CN113435728B

CN113435728B - Farm insect pest searching and killing method and system

Info

Publication number: CN113435728B
Application number: CN202110694425.7A
Authority: CN
Inventors: 孙彤; 黄桂恒
Original assignee: Brake Agricultural Big Data Technology Group Co ltd
Current assignee: Brake Agricultural Big Data Technology Group Co ltd
Priority date: 2021-06-22
Filing date: 2021-06-22
Publication date: 2022-04-12
Anticipated expiration: 2041-06-22
Also published as: CN113435728A; CN114819568B; CN114819568A

Abstract

The embodiment of the specification provides a farm pest searching and killing method, system, device and storage medium. The method comprises the steps of acquiring farm information; determining a plurality of pest types based on farm information; determining a plurality of searching and killing schemes corresponding to a plurality of insect pest types, and acquiring at least one of crop influence characteristics, soil influence characteristics and pesticide effect characteristics corresponding to the plurality of searching and killing schemes; determining a plurality of combination schemes based on a plurality of combination modes of the plurality of searching and killing schemes; obtaining at least one of crop growth potential influence scores and soil quality condition influence scores corresponding to the multiple combination schemes through a prediction model; determining a target combination scheme according to the at least one score.

Description

Farm insect pest searching and killing method and system

Technical Field

The specification relates to the field of intelligent agriculture, in particular to a farm pest searching and killing method and system.

Background

In modern large-scale farms, the pest control industry is at the beginning of the day. Whether pest is checked and killed timely or not influences the ecological environment and the crop growth of the farm to a great extent, and if the pest situation is complex, multiple pests may appear at the same time, the pest risk needs to be pre-warned timely, and a pest checking and killing scheme needs to be accurately determined according to the farm situation.

Therefore, a method for searching and killing farm insect pests is needed.

Disclosure of Invention

One embodiment of the specification provides a farm pest searching and killing method. The farm pest searching and killing method comprises the following steps: acquiring farm information, wherein the farm information comprises crop information and environment information; determining pest conditions based on the farm information, wherein the pest conditions include a plurality of pest types; determining a plurality of searching and killing schemes corresponding to a plurality of insect pest types, and acquiring at least one of crop influence characteristics, soil influence characteristics and pesticide effect characteristics corresponding to the plurality of searching and killing schemes; determining a plurality of combination schemes based on a plurality of combination modes of the plurality of searching and killing schemes; obtaining corresponding fusion characteristics based on the at least one characteristic included in each combination scheme; obtaining at least one of crop growth potential influence scores and soil quality condition influence scores corresponding to the multiple combination schemes through a prediction model based on the fusion characteristics and the farm information; determining a target combination scheme according to the at least one score.

In some embodiments, the determining a plurality of combination scenarios based on a plurality of combinations of the plurality of killing scenarios comprises: obtaining the various combinations of the various searching and killing schemes according to permutation and combination; determining a preset condition based on the farm information; and taking a plurality of combinations which meet the preset condition in the plurality of combinations as the plurality of combination schemes.

In some embodiments, the determining pest conditions based on the farm information comprises: and determining the corresponding multiple insect pest types through an insect pest determination model based on the farm information.

In some embodiments, the obtaining farm information comprises: setting a plurality of sampling points in at least one area of the farm; obtaining multiple groups of corresponding crop data and multiple groups of corresponding environmental data at the multiple sampling points through a monitoring device; the method comprises the steps that multiple groups of crop data and multiple groups of environment data of multiple sampling points are collected through an unmanned aerial vehicle, statistics analysis is conducted on the multiple groups of crop data and the multiple groups of environment data based on corresponding weights of the sampling points, farm information is obtained, and the weights are related to the density of the sampling points in the area where the sampling points are located.

In some embodiments, further comprising: acquiring stage farm information of at least 1 time stage through a monitoring device in a period of time for pest searching and killing by adopting the target combination scheme; determining at least one stage score in stage crop growth potential influence scores and stage soil health influence scores of the target combination scheme through the prediction model based on the fusion features and the stage farm information; determining the density of sampling points and/or adjusting a sampling trajectory of the at least one region for a next time phase based on the at least one phasic score for a current time phase.

