The light waves can be light waves obtained by simulating the components of sunlight. Since different pigments require different types of fertilizers, and different pigments can be expressed by the absorption and reflection intensity of light waves, the amount of pigment in the leaves can be identified by the reflection effect of the leaves of the plant on different light waves, and then the type of fertilizer required can be identified.
[0051] In order to make plants grow according to preset goals, obtain the best amount of fertilizer, and avoid pollution and waste, the plant growth process can be simulated in combination with future temperature and light to screen the best fertilizer formula and amount.
[0052] In the embodiment of this specification, the planning of the fertilization formula, the amount of fertilization and the next fertilization time in combination with the current content of each pigment in the characteristic leaf, the target content of each pigment and the fertility status of the soil includes:
[0053] Scan the soil at the plant site, extract features from the scanned soil image to predict the fertilizer capacity and permeability of the soil at the plant site, construct a search space based on the fertilizer formula, fertilizer amount and next fertilization time, construct a reward function based on the fertilizer capacity of the soil, the permeability of the soil and the target content of each pigment, obtain the future temperature and light sequence to construct the environmental space, search for fertilization strategies in the search space, apply temperature and light in the environmental space to the fertilization strategies, and use the reward function to calculate the reward value of the fertilization strategy after applying the temperature and light, select the fertilization strategy with the largest reward value from multiple fertilization strategies, and determine the fertilization formula and fertilizer amount for this fertilization and the corresponding next fertilization time.
[0054] Fertilizer capacity refers to the water retention capacity of the soil. The stronger the water retention capacity of the soil, the greater the fertilizer capacity.
[0055] Permeability refers to the ability to absorb water and fertilizer. The faster the water is absorbed, the stronger the permeability.
[0056] The indicators of the reward function may include the inverse of the amount of fertilizer applied, the degree of completion of the content of each pigment, the matching degree between the fertilizer formula and the type of pigment, the matching degree between the amount of fertilizer applied and the fertilizer capacity, and the matching degree between the fertilization speed and the permeability.
[0057] Since fertilization records will leave traces during the growth process of the plant, the plant itself can actually represent the fertilization records and can be used as a basis for verifying authenticity.
[0058] Therefore, in the embodiment of this specification, the method may further include:
[0059] Establishing a data table for the identification of each plant, and storing all previous fertilization formulas and fertilization amounts in the data table;
[0060] When inspecting the plants, the fertilization record prediction model is used to learn and predict the images of the plants to obtain a fertilization record sequence, scan the plant identification, search the data table of the identification, obtain the previous fertilization formulas and fertilizer amounts in the data table, calculate the similarity with the predicted fertilization record sequence, and determine whether the currently inspected plants are forged based on whether the similarity is greater than a threshold.
[0061] In the embodiments of this specification, it also includes:
[0062] The experimental samples were selected into groups, and different fertilization formulas and fertilization amounts were applied to different groups of experimental samples. Plant images of the samples were collected, and texture density recognition was performed on the plant images. Training labels were set according to the previous fertilization formulas and fertilization amounts of the samples, and the fertilization record prediction model was trained.
[0063] Texture density can be the distance between plant joints. For example, if less fertilizer is applied, the length of the first joint will be smaller. Even if a large amount of fertilizer is applied later, the length of the first joint will not change, and only the new joint will grow rapidly. In the embodiment of this specification, the comparison of the real-time generated reflection sequence with the pre-configured reference reflection sequence to calculate the current content of each pigment of the characteristic leaf includes:
[0064] The current content of each pigment = (ab)c / b, where a is the sum of the reflected light wave intensities of each band calculated for a certain pigment, b is the sum of the reflected light wave intensities of each band calculated for a certain pigment in the mature leaves of the experiment, and c is the pigment content of the mature leaves.
[0065] Specifically, a reference group can be selected to experimentally measure the reflectivity of mature leaves to different light waves.
[0066] In the embodiment of this specification, the method of calculating the reward value of the fertilization strategy after applying temperature and light by using the reward function includes:
[0067] The reward function is used to combine the association strength between different fertilizers and different pigments to calculate the reward value of the fertilization strategy after applying temperature and light.
[0068] The strength of the association between pigments and fertilizers can refer to the fertilizer composition required for a certain pigment, or the proportion of fertilizer ingredients.
[0069] In the embodiment of this specification, the fertilization control component controls the fertilizer applicator to fertilize according to the fertilization formula and the fertilization amount, including:
[0070] During the fertilization process for the same plant, the fertilization method is adjusted dynamically in real time according to the fertilization performance.
