Method and system for determining a plant protection treatment plan for agricultural plants

Abstract

The present application provides a method for determining a plant protection treatment plan for an agricultural plant, the method being executed by a data processing unit (111), and comprising the steps of: acquiring (S110), by the data processing unit, plant observation data indicative of a current health state of the agricultural plant or a reference plant; acquiring (S120), by the data processing unit, weather data associated with a location where the agricultural plant is grown; predicting (S130), by a computational model (113) executed by the data processing unit, a time-related disease rate of the agricultural plant based on the acquired observation data and the acquired weather data; and determining (S140), by the computational model (113), based on at least the predicted disease rate, at least one plant protection treatment parameter to be included in the plant protection treatment plan.

Description

Technical Field

[0001] This application relates to computer-assisted agricultural plant treatment. In particular, this application relates to a method and device for determining a plant protection treatment plan for agricultural plants. Further, this application relates to a method for adapting a computational model for a changed cultivation condition of an agricultural plant to determine a plant protection treatment plan for the agricultural plant by using a suitable computational model, and a system for treating an agricultural plant based on a plant protection treatment plan assigned to the agricultural plant.

[0002] In agriculture, cultivated plants (especially crops) can be affected by diseases that occur between sowing and harvesting, which can reduce the yield. As a result, plant diseases are mainly caused by three factors. That is, the host plant (affecting the plant's inherent vulnerability), the pathogen (representing the disease-causing agent), and the environmental conditions (which may include the weather preferred by the disease). Mainly these three factors cause diseases, and the manifestation of diseases is a dynamic process among these factors. This complex relationship among the factors alone makes the management of plant diseases difficult. At the same time, agriculture faces the difficulty of feeding an increasing population that is predicted to reach 9 billion by 2050, and agriculture should be as efficient as possible. And important resources such as land, water, and biodiversity are becoming increasingly scarce. Combined with the impact of climate change, extreme weather events are becoming more frequent, increasing plant blights and diseases, and making agriculture difficult.

[0003] However, plants can be maintained healthy by plant protection measures or treatments (such as the application of plant protection agents, insecticides, etc.). Nevertheless, for example, it is difficult to determine the most appropriate timing of plant protection measures, to identify the appropriate plant protection agent, to determine the optimal amount of the plant protection agent, etc.

[0004] Therefore, there is still a need to provide more efficient and effective means to support agriculture in plant disease control and / or health management. Accordingly, one of the objectives of the present invention is to provide more efficient and effective means to support agriculture in plant disease control and / or health management.

[0005] A first aspect of the present invention provides a method (preferably computer-implemented) for determining a plant protection treatment plan for agricultural plants (e.g., crops). The method should be performed by a data processing unit (which may be a processor, a computer device, etc.). The method can be implemented with computer program instructions (for example, provided as computer program elements) and can be performed by, for example, one or more data processing units and / or computer devices. It can also be performed by one or more computer devices included in a distributed computer system. Such a distributed computer system may, in particular, comprise a computing cloud, a client-server system, etc., and computer devices at plant cultivation sites. This means that the distributed computer system may be implemented centrally via cloud computing and / or remotely at plant cultivation sites via edge computing. The method can be performed centrally or remotely, or in a combination of central and remote. In some embodiments, individual computation steps may be performed on different processing units. The computer device may comprise a data processor, memory for storing computer program elements, a data interface, a communication interface, etc. In this specification, data or information may be provided and / or exchanged in electronic form (signals, data packets, etc.) and may be processed electronically by the data processing unit or computer device described above. Data exchange can be performed via a communication network (such as the Internet).

[0006] The method for determining a plant protection treatment plan for agricultural plants comprises the following steps: - The data processing unit acquires plant observation data indicating the current health status of agricultural plants or reference plants. The plant observation data may include one or more individual data points (which may also be subject to data fusion). The observation data does not have to be obtained through direct observation, but may be determined indirectly from databases, datasets, combinations of multiple data sources, etc. Some or all of the above data may be combined or merged with each other. - The data processing unit acquires meteorological data related to the location where agricultural plants are cultivated. As mentioned above, the environmental conditions of plants can typically affect the health of the plants or the development of diseases. Meteorological data can be obtained from meteorological databases, weather records, weather forecasting services, meteorological measurements, satellite data, etc. Meteorological data may also be obtained indirectly through indicators of the geographical location of the plants. Furthermore, meteorological data may include one or more of the following: temperature, humidity, etc. Location may be indicated, for example, by the coordinates of the location, the identifier of the location, or the name of the location (region, city, town name, etc.). Preferably, meteorological data and location are correlated, mapped, combined, etc. - A computational model executed by the data processing unit predicts the time-related disease rates of agricultural plants based on input data (including at least acquired observational and meteorological data). In other words, predictions and / or estimates and / or forecasts are made for the future (e.g., in a time-related manner, with respect to weather) (if so, optionally, to what extent plant diseases are predicted or at least likely to occur). The computational model is understood in a broad sense as a mathematical model used in computational science, particularly those requiring computational resources (provided by the data processing unit above, for example, to study the behavior of complex systems (e.g., agricultural cultivation of plants), especially the probability of plant diseases, for example, by computer simulation). Furthermore, the computational model may be a machine learning model. In this specification, disease rate may include one or more of the following: disease severity, disease incidence, disease risk, etc. Here, disease severity may be understood as the amount of disease, the amount of disease indicators (reflectivity from the plant and / or visible changes on the plant (e.g., visible spots on the plant, so that it can be observed and / or detected by manual visual inspection)), etc. Alternatively, or additionally, observation and / or detection may be performed in at least a semi-automatic manner, for example, using robotic detection devices (aircraft (e.g., drones), agricultural vehicles, etc., having detection means) and / or by satellite imagery. In at least some embodiments, disease detection may be based on so-called normalized difference vegetation index (NDVI) and / or leaf area index (LAI) (image indicators and / or one-sided green leaf area per unit area of ​​ground, adapted for use in the analysis of local measurements (e.g., remote sensing) to assess whether the observed plant contains living green vegetation). Furthermore, disease severity may be understood as a result of a given infection risk over a certain period of time (e.g., days). With respect to disease severity, for example, detection of disease on the upper leaves of a plant is particularly important. This could have a substantial impact on yield. - Determine, by computational model, at least one plant protection treatment parameter to be included in the plant protection treatment plan, based on at least the predicted disease rate. The plant protection treatment parameter and / or plant protection plan are referenced as output data of the computational model and / or as a result of this method. This may subsequently be used as input data for related devices and / or systems, which may further process this data and operate based on this data. For example, the plant protection treatment parameter and / or plant protection plan may also be displayed and / or logged for the user (agricultural company, farmer, etc.). Furthermore, the plant protection treatment parameter and / or plant protection plan may be used as a trigger (trigger signal, message, etc.) configured to trigger a plant protection measure or treatment (e.g., one that can be performed at least partially automated using a robot with detection means (at least semi-automated and / or remotely operable devices (agricultural vehicles, aircraft (e.g., drones), etc.))).

[0007] In this specification, plant diseases can be any undesirable or destructive plant diseases and / or destructive crop diseases. For example, plant diseases may be assigned to or caused by one or more of the following fungi: particularly plant pathogenic fungi, including soilborne pathogens derived from the classes Plasmodiophoromycetes, Peronosporomycetes (synonymous with Oomycetes), Chytridiomycetes, Zygomycetes, Ascomycetes, Basidiomycetes, and Deuteromycetes (synonymous with Fungi imperfecti). The methods described herein are particularly suitable for use in relation to diseases such as wheat leaf blight. Such diseases may, for example, affect plant yields. The impact on yield can be particularly significant if the disease reaches one of the apical leaves, especially one of the top three leaves. Therefore, the method can also map predicted plant protection treatment parameters to estimates for each leaf layer.