In some embodiments, further comprising: obtaining farm information change values before and after the objective combination scheme is adopted for checking and killing based on the multiple groups of crop data and the multiple groups of environmental data of the multiple sampling points; obtaining an evaluation result of the target combination scheme through an evaluation model based on the farm information change value; based on the evaluation result, model parameters of the prediction model are adjusted.

In some embodiments, the model parameters of the prediction model include a plurality of scoring weights corresponding to the plurality of combination schemes, and the adjusting the model parameters of the prediction model based on the evaluation result includes: and adjusting the scoring weight corresponding to the target combination scheme based on the evaluation result.

One of the embodiments of this specification provides a farm pest system of searching and killing, the farm pest system of searching and killing includes: the system comprises an acquisition module, a storage module and a management module, wherein the acquisition module is used for acquiring farm information which comprises crop information and environment information; the judging module is used for determining insect pest situations based on the farm information, wherein the insect pest situations comprise various insect pest types; the characteristic extraction module is used for determining a plurality of searching and killing schemes corresponding to a plurality of insect pest types and acquiring at least one characteristic of crop influence characteristics, soil influence characteristics and pesticide effect characteristics corresponding to the plurality of searching and killing schemes; the combination module is used for determining a plurality of combination schemes based on a plurality of combination modes of the plurality of searching and killing schemes; a feature fusion module, configured to obtain a corresponding fusion feature based on the at least one feature included in each combination scheme; the scoring module is used for obtaining at least one of crop growth potential influence scores and soil quality condition influence scores corresponding to the multiple combination schemes through a prediction model on the basis of the fusion characteristics and the farm information; and the determining module is used for determining a target combination scheme according to the at least one score.

A farm pest killing device, the device comprising at least one processor and at least one memory; the at least one memory is for storing computer instructions; the at least one processor is used for a farm pest searching and killing method.

One embodiment of the present disclosure provides a computer-readable storage medium storing computer instructions, wherein when the computer instructions in the storage medium are read by a computer, the computer executes a farm pest searching and killing method.

Drawings

The present description will be further explained by way of exemplary embodiments, which will be described in detail by way of the accompanying drawings. These embodiments are not intended to be limiting, and in these embodiments like numerals are used to indicate like structures, wherein:

fig. 1 is an exemplary flow chart of a method for farm pest extermination according to some embodiments of the present disclosure;

FIG. 2 is an exemplary flow diagram of a method of adjusting model parameters of a predictive model according to some embodiments described herein;

fig. 3 is a schematic view of an application scenario of a farm pest killing system according to some embodiments of the present disclosure;

fig. 4 is a block diagram of a farm pest killing system according to some embodiments of the present disclosure.

Detailed Description

In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the drawings used in the description of the embodiments will be briefly described below. It is obvious that the drawings in the following description are only examples or embodiments of the present description, and that for a person skilled in the art, the present description can also be applied to other similar scenarios on the basis of these drawings without inventive effort. Unless otherwise apparent from the context, or otherwise indicated, like reference numbers in the figures refer to the same structure or operation.

It should be understood that "system", "apparatus", "unit" and/or "module" as used herein is a method for distinguishing different components, elements, parts, portions or assemblies at different levels. However, other words may be substituted by other expressions if they accomplish the same purpose.

As used in this specification and the appended claims, the terms "a," "an," "the," and/or "the" are not intended to be inclusive in the singular, but rather are intended to be inclusive in the plural, unless the context clearly dictates otherwise. In general, the terms "comprises" and "comprising" merely indicate that steps and elements are included which are explicitly identified, that the steps and elements do not form an exclusive list, and that a method or apparatus may include other steps or elements.