[0071] In the embodiments of this specification, during the fertilization process for the same plant, dynamically adjusting the fertilization method in real time according to the fertilization performance includes:
[0072] After releasing water and fertilizer to the current plant, the water and fertilizer flow shape image is collected, and the flow morphology characteristics are extracted to predict the water absorption parameters of the soil at the current plant. The fertilizer concentration, water and fertilizer release rate and fertilizer spreader movement speed are planned according to the water absorption parameters. The water and fertilizer release rate is lowered for the soil with slower water absorption parameters to reduce the lateral flow ratio of water and fertilizer and increase the longitudinal flow ratio of water and fertilizer.
[0073] The water-fertilizer flow shape can characterize water permeability and water retention, and thus can be used to predict water absorption parameters. Water absorption parameters can include water absorption speed.
[0074] For example, if a piece of soil absorbs water too slowly but uses a higher tax release rate, then after the water and fertilizer remain on the ground, a considerable portion of the water and fertilizer will flow along the ground, and the amount that penetrates deep into the lower part of the plant will be reduced, causing pollution and waste. Therefore, we need to combine the soil's water absorption parameters to plan the water and fertilizer release rate and the movement speed of the fertilizer spreader to absorb water and fertilizer with a smaller area.
[0075] The flatter the flow pattern characteristics are, the smaller the water and fertilizer release rate should be planned for it.
[0076] The extracting of flow morphology characteristics to predict the water absorption parameters of the soil at the current plant includes:
[0077] The water and fertilizer flow shape image is input into the pre-built water absorption prediction model to predict the water absorption parameters of the soil at the current plant.
[0078] Among them, soils with different sample permeability and water retention can also be collected and grouped, and different moisture contents are applied to each group of soil to group them again as experimental samples, and water is applied to the soil in each group of experimental samples and the saturation time and water capacity are recorded, and the water and fertilizer flow shape images during water application are collected, and the flow state is marked as training samples, and the water absorption parameters are calculated according to the saturation time and water capacity and the labels of the training samples are set accordingly, and the water absorption prediction model is trained using the training samples and labels.
[0079] Water absorption parameter a = water flow thickness after reaching the ground divided by the water flow spreading area.
[0080] The above-ground and aerial parts of a plant are often associated, so the distribution of roots below ground can be predicted from the aerial parts of the plant.
[0081] In the embodiment of this specification, in the process of fertilizing the same plant, dynamically adjusting the fertilization method in real time according to the fertilization performance also includes:
[0082] Constructing a root distribution prediction model, acquiring the above-ground plant morphology of the current plant by collecting images, inputting the morphology into the root distribution prediction model, predicting the root distribution boundary of the current plant, and determining the current geometric center of the root distribution boundary;
[0083] The outlet depth of the pipe for releasing water and fertilizer is adjusted according to the current geometric center of the root distribution boundary to carry out underground fertilization.
[0084] The geometric center may be the centroid of the solid geometric shape formed by the root distribution boundary.
[0085] In an embodiment of the present specification, the step of adjusting the outlet depth of the pipe for releasing water and fertilizer according to the current geometric center of the root distribution boundary includes:
[0086] Construct a water and fertilizer release simulation model, configure gravity parameters, predicted water absorption parameters and the current geometric center position, search for the pipeline outlet depth within a preset range, simulate the water and fertilizer diffusion release trajectory after applying water and fertilizer pressure parameters and release time, and select the corresponding water and fertilizer pressure parameters and release time and pipeline outlet depth when the boundary of the water and fertilizer release trajectory is the root distribution boundary;
[0087] And according to the water and fertilizer pressure parameters, the corresponding water and fertilizer release rate is obtained, the volume of the water and fertilizer solution is calculated in combination with the water and fertilizer release rate and the release time, the fertilizer concentration is calculated in combination with the water and fertilizer solution volume and the obtained fertilizer application amount, and the movement speed of the fertilizer spreader is calculated in combination with the plant spacing and the release time.
[0088] When the boundary of the simulated water and fertilizer release trajectory is the root distribution boundary, all the released water and fertilizer can be located within the absorption range of the root system, thereby avoiding pollution caused by the fertilization position exceeding the root system range.
[0089] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.
[0090] The above is only an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.