[0008] The method allows for the automated adjustment of rules for determining plant protection treatment parameters and / or plant protection plans. This means that the method, and in particular the models used in this method, can be adapted to different situations. For example, it can be adapted without the intervention of human experts to new cultivation sites, new regions, changes in weather conditions, climate change, new or altered diseases, etc. Therefore, according to the method, the timing of application of plant protection agents, insecticides, etc., can be optimized, so agricultural companies, farmers, etc., only need to apply the actual amount, i.e., the smallest possible amount, to the plants, thus protecting yields, reducing costs, and protecting the environment.

[0009] For example, the calculation model can be adapted to other diseases by using the method according to the second embodiment. For example, the plant diseases that can be subsequently predicted may be assigned to or caused by one or more of the following pesticides: Albugo spp. (white rust) in ornamental plants, vegetables (e.g., A. candida) and sunflowers (e.g., A. tragopogonis); Alternaria species. (Alternaria leaf spot disease) spp.) in vegetables (e.g., A. dauci or A. porri), rapeseed (e.g., A. brassicicola or brassicae), sugar beet (A. tenuis), fruit (e.g., A. grandis), rice, soybean, potato and tomato (e.g., A. solani, A. grandis or A. alternata), tomato (e.g., A. solani or A. alternata) and wheat (e.g., A. triticina); species of the genus Aphanomyces in sugar beets and vegetables. spp.); Species of the genus Ascochyta in cereals and vegetables, such as A. tritici (anthracnose) in wheat and A. hordei in barley; Aureobasidium zeae (synonymous with maize brown spot disease (Kapatiella zeae)) in maize; Species of the genus Bipolaris and Drechslera (Teleomorph: Species of the genus Cochliobolus) spp.)) such as sesame leaf spot (D. maydis) or sooty spot (B. zeicola) in maize, for example, leaf spot (B. sorokiniana) in cereals, and for example, B. oryzae (B.oryzae; Blumeria (formerly Erysiphe) graminis (powdery mildew) in cereals (e.g., wheat or barley); Botrytis cinerea (teleomorph: Botryotinia fuckeliana: gray mold) in fruits and berries (e.g., strawberries), vegetables (e.g., lettuce, carrots, celery and cabbage); white spot leaf blight (B. squamosa) or gray mold (B. allii) in onions, rapeseed, ornamental plants (e.g., B. eliptica), grapes, forest plants and wheat; Bremia lactucae in lettuce (Lactucae) (downy mildew); species of the genus Ceratocystis (synonymous with the genus Ophiostoma) (root rot, damping-off disease) in broad-leaved and evergreen trees, e.g. C. ulmi (Ulmus dulcis) in elm; species of the genus Cercospora (spp.) (brown spot disease), such as maize (e.g., gray spot disease: C. zeae-maydis), rice, sugar beet (e.g., C. beticola), sugarcane, vegetables, coffee, soybean (e.g., C. sojina or C. kikuchii) and in rice; species of the genus Cladobotryum (synonymous with the genus Dactylium) in mushrooms. spp.) (e.g., C. mycophilum (formerly Dactylium dendroides, teleomorph: Nectria albertinii, Nectria rosella (synonymous with Hypomyces rosellus)); species of the genus Cladosporium (Cladosporium spp.), such as C. fulvum (leaf mold) in tomatoes and grains, such as C. herbarum (C.herbarum) (black spot disease); Claviceps purpurea in cereals purpurea) (corn corn); species of the genus Cochliobolus (anamorph: Helminthosporium of the genus Bipolaris) (leaf spot disease), such as in maize (C. carbonum), in cereals (e.g. C. sativus, anamorph: B. sorokiniana), and in rice (e.g. C. miyabeanus, anamorph: H. oryzae); species of the genus Colletotrichum (teleomorph: Glomerella) (anthracnose disease), such as in cotton (e.g. C. gossypii), and in maize (e.g. C. graminico) la: anthracnose (root rot), in soft fruits, in potatoes (e.g., C. coccodes: black spot), in legumes (e.g., C. lindemuthianum), in soybeans (e.g., C. truncatum or C. gloeosporioides), in vegetables (e.g., C. lagenarium or C. capsici), in fruits (e.g., C. acutatum), in coffee (e.g., C. coffeanum or C. kahawae), and C. gloeosporioides in various crops; Corticium species in rice. spp.), for example C. sasakii (sheath blight); Corynespora cassiicola (leaf spot) in soybeans, cotton and ornamental plants; species of the genus Cycloconium, for example C. oleaginum (C.oleaginum); species of the genus Cylindrocarpon (e.g., fruit tree ulcer or young grape weakening, teleomorph: species of the genus Nectria (spp.) or Neonectria (spp.)) in fruit trees, in grapes (e.g. C. liriodendri, teleomorph: Neonectria liriodendri, black foot disease) and in ornamental plants; Dematophora (teleomorph: Roselinia) necatrix (root and stem rot) in soybeans; species of the genus Diaporthe in soybeans (spp.), for example, D. phaseolorum (seedling blight); species of the genus Drechslera (synonymous with Helminthosporium, teleomorph: Pyrenophora) in maize, cereals such as barley (e.g. D. teres, net-like spot disease) and wheat (e.g. D. tritici-repentis: yellow-brown spot disease), rice and grass; in grapes, Formitiporia (synonymous with Phellinus) punctata, F. mediterranea, Phaeomoniella chlamydospora Esca disease (canker, apoplexy) caused by chlamydospora (formerly Phaeoacremonium chlamydosporum), Phaeoacremonium aleophilum and / or Botryosphaeria obtusa; species of the genus Elsinoe, such as E. piri in apple fruits and E. beneta in soft fruits.E. ampelina (anthrax) in veneta; Entyloma oryzae (sooty mold) in rice; Epicoccum spp. (black mold) in wheat; Erysiphe spp. (powdery mildew), including E. betae in sugar beets, E. pisi in vegetables, E. cichoracearum in cucurbits, and E. cruciferarum in cabbage and rapeseed; Eutypa rata in fruit trees, grapes and ornamental trees. lata) (euchipah ulcer or canker, anamorph: synonymous with Cytosporina lata, Libertella blepharis); species of the genus Exserohilum (synonymous with Helminthosporium) in maize (e.g., E. turcicum); species of the genus Fusarium (teleomorph: Gibberella) in various plants (damping-off, root or stem rot), for example, F. graminearum or F. curum in cereals (e.g., wheat or barley) F. culmorum) (root rot, black spot, or Fusarium head blight), F. oxysporum in tomatoes, F. solani (differentiated form of glycines, synonymous with the current F. virguliforme), F. tucumaniae and F. brasiliense in soybeans, each causing sudden death syndrome, and F. vertisilioides in maize (F.verticillioides; Gaeumannomyces graminis (damping-off disease) in cereals (e.g., wheat or barley) and maize; Gibberella species (Gibberella spp.) in cereals (e.g., G. zeae) and rice (e.g., G. fujikuroi, bakanae disease); Glomerella cingulata in grapes, apple fruits and other plants, and G. gossypii in cotton; grain stain complex disease in rice; Guignardia bidwellii (black rot) in grapes; Gymnosporangium species in rose family plants and junipers. (spp.), for example, G. sabinae (rust disease) in pears; species of the genus Helminthosporium (synonymous with Drechslera, teleomorph: Cochliobolus) in maize, cereals, potatoes and rice; species of the genus Hemileia (spp.), for example, H. vastatrix (leaf rust disease) in coffee; Isariopsis clavispora (synonymous with Cladosporium vitis) in grapes; Macrophomina phaseolina in soybeans and cotton. Phaseolina (synonymous with phaseoli) (root and stem rot); Microdochium (synonymous with Fusarium) nivale (pink snow mold) in