Flow charts are used in this description to illustrate operations performed by a system according to embodiments of the present description. It should be understood that the preceding or following operations are not necessarily performed in the exact order in which they are performed. Rather, the various steps may be processed in reverse order or simultaneously. Meanwhile, other operations may be added to the processes, or a certain step or several steps of operations may be removed from the processes.

The embodiment of the specification relates to a farm pest searching and killing method and system. The method and the system for searching and killing the farm insect pests can be applied to searching and killing the insect pests in farmlands, flower fields, vegetable fields, forest lands, terraced fields, wetlands, grasslands, orchards, tea gardens, nurseries, gardens, green belts and the like. In some embodiments, the farm pest extermination methods and systems can be applied to food crops (e.g., rice, corn, beans, potatoes, highland barley, broad bean, wheat, etc.), oil crops (e.g., oilseeds, cranberries, mustard, peanuts, flax, hemp, sunflowers, etc.), vegetables (e.g., radish, cabbage, celery, leeks, garlic, shallots, carrots, melons, lotus, jerusalem artichoke, sword bean, coriander, lettuce, dayflowers, peppers, cucumbers, tomatoes, coriander, etc.), fruits (e.g., pears, greens, apples, peaches, apricots, walnuts, plums, cherries, strawberries, amomum fruits, red dates, etc.), feed crops (e.g., corn, green manure, milk vetch, etc.), medicinal crops (e.g., ginseng, angelica, honeysuckle, mint, artemisia wormwood, etc.), and wild fruits (e.g., sour pears, wild apricots, wild peaches, wild jujubes, mountain dates, cherries, wild cherries, etc.) Sea buckthorn, etc.) and the like. Through the method and the system for searching and killing the farm insect pests, the following effects can be achieved: the method has the advantages of rapidly and accurately reporting the farm site condition, giving early warning to farmers in time, providing an optimal searching and killing scheme, monitoring the searching and killing process and the like. In some embodiments, the farm pest killing method and system can be applied to other fields, such as breeding, livestock raising, environmental protection and the like. The farm pest killing method and system can provide services such as soil monitoring, water quality monitoring, air quality monitoring and the like.

Fig. 1 is an exemplary flow chart of a method for farm pest extermination according to some embodiments of the present disclosure. As shown in fig. 1, the process 100 includes the following steps. In some embodiments, process 100 may be performed by processor 312.

Step 110, acquiring farm information, wherein the farm information comprises crop information and environment information.

The farm information is information related to a farm. In some embodiments, farm information may include crop information and environmental information. In some embodiments, the farm information may also include farm remote sensing images, fertigation information, historical pest killing information, historical pesticide parameters, and the like.

Crop information is information about crops grown in a farm. In some embodiments, the crop information may include the type of crop, foliage conditions, growth period, planting area, spatial distribution, and the like. The environmental information is information related to farm environment. In some embodiments, the environmental information may include soil information (e.g., soil type such as brick red soil, yellow brown soil, gray brown soil, etc., soil texture such as the type and combination of different mineral particles in the soil, soil layer thickness, content of various elements in the soil, etc.), air humidity information, climate information, cultivation conditions, etc.

In some embodiments, crop related data, soil related data, air humidity related data may be collected by monitoring by a monitoring device, by unmanned aerial vehicle patrol, and the like. In some embodiments, the data may be analyzed based on the collected data, either directly or indirectly (e.g., image recognition of the acquired image, analysis of the data, etc.) to obtain crop information and environmental information. Wherein, monitoring devices can refer to equipment with monitoring functions, such as a monitoring camera, an infrared detector, a soil content detection device, and the like.

In some embodiments, the monitoring device can retain the monitoring data of the sampling point, and the unmanned aerial vehicle can acquire the monitoring data of the sampling point through the scanning monitoring device.

For more on the basis of sampling points, monitoring devices, unmanned aerial vehicles to obtain farm information, reference may be made to fig. 3 and its associated description.

In some embodiments, farm information may also be obtained by other means. For example, farm information may be obtained by remote sensing images. Specifically, the remote sensing image can be obtained by modes of aerial photography, aerial scanning, microwave radar and the like, and farm information can be obtained through image processing.