Claims
1. A precision fertilization method based on real-time detection of soil fertility, characterized in that: include: An image acquisition module, a light wave generation module, a light wave recovery module, a data processing module and a fertilization control component are configured in the fertilizer spreader; The image acquisition module acquires field images and identifies multiple leaves of each plant, and determines characteristic leaves according to the states of the multiple leaves. The data processing module generates an absorption spectrum according to the light wave absorption characteristics of different pigments, and the absorption spectrum contains the proportion of light wave absorption rates of different frequency bands; The corresponding absorption spectrum is determined for each pigment, and the light wave generating module is controlled to generate and emit light waves to the characteristic blade in the order of the proportion in the absorption spectrum from high to low, and after the light wave recovery module receives the reflected light wave, the corresponding absorption spectrum is determined for the next pigment, and the light wave is generated and emitted for recovery, and the light wave recovery module generates a reflection sequence in the order of reception; Comparing the real-time generated reflection sequence with the pre-configured reference reflection sequence, and calculating the current content of each pigment of the characteristic leaf, including: Current content of each pigment = (ab) c / b, where a is the sum of the reflected light wave intensities of each band calculated for a certain pigment, the light wave intensity calculated according to the reflection sequence generated in real time, b is the sum of the reflected light wave intensities of each band calculated for a certain pigment in the mature leaves of the experiment, and c is the pigment content of the mature leaves; The data processing module compares the real-time generated reflection sequence with a pre-configured reference reflection sequence to calculate the current content of each pigment in the characteristic leaf; Acquire the fertility status of the soil at the plant, and plan the fertilization formula, fertilization amount and next fertilization time in combination with the current pigment content of the characteristic leaves, the target content of each pigment and the fertilization status of the soil, wherein the fertilization control component controls the fertilizer machine to fertilize according to the fertilization formula and fertilization amount and triggers the next fertilization task at the next fertilization time; The method of planning a fertilization formula, a fertilization amount and a next fertilization time in combination with the current content of each pigment in the characteristic leaf, the target content of each pigment and the fertility status of the soil includes: Scan the soil at the plant site, extract features from the scanned soil image to predict the fertilizer capacity and soil permeability of the soil at the plant site, construct a search space based on the fertilizer formula, fertilizer amount and next fertilization time, construct a reward function based on the fertilizer capacity of the soil, the permeability of the soil and the target content of each pigment, obtain the future temperature and light sequence to construct the environmental space, search for fertilization strategies in the search space, apply temperature and light in the environmental space to the fertilization strategies, and use the reward function to calculate the reward value of the fertilization strategy after applying the temperature and light, select the fertilization strategy with the largest reward value from multiple fertilization strategies, and determine the fertilization formula and fertilizer amount for this fertilization and the corresponding next fertilization time; The method of calculating the reward value of the fertilization strategy after applying temperature and light by using the reward function includes: The reward function is used to combine the association strength between different fertilizers and different pigments to calculate the reward value of the fertilization strategy after applying temperature and light; The fertilization control component controls the fertilizer applicator to fertilize according to the fertilization formula and the fertilization amount, including: During the process of fertilizing the same plant, the fertilization method is adjusted dynamically in real time according to the fertilization performance.
2. The method according to claim 1, characterized in that Also includes: Establishing a data table for the identification of each plant, and storing all previous fertilization formulas and fertilization amounts in the data table; When inspecting the plants, the fertilization record prediction model is used to learn and predict the images of the plants to obtain a fertilization record sequence, scan the plant identification, search the data table of the identification, obtain the previous fertilization formulas and fertilizer amounts in the data table, calculate the similarity with the predicted fertilization record sequence, and determine whether the currently inspected plants are forged based on whether the similarity is greater than a threshold.
3. The method according to claim 2, characterized in that Also includes: The experimental samples were selected into groups, and different fertilization formulas and fertilization amounts were applied to different groups of experimental samples. Plant images of the samples were collected, and texture density recognition was performed on the plant images. Training labels were set according to the previous fertilization formulas and fertilization amounts of the samples, and the fertilization record prediction model was trained.
4. The method according to claim 3, characterized in that In the process of fertilizing the same plant, the fertilization method is adjusted dynamically in real time according to the fertilization performance, including: After releasing water and fertilizer to the current plant, the water and fertilizer flow shape image is collected, and the flow morphology characteristics are extracted to predict the water absorption parameters of the soil at the current plant. The fertilizer concentration, water and fertilizer release rate and fertilizer spreader movement speed are planned according to the water absorption parameters. The water and fertilizer release rate is lowered for the soil with slower water absorption parameters to reduce the lateral flow ratio of water and fertilizer and increase the longitudinal flow ratio of water and fertilizer.