cereals (e.g., wheat or barley); Microsphaera diffusa (powdery mildew) in soybeans; Monilinia spp. species in drupes and other Rosaceae plants, e.g., M. laxa, M. fructicola and M. fructigena (M.fructigena) (synonymous with species of the genus Monilia: flower blight and branch blight, brown rot); species of the genus Mycosphaerella in grains, bananas, soft fruits and peanuts, for example M. graminicola (anamorph: Zymoseptoria tritici, formerly Septoria tritici: Septoria spot disease) in wheat or M. fijiensis (Pseudocercospora fijiensis) in bananas Synonymous with fijiensis: black sigatoka disease) and M. musicola, M. arachidicola (synonymous with M. arachidis or Cercospora arachidis), M. berkeleyi in peas, M. pisi in peas, and M. brassiciola in plants of the Brassicaceae family; species of the genus Peronospora (spp.) (downy mildew), including those in cabbage (e.g., P. brassicae), rapeseed (e.g., P. parasitica), onions (e.g., P. destroyer), tobacco (P. tabacina), and soybeans (e.g., P. manshurica); Phakopsora pachyrhizi and P. meibomiae in soybeans (soybean rust); species of the genus Phialophora (spp.), for example, in grapes (e.g., P. tracheiphila and P. tetraspora) and in soybeans (e.g., P. gregata: stem rot); Phoma lingam in rapeseed and cabbage (Leptosphericum biglobosa (. (Synonymous with Leptosphaeria biglobosa and L. maculans: root and stem rot), and P. betae in sugar beets (root rot, leaf spot, and seedling blight), and P. zeae-maydis in maize (synonymous with Phyllostica zeae); species of the genus Phomopsis (Phomopsis spp.) in sunflowers, grapes (e.g. P. viticola: leaf spot), and soybeans (e.g., stem rot: P. phaseoli, teleomorph: Diaporthe phaseolorum); Physoderma maydis in maize (brown spot); species of the genus Phytophthora in various plants Plasmodiophora brassicae (clubroot) in cabbage, rapeseed, radish, and other plants; species of the genus Plasmopara (Plasmopara). spp.), for example, P. viticola (grape downy mildew) in grapes and P. halstedii in sunflowers; species of the genus Podosphaera (Podosphaera spp.) (powdery mildew), in Rosaceae plants, hops, pome fruits and soft fruits (for example, P. leucotricha in apples) and in Cucurbitaceae plants (P. xanthii); species of the genus Polymyxa (Polymyxa spp.), for example, in cereals, for example, barley and wheat (P. graminis (P.Pseudocercosporella betae in graminis and sugar beets, and the viral diseases transmitted thereby; Pseudocercosporella herpotrichoides in cereals, such as wheat or barley (synonymous with Oculimacula yallundae, O. acuformis: eye spot disease, teleomorph: Tapesia yallundae); Pseudoperonospora in various plants (downy mildew), such as P. cubensis in cucurbitaceae or P. humili in hops; Pseudopezicula tracheiphylla in grapes tracheiphila) (red burn or rotbrenner, anamorph: phialophora); species of the genus Puccinia in various plants P. triticina (brown rust or leaf rust), P. striiformis (striped rust or yellow rust), P. hordei (dwarf rust), P. graminis (stem rust or black rust) or P. recondita (brown rust or leaf rust) in grains such as wheat, barley or rye; P. kuehnii (orange rust) in sugarcane; and P. asparagi in asparagus; species of the genus Pyrenopeziza in rapeseed. spp.), e.g., P. brassicae; Pyrenophora (anamorph: Drechslera) · tritici-repentis (yellowish-brown spot disease) in wheat or P. teres (nettle spot disease) in barley; species of the genus Pyricularia (Pyricularia spp.), e.g., P. oryzae (P.P. grisea (oryzae) (Teleomorph: Magnaporthe grisea, rice rot disease) and P. grisea in grass and cereals; species of the genus Pythium (Pythium spp.) in grass, rice, maize, wheat, cotton, rapeseed, sunflower, soybean, sugar beet, vegetables and various other plants (seedling blight) (e.g. P. ultimum or P. aphanidermatum) and P. oligandrum in mushrooms; species of the genus Ramularia (spp.), for example, R. collo-cygni (ramularia leaf spot disease, physiological leaf spot disease) in barley, R. areola (teleomorph: Mycosphaerella areola) in cotton, and R. beticola in sugar beet; species of the genus Rhizoctonia in cotton, rice, potato, grass, maize, rapeseed, potato, sugar beet, vegetables and various other plants. (spp.), for example, R. solani (root and stem rot) in soybeans, R. solani (sheath blight) in rice, or R. cerealis (Rhizoctonia spring blight) in wheat or barley; Rhizopus stolonifer (black mold, soft rot) in strawberries, carrots, cabbage, grapes and tomatoes; Rhynchosporium secalis and R. commune (burn disease) in barley, rye and rye; Sarocladium oryzae and S. attenuatum (coat rot) in rice; species of the genus Sclerotinia (spp.) (stem rot or white mold disease), which affects vegetables (S. minor and S. sclerotiorum) and crops, such as rapeseed and sunflowers (e.g., S.S. rolfsii (synonymous with Athelia rolfsii) in soybeans, peanuts, vegetables, maize, grains and ornamental plants; Septoria spp. species in various plants, such as S. glycines (brown spot disease) in soybeans and S. tritici (Zymoseptoria tritici) in wheat. (Synonymous with tritici, Septoria spot disease) and S. (synonymous with Stagonospora) nodorum (Stagonospora spot disease) in cereals; Uncinula (synonymous with Erysiphe) necator (powdery mildew, anamorph: Oidium tuckeri) in grapes; species of the genus Setosphaeria (black leaf blight), in maize (e.g., S. turcicum, synonymous with Helminthosporium turcicum) and in turfgrass; species of the genus Sphacelotheca (sooty mold) spp., including those in maize (e.g., S. reiliana: synonymous with Ustilago reiliana: smut), millet and sugarcane; Sphaerotheca fuliginea (synonymous with Podosphaera xanthii: powdery mildew) in cucurbits; Spongospora subterranea (floury crust) in potatoes and the viral diseases transmitted by it; species of the genus Stagonospora in cereals (Stagonospora spp.), such as S. nodorum in wheat (S.(nodorum) (Stagonospora spot disease, teleomorph: Leptospheria [synonymous with Phaeosphaeria] nodorum, synonymous with Septoria nodorum); Synchytrium endobioticum in potatoes (potato crown disease); species of the genus Taphrina, e.g. T. deformans in peaches (leaf curl disease) and T. pruni in plums (plum swelling disease); species of the genus Thielaviopsis in tobacco, apple fruits, vegetables, soybeans and cotton (black root rot), e.g. T. basicola (Chalara elegans (Chalara elegans) Synonymous with elegans); species of the genus Tilletia in cereals (smut or smut), e.g., T. tritici (synonymous with T. caries, wheat smut) and T. controversa (dwarf smut) in wheat; Trichoderma harzianum in mushrooms; Typhula incarnata (gray snow mold) in barley or wheat; species of the genus Urocystis, e.g., U. occulta (smut) in rye; species of the genus Uromyces in vegetables (Rust disease), for example, in legumes (e.g., synonymous with U. appendiculatus and U. phaseoli), in sugar beets (e.g., U. betae or U. beticola) and in leguminous plants (e.g., U. vignae, U. pisi, U. viciae-fabae and U. fabae); species of the genus Ustilago (Ustilago spp.) (Nudity smut), for example, in cereals (e.g., U. nuda and U. avenae).(avaenae), those in maize (e.g., U. maydis: maize smut) and sugarcane; species of the genus Venturia (black spot), those in apples (e.g., V. inaequalis) and pears; and species of the genus Verticillium (damping-off), such as V. longisporum in rapeseed, V. dahliae in strawberries, rapeseed, potatoes and tomatoes, and V. fungicola in mushrooms; and Zymoseptoria tritici in cereals.