And 120, determining pest situations based on the farm information, wherein the pest situations comprise multiple pest types.

The pest situation is a situation where pests occur in a farm and may include a pest situation, a situation where the pest inflicts harm on crops, and the like. For example, the pest condition may include a condition that the pest is a red spider and leaves are bitten, leaves are shriveled, leaves are green, leaves are moldy, leaves are dropped, roots are distorted, roots are rotted, crop dysplasia, and the like.

In some embodiments, the pest condition may include a plurality of pest types, and the pest types may be pest species that cause damage to the crop. For example, for wheat, the pest types may include soil insects, wheat aphids, red spiders, sucking worms, armyworms, and the like. For example, for rice, the pest types may include plant hoppers, rice leaf rollers, chilo suppressalis, tryporyza incertulas, thrips oryzae, and the like. For another example, for cotton, the pest types may include cotton bollworm, cotton aphid, cotton stink bug, prodenia litura, beet armyworm, and the like. For another example, for vegetables, the pest types may include bemisia tabaci, vegetable diamond back moth, vegetable beet armyworm, vegetable thrips, cabbage caterpillar, vegetable aphid, and the like.

In some embodiments, pest conditions may be determined based on farm information, such as crop information, environmental information, remote farm sensing images, fertigation information, historical pest killing information, and the like. In some embodiments, pest conditions can be determined through crop information and environmental information. In some embodiments, pest conditions may be determined by crop species, foliage conditions, growth conditions, and the like. For example, when the crop is rice, when the collected rice leaves include the conditions of white spots, withered and yellow leaves, twisted leaves, inextensible leaf sheaths and the like, the insect pest type can be determined to be thrips oryzae. For example, when the collected vegetable leaves are white, the leaves are not full of a silver white film, the leaf vein stems are white and semitransparent, and plants are withered when the crops are vegetables, the insect pest type can be determined to be bemisia tabaci.

In some embodiments, a corresponding plurality of pest types may be determined by a pest determination model based on farm information.

In some embodiments, the pest determination model may be a machine learning model. In some embodiments, the machine learning model may include, but is not limited to, a neural network model (e.g., CNN model, DNN model, RNN model), a support vector machine model, a lambdarak model, a GBDT + LR model, and the like.

In some embodiments, farm information, such as crop information, environmental information, may be input into a pest determination model, which may determine a corresponding pest type. In some embodiments, farm information, such as crop information, environmental information, etc., is input into a pest determination model, which may determine a corresponding pest type or types. For example, the pest determination model may determine the pest types of the farm by outputting probabilities that a plurality of pest types, such as bemisia tabaci, vegetable diamond back moth, vegetable beet armyworm, vegetable thrips, cabbage caterpillars, and vegetable aphids, respectively correspond to each other, and determining the pest types of the farm based on the probabilities, such as determining the pest types corresponding to the probabilities greater than a threshold as the pest types corresponding to the farm, or determining the pest types corresponding to the probabilities ranked to the TopN as the pest types corresponding to the farm.

In some embodiments, an initial pest determination model may be trained based on a number of training samples with identifications to iteratively update parameters of the initial pest determination model to arrive at a pest determination model. In some embodiments, training samples with the identification are input into the initial pest determination model, and parameters of the initial pest determination model are updated through training iteration. In some embodiments, the training samples may be farm information samples such as crop information samples and environmental information samples, and the sample identifiers may include crop information and one or more types of insect pests corresponding to the environmental information.

In some embodiments, the model parameters may be iteratively updated by training through various methods based on the training samples. For example, the training may be based on a gradient descent method.

In some embodiments, the corresponding multiple pest types in the field can be determined according to historical pest killing information and historical pesticide application parameters. For example, the rice planthopper is checked and killed 2 times at intervals of 7-10 days, according to historical pest checking and killing information, the first time of checking and killing of the rice planthopper is carried out 3 days before the current date, and then the current pest type at least comprises the rice planthopper. In some embodiments, pest type may also be determined using pest activity monitored by the monitoring device. For example, the pest type is determined according to the monitoring of the wing flapping frequency by the monitoring device.