[0010] In one embodiment, the input data further includes a soil moisture index, which is acquired by a data processing unit and associated with the location where the agricultural plants are cultivated. For example, the soil moisture index may indicate how wet or moist the soil was at a given time and / or for future times. In other words, the method may include an optional step in which the data processing unit acquires a soil moisture index associated with the location where the agricultural plants are cultivated (provided to the computational model as part of the input data). This would enable more accurate prediction of disease rates.

[0011] In one embodiment, disease rates may be predicted in quarantine values. In other words, the calculation model may predict a value assigned to a specific probability value or range of disease occurrence on the plant, or the course of disease that could spread beyond a specific threshold. The quantitative value representing the disease rate may be, for example, between 0 and 1, between 0 and 100, etc. Alternatively or additionally, the predicted value may be provided as a percentage value from 0 to 100%. Thus, disease rates can be accurately predicted and / or estimated with specific values ​​for a particular time or period within the prediction period. Furthermore, based on the predicted value, a time-related threshold (where the course of disease or the level of disease becomes unacceptable, for example, due to movement to the apical leaves) may be determined.

[0012] In one embodiment, at least one plant protection treatment parameter may include a treatment period or treatment time. In other words, the method can discover the appropriate (preferably most appropriate) treatment timing (for example, the spray timing at which a plant protection agent or insecticide can be applied in order to at least control, eradicate, or prevent the disease, on the one hand, and to use the minimum possible amount of the required plant protection agent or insecticide). The plant protection treatment parameter can therefore also be referred to as the optimal treatment and / or application time.

[0013] In one embodiment, at least one plant protection treatment parameter may further include a day or time window in which the controllability of the disease with respect to a particular plant protection measure exceeds a minimum threshold. This may also be referred to as the optimal treatment and / or application time.

[0014] In one embodiment, the location where agricultural plants are cultivated may be a field (e.g., a soil surface), and the field may be divided into subfields, and disease rates may be predicted for at least some of the subfields in a manner specific to each subfield. For example, a map may be generated which divides the field into multiple subfields and assigns geographic reference data to them. Then, disease rates may be predicted for one, some, or all of the subfields. In this way, plant treatment can be controlled individually for each subfield, which can save, for example, mechanical operation time as a result of protective chemicals and / or mechanical treatment of the plant(s).

[0015] According to one embodiment, at least one plant protection treatment parameter may be determined in a manner specific to each subfield. In this way, at least one plant protection treatment parameter can be determined individually for one, some, or all subfields, which may, for example, save mechanical operation time as a result of protective chemicals and / or mechanical treatment of the plant(s).

[0016] In one embodiment, the soil moisture index may include soil moisture values ​​associated with one or more soil depths. For example, the soil moisture value may indicate how wet or wet the soil in a field or one or more subfields is at a given time and / or for future times. Furthermore, as an example, the soil moisture value may be derived from microwave radiation measurements, and different wavelengths may be used to provide soil moisture values ​​for different depths of soil. Optionally, C-band microwave radiation may be used to provide and / or determine the soil moisture value at the top 2 cm of the soil, X-band microwave radiation may be used to provide and / or determine the soil moisture value at the top 1 cm of the soil, and L-band microwave radiation may be used to provide and / or determine the soil moisture value at the top 5 cm of the soil. The soil moisture value may be used as input data for a computational model. This allows for more accurate prediction of disease rates.

[0017] In one embodiment, the soil moisture index may include soil type. This allows different soil types to be assigned to different disease predispositions for disease probability (for example, by corresponding classifications). This makes it possible to provide a more accurate prediction of disease rates.

[0018] In one embodiment, the soil moisture index may be modeled and / or calculated based on soil data and meteorological data, and the soil data may be one or more of the following data types (related to a field or subfield): soil type, soil quality, soil granularity, soil moisture, soil temperature, soil surface temperature, soil density, soil texture, soil conductivity, soil pH value, and / or soil water retention capacity. This allows different soil data to be assigned to different disease predispositions for the probability of disease (e.g., by corresponding classifications). In this way, a more accurate prediction of disease rates can be provided.

[0019] In one embodiment, the soil moisture index is modeled and / or calculated based on at least soil type and meteorological data. This allows different soil types to be assigned to different disease predispositions for disease probability (e.g., by corresponding classifications). This makes it possible to provide a more accurate prediction of disease rates.

[0020] According to one embodiment, the soil moisture index may be derived at least partially from remote measurements performed on the site where agricultural plants are cultivated. For example, soil moisture (e.g., soil moisture value) may be obtained from satellite data. Optionally, C-band microwave radiation may be used to provide and / or determine the soil moisture value at the top 2 cm of the soil, X-band microwave radiation may be used to provide and / or determine the soil moisture value at the top 1 cm of the soil, or L-band microwave radiation may be used to provide and / or determine the soil moisture value at the top 5 cm of the soil. The soil moisture value may be used as input data for a computational model. This allows for more accurate prediction of disease rates.

[0021] In one embodiment, the soil moisture index may be derived at least partially from local measurements performed at the site where the agricultural plants are cultivated. For example, soil moisture and / or the soil moisture index may be determined using at least one soil moisture sensor. The soil moisture values ​​may be used as input data for a computational model. This would allow for more accurate predictions of disease rates.

[0022] According to one embodiment, the method may further include obtaining a biomass index associated with the location where the agricultural plants are cultivated, and the biomass index may be additionally provided to the computational model as additional input data for predicting disease rates. For example, the biomass index may be a normalized difference vegetation index (NDVI) and / or leaf area index (LAI). This would provide a more accurate prediction of disease rates.

[0023] In one embodiment, the computational model may be a computational regression model. In one embodiment, the specifications of the plant protection agent to be used may be identified using at least one plant protection treatment parameter. For this purpose, an appropriate computational model (or data obtained from a database, etc.) may be used. This further improves the quality of plant protection.

[0024] According to one embodiment, at least one plant protection treatment parameter and / or plant treatment plan may be used as a trigger to inform the user (for example, through a message, warning, etc., relating to the at least one plant protection treatment parameter and / or plant treatment plan) or to instruct the user to perform a specific action (for example, at a specific time) based on the at least one plant protection treatment parameter and / or plant treatment plan.

[0025] In one embodiment, at least one plant protection treatment parameter and / or plant treatment plan may be used to generate a control dataset configured to be provided to a robotic device (which may be configured to automatically execute the plant treatment plan or to apply, for example, plant protection agents, insecticides, etc., on a specific day or time). This can further improve agricultural efficiency.

[0026] According to one embodiment, predicting the disease rate of agricultural plants is further possible. - The calculation model may include predicting the disease progression window, calculating the possible course of the disease in agricultural plants over time within that disease progression window, and providing an index of the disease rate for a specific time within that disease progression window. - Predicted disease rates may be extracted from the disease progression window.

[0027] Stated another way, the computational model can determine a disease progress curve (which may be part of a window). This disease progress curve may be used to identify a threshold (after which the level of disease is unacceptable, e.g., for movement to upper or top leaves of the plant, etc.). Thereafter, this may be used to determine an optimal application time.

[0028] In one embodiment, the plant observation data may include one or more of growth stage, field data, and observed infestation data for an agricultural plant. These data may be combined and / or correlated. According to one embodiment, the field data may include one or more of geographical location information (indicating the geographical location where the agricultural plant is cultivated), and field data (including one or more of soil data, field dimensions, field orientation, field environmental data).

[0029] According to one embodiment, the field data may include one or more of geographical location information (indicating the geographical location where the agricultural plant is cultivated), soil data, field dimensions, field orientation, field environmental data. These data may be combined and / or correlated. In at least some embodiments, the computational model may include several layers (the number of which may be based on the number, plurality, etc. of input data to be processed). If additional layers are introduced, additional data and / or parameters may be considered for prediction. For example, soil moisture, etc. may additionally be considered.

[0030] In one embodiment, the predicted disease rate indicates or includes one or more of disease severity, disease incidence, and disease risk. Preferably, it indicates disease severity. The reason is that a high infection risk on a given day is not necessarily a problem, but a continuously high risk over several days can lead to the emergence of disease. However, disease severity, which is the amount of disease visible on the plant, can be seen as the result of several infection risk days.

[0031] According to one embodiment, the computational regression model utilizes an artificial neural network that outputs data in response to input plant observation data and meteorological data. The neural network may consist of multiple layers, each containing one or more neurons. Neurons between adjacent layers are linked in the sense that the output of a neuron in the first layer becomes the input of one or more neurons in the adjacent second layer. Each such link is given a "weight," and the corresponding input is fed into an "activation function" that gives the neuron's output as a function of the input, with this weight. The activation function is typically a nonlinear function of its input. For example, the activation function may include a "pre-activation function," which is a weighted sum of each input or another linear function, and a threshold function or other nonlinear function that generates the neuron's final output from the values ​​of the pre-activation function. In the neural network used to perform this method, the weights are set or adjusted by training with appropriate training data before making predictions. In at least some embodiments, the neural network can be implemented using techniques such as PyTorch and / or FastAI.

[0032] In at least some embodiments, the computational model may be a time-aware computational model. For example, the model may have a further layer configured to process time-dependent input data (e.g., timestamps that map input values ​​to time or duration). This further improves the learning and / or predictive capabilities of the computational model.

[0033] Furthermore, alternatively or additionally, the computational model may include, or be formed as, a recurrent neural network (RNN) where the connections between nodes form a graph oriented along a temporal sequence. This further improves the learning and / or predictive capabilities of the computational model.

[0034] Alternatively, or additionally, the computational model may include, or be formed as, an artificial recurrent neural network (RNN) architecture, specifically a long short-term memory (LSTM) architecture. This further improves the learning and / or predictive capabilities of the computational model.

[0035] In one embodiment, plant protection treatment parameters and / or plant protection treatment plans may be provided as a computer-readable dataset configured to be executable by a data processing device. For example, plant protection treatment parameters and / or plant protection treatment plans may be provided as a message (e.g., one received by a terminal (e.g., a smartphone or any other suitable computer device)). It may also be used to control a robot for using the protection treatment parameters and / or for executing the plant protection treatment plan.

[0036] In one embodiment, the computational model may further acquire reconnaissance information and / or user feedback collected and / or captured during the growing season. Reconnaissance information may be obtained, for example, from a computer application (App). Reconnaissance information may include, for example, one or more reconnaissance images acquired at the location of the plant. Reconnaissance images may be subject to image processing (e.g., image analysis, pattern recognition, etc.). User feedback may be based, for example, on manual observation, at least semi-automatic detection, etc. It may be entered via the same or a different computer App. In at least some embodiments, the acquired reconnaissance information and / or user feedback may be used to fit and / or calibrate the computational model during the plant growing season. This further improves the learning and / or predictive ability of the computational model.

[0037] A second aspect of the present invention provides a method for adapting a computational model to modified growing conditions for agricultural plants in order to determine a plant protection treatment plan for agricultural plants by using the adapted computational model. The method is preferably computer-implemented and executed by a data processing unit, which may be the data processing unit or computer device described above. The method comprises the step of providing the computational model with training data as input data, the input data including at least one or more of field-specific data, observed disease severity, growth stage data, and meteorological data, and the input data is associated with modified growing conditions for agricultural plants.

[0038] Field-specific data may refer to data collected in experimental trials. Field trials are a standard method in agriculture for studying species diversity, susceptibility, fungicide effectiveness, and the impact of other specific farming activities. For example, as part of these studies, trial operators may devise various plot designs or trial setups, recording various aspects throughout the growing season, which are later analyzed and studied by the trial operators. This study uses data from untreated plots in the trial to study how diseases progress if farmers take no action. This allows for the study of disease dynamics and thus improves device management strategies. As part of this process, planting dates, crop data (crop type, etc.), and trial location details may be further used. These field-specific data may be computer-processed and provided as input data in electronic form. Field-specific data may be acquired electronically by appropriate detection and / or acquisition means (optical detection means, etc.) (e.g., by remote control, or by at least partially autonomous robots, satellite imaging, etc.). Furthermore, field-specific data may be correlated with observed disease severity and / or growth stage data and / or meteorological data.

[0039] In one embodiment, field-specific data includes data relating to planting date and crop data, and crop data includes data relating to crop type, crop variety, crop diversity, crop genetic information, and crop susceptibility to specific diseases. Field-specific data, data relating to planting date, and crop data can be obtained through measurement (including sensor measurements based on, for example, remote sensing, proximal sensing, etc.), modeling, or user input. "Genetic information" is understood as any kind of information relating to the genetic characteristics of an organism and includes, but is not limited to, DNA sequences, RNA sequences, portions of DNA and / or RNA sequences, molecular structures of DNA and / or RNA, epigenetic information (e.g., methylation of DNA portions), information relating to gene mutations, information relating to changes in gene copy number, information relating to gene overexpression, information relating to gene expression levels, information relating to gene shifting, information relating to wild-type and mutant ratios, information relating to ratios between different mutations, information relating to ratios between mutations and other variants (e.g., epigenetic variants), and information relating to ratios between different variants (e.g., epigenetic variants). Furthermore, "genetic information" includes information about the absence of a particular wild type, mutation, or variant (e.g., an epigenetic variant), or a DNA / RNA sequence, or a part of a DNA / RNA sequence, or a particular epigenetic variant.

[0040] Observed disease severity and / or growth stage data may be obtained by observation, determining disease severity at various growth stages of agricultural plants throughout the growing season. Furthermore, observed disease severity and / or growth stage data may be correlated with field-specific data and / or meteorological data.

[0041] Meteorological data may refer to historical meteorological data, or it may be derived from simulated data from, for example, meteorological databases, weather station networks, and appropriate weather models via an appropriate application programming interface (API). Furthermore, meteorological data may be correlated with field-specific data and / or observed disease severity and / or growth stage data.

[0042] Before providing the above data to the computational model as training data, one or more of the following may be pre-processed: field-specific data, observed disease severity, growth stage data, and meteorological data. For example, observations may be made for each observation day, or each value from the same day may be averaged. Furthermore, since the planting date is an important parameter, trials without a planting date may be discarded. In addition, trials may lack geographic coordinate details, and for such trials, the location estimate may be estimated based on further available information about the location (city or town designation) and / or by reverse geographic coding.

[0043] Furthermore, before providing the above data to a computational model as training data, the data may be subjected to disease progression analysis. For example, observations made in trials may aim to capture the temporal disease manifestation (the amount of disease present in the plant population when assessed several times over the growing season). Such assessments can be made for disease severity on different foliage layers. In particular, a weighted sum method can be used to sum the disease severity values ​​specific to each different foliage layer based on their impact on the final yield. This results in a more understandable, simplified, and smooth curve of disease progression over time. Based on these temporal disease manifestation values, a disease progression curve (a collective presentation that plots the dynamics of disease manifestation over time) may be prepared. This temporal progression curve represents the result of complex interactions between the host, pathogen, environment, and crop farming. The method used to describe the temporal disease progression curve is the use of an appropriate growth model.

[0044] - Based on training data, use backpropagation to adjust the parameters or weights of the computational model to fit the modified cultivation conditions of agricultural plants.

[0045] For example, the computational model may be formed as a neural network, or a neural network may be utilized. Generally, a neural network may consist of multiple layers, each containing one or more neurons. Neurons between adjacent layers are linked in the sense that the output of a neuron in the first layer becomes the input of one or more neurons in the adjacent second layer. Each such link is given a "weight," and the corresponding input is fed into an "activation function" that gives the neuron's output as a function of the input, with this weight. The activation function is typically a nonlinear function of its input. For example, the activation function may include a "pre-activation function," which is a weighted sum of each input or another linear function, and a threshold function or another nonlinear function that generates the neuron's final output from the values ​​of the pre-activation function. In the neural network used to perform this method, the weights are set or adjusted by training with appropriate training data before making predictions.

[0046] Furthermore, backpropagation is known in the field of machine learning, for example, referring to algorithms used to train feedforward neural networks for supervised learning. Backpropagation may involve calculating the gradient of the loss function with respect to the network weights to fit the neural network. Backpropagation used to train feedforward networks can perform multiple nonlinear regressions. The goal of the feedforward network is to approximate a function f such that y=f(x;θ), where f maps all input data and parameter θ to a disease rate value y (between 0 and 1, or between 0 and 100%). The backpropagation technique may involve iterative adjustments to parameter θ to minimize the difference between the actual output and the desired output. The input data may be a combination of categorical and continuous values. Continuous values ​​can be used as input data without further preprocessing, while categorical values ​​may benefit from preprocessing. In at least some embodiments, backpropagation can be further improved by representing the values ​​of the categorical column in the form of an N-dimensional vector instead of a single integer. Vectors allow for the capture of more information and the discovery of relationships between multiple categorical values ​​in a more appropriate way. Input data may be fed into a multilayer feedforward neural network. This means that the network contains multiple layers of hidden neurons. Hidden layers are used to increase nonlinearity and modify the better generalization representation of the data on a function. Because this is a complex tabular data analysis task, this layer contains a large number of output neurons (e.g., hundreds or thousands), preferably, in a two-layer network, 500-1500, preferably about 1000, and 200-800, preferably about 500, respectively. This makes, for example, matrix multiplication a linear function. The nonlinearity used is, for example, a rectified linear unit (ReLU). As generalization ability increases, the risk of data overfitting increases. To avoid this, dropout regularization may be used.Alternatively or additionally, so-called batch normalization may be applied after nonlinearity to avoid overfitting. Furthermore, so-called batch normalization may be performed to improve the speed, performance, and stability of the neural network. This is used, in particular, to normalize the input layer by adjusting and scaling the activations. The output layer receives the output activations of the preceding layer (a corresponding number of inputs, e.g., 200-800, preferably about 500) as its input. Optionally, a linear transformation may be performed to obtain a single output, namely a predicted disease rate value in the range between 0 and 1 (which may then be mapped to a value between 0 and 100%). The computational model may be trained using backpropagation in this way.

[0047] - Use a fitted computational model to determine a plant protection treatment plan for agricultural plants by predicting at least one time-related disease rate for agricultural plants.

[0048] After training and adaptation, the calculation model is adapted to the modified cultivation conditions. Therefore, it can be used to accurately determine plant protection measures for agricultural plants in the context of new weather conditions, new diseases, and / or new regions.

[0049] According to one embodiment, this method provides training data as input data before, - A step of combining field-specific data, observed disease severity, growth stage data, and meteorological data into combined data, - The combined data is processed using a weighted sum function, and the disease rate values ​​specific to each different foliage layer are summed based on their impact on the final yield of the agricultural plant. It may be provided.

[0050] A third aspect of the present invention provides a device for determining a plant protection treatment plan for agricultural plants. The device comprises a data interface configured to receive and / or output data, and a data processing unit. The data processing unit - Using a computational model executed by the data processing unit, Based on acquired observational data, and optionally selected soil moisture data and acquired meteorological data, the time-related disease rates of agricultural plants are predicted. - Using a computational model, determine at least one plant protection treatment parameter that should be included in the plant protection treatment plan, based on at least the predicted disease rate. It is configured in this way.

[0051] Preferably, the device may be configured to perform the method of the first embodiment.

[0052] A fourth aspect of the present invention provides a device for adapting a computational model to modified growing conditions for agricultural plants in order to determine a plant protection treatment plan for agricultural plants using the adapted computational model. The device comprises a data interface configured to receive and / or output data, and a data processing unit. The data processing unit is - The computational model obtains training data associated with modified cultivation conditions for agricultural plants, and the training data includes field-specific data, observed disease severity, growth stage data, and optionally, at least one of soil moisture data and meteorological data. - Using backpropagation, adjust the parameters or weights of the computational model to fit the model to the changed cultivation conditions of agricultural plants based on the training data. - Determine a plant protection treatment plan for agricultural plants by predicting at least one time-related disease rate for agricultural plants using a fitted computational model. It is configured in such a way. Preferably, the device may be configured to perform the method of the second embodiment.

[0053] A fifth aspect of the present invention provides a system for treating agricultural plants based on a plant protection treatment plan assigned to the agricultural plants. The system is: - A first device comprising a data interface configured to receive and / or output data, and a data processing unit, wherein the data processing unit is - Based on the acquired observational data, and optionally acquired soil moisture data and meteorological data, the time-related disease rates of agricultural plants are predicted using a computational model executed by the first data processing unit. - Using a computational model, determine at least one plant protection treatment parameter that should be included in the plant protection treatment plan, based on at least the predicted disease rate. - Provides output data that includes at least one plant protection treatment parameter. The first device is configured as follows: - A second device comprising a data interface configured to receive and / or output data, and a data processing unit. The data processing unit is - Obtain output data from the first data processing unit, - Process the output data obtained to use at least one plant protection treatment parameter. It is configured in this way.

[0054] The system may be a distributed computer system, and the first and second devices may be connected via a communication network (such as the Internet). For example, the first device may be a server, a cloud, etc., and may be configured to centrally execute each of the steps described above. Furthermore, the second device may be located remotely from the first device. The second device may be any computer device, terminal (such as a smartphone), controller of a robot device, etc. For example, if the plant protection treatment parameters indicate the timing of plant treatment, the output data of the first device may include messages sent and received by, for example, a terminal, thereby informing the user of the expected timing of plant treatment. Furthermore, the output data of the first device may trigger a robot device to perform plant treatment.

[0055] A sixth aspect of the present invention provides a computer program element for determining a plant protection treatment plan for agricultural plants. The computer program is configured to perform the methods according to the first and / or second aspects when executed by a data processing unit and / or computer device.

[0056] These and other embodiments of the present invention will become apparent and will be described by referring to the following embodiments.

[0057] Hereinafter, exemplary embodiments of the present invention will be described with reference to the accompanying drawings. [Brief explanation of the drawing]

[0058] [Figure 1] A schematic block diagram of a system for treating agricultural plants according to one embodiment of the present invention. [Figure 2] A schematic block diagram of the architecture of a computational model adapted to determine a plant protection treatment plan for agricultural plants, according to one embodiment of the present invention. [Figure 3] A flowchart illustrating a method for determining a plant protection treatment plan for agricultural plants, according to one embodiment of the present invention. [Figure 4] A flowchart illustrating a method for adapting a computational model to modified cultivation conditions for agricultural plants, according to one embodiment of the present invention. [Modes for carrying out the invention]

[0059] The accompanying drawings are for illustrative purposes only and serve only to illustrate the present invention. Identical or equivalent elements are consistently provided by the same reference numerals.

[0060] Figure 1 shows a schematic block diagram of system 100 for treating agricultural plants.

[0061] System 100 includes a first device 110 configured to determine a plant protection treatment plan for agricultural plants (described in more detail later). The first device 110 may be a suitable type of computer and includes a data interface 111 configured to receive and / or output data, and a data processing unit 112. It may also include data storage, memory, etc. Optionally, the data interface 111 may be configured to communicate over a communication network (such as the Internet). In some embodiments, the first device 110 may form or be part of a computing cloud, server, etc. In other embodiments, the first device 110 may be a local computer device. The first device 110 is configured to computationally execute a computational model 113 (see, for example, Figure 2) adapted to determine a plant protection treatment plan for agricultural plants, as will be described in more detail later.

[0062] Furthermore, the system 100 includes a second device 120 configured to at least acquire and / or process output data acquired from the first device 110. In other words, the output data from the first device 110 may be used by the second device 120 for the treatment of agricultural plants. For example, the second device 120 may receive a plant protection treatment plan for agricultural plants from the first device 110. The second device 120 may be a suitable type of computer and includes a data interface 121 configured to receive and / or output data, and a data processing unit 122. It may also include data storage, memory, etc. Optionally, the data interface 121 may be configured to communicate over a communication network (such as the Internet). In at least some embodiments, the second device 140 may be located remotely from the first device 110 and / or the second device 120 (for example, at or near the location of agricultural plants). Furthermore, optionally, the second device 120 may be a terminal such as a smartphone or a robotic device.

[0063] Furthermore, the system 100 includes, or is operably connected to, at least one data source 130 configured to collect or provide data to be input to the first device 110 and / or the second device 120. The data source 130 may, exemplary, represent several different data sources (weather stations, weather station networks, databases containing observed plant data, etc.). It may also include training data comprising at least one or more of field-specific data, observed disease severity, growth stage data, and meteorological data, the training data being associated with modified cultivation conditions for agricultural plants. Furthermore, the data source 130 may also include plant observation data indicating the current health status of agricultural plants or reference plants and meteorological data associated with the location where the agricultural plants are cultivated.

[0064] The data source 130, the first device 110, and / or the second device 120 are at least partially operably connected to each other, as indicated by the arrows between each entity shown in Figure 1, and the data flow between entities is identified by the direction of the arrows.

[0065] The system 100 described above may operate as described below.

[0066] The first device 110 may be configured to determine a plant protection treatment plan for agricultural plants. Specifically, the first device 110 is configured, for example, by a data processing unit 112, to acquire plant observation data from a data source 130 via a data interface 111 that indicates the current health status of agricultural plants or reference plants. Furthermore, the first device 110 is configured, by a data processing unit 112, to acquire meteorological data from the data source 130 via the data interface 111 that is associated with the location where the agricultural plants are cultivated. The first device 110 is further configured, by the calculation model 113 (preferably stored or loaded in, for example, the data storage unit of the first device 110 and executed by the data processing unit 112), to predict the time-related disease rate of agricultural plants based on the acquired observation data, the acquired meteorological data, and optionally a soil moisture index. Furthermore, the first device 110 is configured, by the calculation model 113, to determine at least one plant protection treatment parameter that should be included in the plant protection treatment plan based on at least the predicted disease rate. The computational model 113 may further be formed as a neural network adapted to output data in response to input plant observation data and meteorological data, or it may utilize such a neural network. The computational model 113 may be configured to process the input data to calculate the disease rate as a quantitative value (e.g., a value from 0 to 1, or from 0 to 100). In at least some embodiments, at least one plant protection treatment parameter includes a treatment period or treatment time. For example, at least one plant protection treatment parameter includes a day or time window in which the controllability of disease with respect to a particular plant protection measure exceeds a minimum threshold. Furthermore, in at least some embodiments, the first device 110 and / or computational model 113 may be adapted by the computational model 113 to predict a disease progression window (in which the possible course of disease in agricultural plants over a period of time is calculated and the disease rate for a specific time within the disease progression window is shown), and the predicted disease rate is extracted from the disease progression window.Furthermore, in at least some embodiments, the first device 110 and / or the computational model 113 are configured to process plant observation data using a weighted sum function configured to sum disease rates specific to each different foliage layer based on their impact on plant yield. For example, the plant observation data includes one or more of the following: field data, observation infestation data, and growth stage data associated with agricultural plants. Furthermore, the predicted disease rate represents or includes one or more of the following: disease severity, disease incidence, and disease risk. Plant protection treatment parameters and / or plant protection treatment plans are provided as computer-readable datasets configured to be executed by a data processing device (e.g., a second device 120).

[0067] Optionally, the place where agricultural plants are cultivated may be a field, and the field may be divided into several subfields, and disease rates may be predicted in a manner specific to each subfield for at least some of the subfields. For example, the field may be divided using a map (e.g., a digital and / or computer-readable map) that shows several different subfields. Based on the division into several subfields, at least one plant protection treatment parameter may be determined in a manner specific to each subfield, for example, at least one protection treatment parameter may be determined individually for each subfield.

[0068] Optionally, the soil moisture index includes soil moisture values ​​associated with one or more soil depths. In at least some embodiments, the soil moisture index may include soil type. Optionally, the soil moisture index is modeled, predicted, and / or calculated based on at least one soil type and meteorological data. Furthermore, the soil moisture index may be derived, at least in part, from remote measurements performed on the site where agricultural plants are cultivated. Alternatively, or additionally, the soil moisture index may be derived, at least in part, from local measurements performed on the site where agricultural plants are cultivated.

[0069] In at least some embodiments, a biomass index associated with the location where the agricultural plants are cultivated (e.g., LAI and / or NDVI, which may be derived from satellite data) may be obtained. This may provide the biomass index to the computational model 113 as additional input data for predicting disease rates.

[0070] The computational model 113 performed by the first device 110 may be adapted to the modified growing conditions of agricultural plants in order to determine an appropriate plant protection treatment plan for agricultural plants using the adapted computational model 113. For this purpose, the first device 110 is configured to acquire training data (including at least one of field-specific data, observed disease severity, growth stage data, and meteorological data) by the computational model 113 (for example, via the data interface 112), and the training data is associated with the modified growing conditions of agricultural plants. Furthermore, the first device 110 is configured to adjust the parameters or weights of the computational model 113 to adapt the computational model 113 to the modified growing conditions of agricultural plants using backpropagation based on the training data. Subsequently, as described above, a plant protection treatment plan for agricultural plants may be determined by predicting at least the time-related disease rate of agricultural plants using the adapted computational model 113.

[0071] Figure 2 is a schematic block diagram of an exemplary architecture (here a multilayer neural network) of the computational model 113 described above. For example, computational model 113 is a two-layer feedforward neural network configured to be trained by backpropagation (e.g., a backpropagation algorithm). Thus, computational model 113 comprises a first layer 113A and a second layer 113B. Input data (which may include categorical values ​​(see block 113C) and continuous values ​​(see block 113D)) is fed into the neural network, and in particular into the first layer 113A, as shown as blocks 113C and 113D in Figure 2. The first layer 113A and the second layer 113B may be interconnected. The first layer 113A and the second layer 113B may each include a linear function (e.g., matrix multiplication) and a nonlinear function (e.g., rectified linear unit (ReLU)). The output of the calculation model 113 via block 113E may be the predicted at least one plant protection treatment parameter to be included in the plant protection treatment plan described above, or it may be a complete plant protection treatment plan that includes at least one plant protection treatment parameter.

[0072] Figure 3 shows a flowchart of a method for determining a plant protection treatment plan for agricultural plants. It should be noted that the following method steps (in particular acquiring input data) do not need to be performed in a specific order, and the input data may be acquired in different orders. In step S110, plant observation data indicating the current health status of the agricultural plant or reference plant is acquired, for example, by the data processing unit 111. In step S120, meteorological data associated with the location where the agricultural plant is cultivated is acquired, for example, by the data processing unit 111. Optionally, soil moisture indices associated with the location where the agricultural plant is cultivated may also be acquired, for example, by the data processing unit 111. In step S130, based on the input data, which includes at least the acquired observation data and acquired meteorological data, and optionally the acquired soil moisture indices, the time-related disease rate of the agricultural plant is predicted by a calculation model 113 run by the data processing unit 111. In step S140, based on at least the predicted disease rate, at least one plant protection treatment parameter to be included in the plant protection treatment plan is determined, for example, by the calculation model 113 (run by the data processing unit 111).

[0073] Figure 4 is a flowchart illustrating how to adapt the adapted computational model 113 to modified cultivation conditions for agricultural plants in order to determine a plant protection treatment plan for agricultural plants. In step S210, training data is acquired by the computational model 113, which includes at least one of the following: field-specific data, observed disease severity, growth stage data, and meteorological data, and optionally includes soil moisture index, and the training data is associated with the modified cultivation conditions for agricultural plants. In step S220, based on the training data, the parameters or weights of the computational model 113 are adjusted using backpropagation to adapt the computational model 113 to the modified cultivation conditions for agricultural plants. In step S230, the adapted computational model 113 is used to determine a plant protection treatment plan for agricultural plants, at least by predicting the time-related disease rate of agricultural plants.

[0074] It should be noted that each embodiment of the present invention is described by reference to various subjects. In particular, some embodiments are described by reference to method-type claims, and other embodiments are described by reference to device-type claims. However, those skilled in the art will understand from the above and below descriptions that, unless otherwise notified, any combination of features belonging to one type of subject, as well as combinations of features relating to different subjects, are disclosed in this application. However, all features, when combined, provide a synergistic effect that goes beyond the simple addition of features.

[0075] Although the present invention has been illustrated and described in detail in the drawings and the above description, such illustrations and descriptions should be considered illustrative or illustrative, and not limiting. The present invention is not limited to the embodiments disclosed. Other variations of the embodiments disclosed will be understood and realized by those skilled in the art in carrying out the invention described in the claims, from a study of the drawings, disclosure and dependent claims.

[0076] In the claims, the term “equipped with” does not exclude other elements or steps, and the indefinite article “a” or “an” does not exclude plurals. A single processor or other unit may fulfill the functions of several items described in the claims. The mere fact that certain means are described again in different dependent claims does not imply that a combination of those means cannot be used for benefit. No reference numeral in the claims should be considered limiting.

Claims

1. 1. A method for determining a plant protection treatment regimen for an agricultural plant, comprising: The method is performed by a data processing unit (111), The method comprises: obtaining plant observation data indicating the current health status of the agricultural plants or reference plants by the data processing unit (S110); obtaining, by the data processing unit, meteorological data associated with a location where the agricultural plants are grown (S120); predicting (S130) a time-related disease rate of the agricultural plants based on input data comprising at least the acquired observation data and the acquired meteorological data by a computational model (113) executed by the data processing unit; determining (S140) at least one plant protection treatment parameter to be included in the plant protection treatment plan based on at least the disease rate predicted by the computational model (113); A method comprising:

2. The method of claim 1 , wherein the input data further comprises a soil moisture index obtained by the data processing unit and associated with the location where the agricultural plants are grown.

3. 3. The method according to claim 1 or 2, wherein the at least one plant protection treatment parameter comprises a treatment duration or a treatment time.

4. 4. The method according to any one of claims 1 to 3, wherein said at least one plant protection treatment parameter comprises a day or time window during which controllability of said disease for a particular plant protection measure exceeds a minimum threshold.

5. the location where the agricultural plants are grown is a field; The field is divided into several subfields, The disease rate is predicted in a subfield-specific manner for at least some of the subfields; The method according to any one of claims 1 to 4.

6. The method of claim 5 , wherein the at least one plant protection treatment parameter is determined in a specific manner for each subfield.

7. The method of any one of claims 1 to 6, wherein the soil moisture index comprises a soil moisture value associated with one or more soil depths.

8. The method of any one of claims 1 to 7, wherein the soil moisture indicator comprises soil type.

9. The method of any one of claims 1 to 8, wherein the soil moisture index is modelled based on at least soil type and the meteorological data.

10. The method of any one of claims 1 to 9, wherein the soil moisture index is at least partly derived from remote measurements performed on the location where the agricultural plants are grown.

11. The method according to any one of claims 1 to 10, wherein the soil moisture index is at least partly derived from local measurements carried out at the location where the agricultural plants are grown.

12. obtaining a biomass index associated with the location where the agricultural plants are grown; The biomass index is additionally provided to the computational model (113) as additional input data for predicting the disease rate. The method according to any one of claims 1 to 11.

13. The step of predicting the disease rate of the agricultural plant further comprises predicting, by the computational model (113), a disease progression window in which the likely course of a disease of the agricultural plant over a time period is calculated, and an index relating to the disease rate for a specific time within the disease progression window; The predicted disease rate is extracted from the disease progression window. The method according to any one of claims 1 to 12.

14. The method according to any one of claims 1 to 13, wherein the plant observation data is obtained and / or processed leaf layer by leaf layer.

15. the plant observation data is weighted or classified for different leaf layers of the agricultural plant or the reference plant based on the influence of the different leaf layers on the yield of the agricultural plant; the disease rate is predicted based on the weighted or categorized plant observation data; The method according to any one of claims 1 to 14.

16. The method of any one of claims 1 to 15, wherein the plant observation data comprises one or more of growth stage data, field data, and observed infestation data associated with agricultural plants.

17. 17. The method of any one of claims 1 to 16, wherein the predicted disease rate indicates or comprises one or more of disease severity, disease prevalence, and disease risk.

18. 18. The method of any one of claims 1 to 17, wherein the computational model (113) utilizes a neural network adapted to output data in response to the plant observation data and meteorological data as inputs.

19. 19. The method according to any one of claims 1 to 18, wherein the at least one plant protection treatment parameter and / or the plant protection treatment plan are provided as a computer-readable data set configured to be executed by a data processing device of a robotic device for applying a plant protection agent on a specific day or time.

20. 20. The method according to any one of claims 1 to 19, wherein the at least one plant protection treatment parameter and / or the plant treatment plan is used as a trigger for automatically informing a user of the at least one plant protection treatment parameter and / or the plant treatment plan and / or for instructing the user to perform a certain action at a certain time.

21. 1. A method for adapting a computational model (113) to changed growing conditions of an agricultural plant in order to determine a plant protection treatment regime for the agricultural plant using said adapted computational model (113), comprising: The method is performed by a data processing unit (111), The method comprises: obtaining, by the computational model (113), training data including at least one or more of field-specific data, observed disease severity, growth stage data, and meteorological data, wherein the training data is associated with modified cultivation conditions of the agricultural plants; - adjusting parameters or weights of the computational model (113) based on the training data using back propagation to adapt the computational model (113) to the changed cultivation conditions of the agricultural plants; - determining the plant protection treatment regime for the agricultural plants by predicting at least one time-related disease rate of the agricultural plants using the adapted computational model (113); A method comprising:

22. The method further comprises, before providing the training data as input data, processing the plant observation data with a weighted sum function configured to sum disease rates specific to each different leaf layer based on their impact on yield of the plant; 22. The method of claim 21.

23. A device (110) for determining a plant protection treatment regime for an agricultural plant, comprising: The device comprises a data interface (112) configured to receive data and / or output data, and a data processing unit (111), The data processing unit (111) predicting a time-related disease rate of said agricultural plants using a computational model (113) executed by said data processing unit based on the acquired observational data and the acquired meteorological data; determining, based on at least the predicted disease rate, using the computational model (113) at least one plant protection treatment parameter to be included in the plant protection treatment plan; The device is configured to:

24. 1. A system for treating agricultural plants based on a plant protection treatment plan assigned to the agricultural plants, comprising: A first data processing unit (111) and a second data processing unit (121), The first data processing unit (111) predicting a time-related disease rate of said agricultural plants using a computational model (113) executed by said first data processing unit (111) based on the acquired observational data and the acquired meteorological data; determining, using said computational model (113), based on at least said predicted disease rate, at least one plant protection treatment parameter to be included in said plant protection treatment plan; providing output data including at least said at least one plant protection treatment parameter. It is configured as The second data processing unit (121) Obtaining the output data from the first data processing unit (111); processing the obtained output data to use the at least one plant protection treatment parameter; It is configured as follows: system.

25. 23. A computer program element for determining a plant protection treatment regime for agricultural plants, said computer program being configured to perform the method according to any one of claims 1 to 22 when executed by a data processing unit and / or a computing device.

Images