In some embodiments, the corresponding multiple pest types may also be determined based on farm information according to other methods, which are not limited herein.

Step 130, determining a plurality of searching and killing schemes corresponding to a plurality of insect pest types, and acquiring at least one of crop influence characteristics, soil influence characteristics and pesticide effect characteristics corresponding to the plurality of searching and killing schemes.

The checking and killing scheme is a control means aiming at different insect pest types. In some embodiments, the killing regimen may include agricultural control, physical control, chemical control, and the like, or any combination thereof.

In some embodiments, agricultural control includes keeping the field clean, thoroughly cleaning the field of straw, dead branches, fallen leaves, weeds, clutter, etc. in time after harvesting of previous crops or after picking of fruits or pruning. In some embodiments, physical control includes the use of insect nets, insect killing lights, and the like. For example, a 24-30 mesh insect net can be used to prevent the invasion of pests such as diamondback moth, cabbage caterpillar, prodenia litura, beet moth, aphid, leaf miner, etc. In some embodiments, chemical control comprises controlling pests using an agent.

In some embodiments, the invasion of pests can be prevented in advance by agricultural control and physical control. For example, keeping the field clean, thoroughly clearing dead branches, fallen leaves, weeds, mess, etc. after harvesting, installing insect nets during crop growth, insect killing lamps to control pest invasion.

In some embodiments, after the crop is damaged by the pest, different chemical control means can be adopted according to different pest types, and the chemical control means can comprise the medication type, the formula and the dosage of various medicaments, the combined formula and the proportion of various medicaments and the like. For example, on aphids that are damaging to wheat, carbosulfan can be applied at a field aphid prime stage (e.g., on the order of 500 aphids per hundred plants) at an application rate of 6-8 grams per acre (e.g., 30-40 ml/acre of 20% carbosulfan emulsifiable concentrate). For example, for the rice leaf roller which is harmful to rice, 2% abamectin can be applied from the peak period of hatching of eggs of the rice leaf roller to the peak period of 1 and 2 instar larvae, the application amount is 100 ml, 15-30 kg of water is added, and the abamectin is sprayed once more at intervals of 15 days.

In some embodiments, the multiple searching and killing schemes corresponding to the multiple insect pest types can be determined by searching a table, inputting the insect pest types into a trained machine learning model to obtain the corresponding searching and killing schemes, and the like.

The crop influence characteristics refer to relevant data of the influence of the killing scheme on the growth of crops. In some embodiments, the crop impact characteristic may include an impact of a killing protocol on leaf feeding, leaf shrinkage, leaf chlorosis, leaf mildew, leaf drop, rhizome distortion, rhizome rot, crop dysplasia, or the like, or any combination thereof.

The soil influence characteristics refer to relevant data of the influence of the killing scheme on the soil. In some embodiments, the soil-affecting characteristic may include an effect of a hunting regime on the porosity, pH, fertility, salt concentration, etc., of the soil, or any combination thereof.

Pharmacodynamic characteristics are characteristics of the agents used in the killing protocol. In some embodiments, the pharmacodynamic characteristics may include formulation and amount of each agent, combined formulation and ratio of each agent corresponding to the type of pesticide to be killed, time to kill the pest, and the like.

In some embodiments, at least one of a crop impact feature, a soil impact feature, and a pharmacodynamic feature may be extracted by the feature extraction module 430. In some embodiments, the feature extraction module 430 includes a neural network unit (hereinafter simply referred to as a feature extraction unit) for feature extraction.

In some embodiments, the feature extraction unit may include, but is not limited to, CNN, DNN, RNN, or any combination thereof.

In some embodiments, the input of the feature extraction unit may be a killing scheme such as a formula and an amount of each agent, a combined formula and a proportion of each agent, and the output may be at least one of a crop impact feature, a soil impact feature, and a pesticide effect feature. In some embodiments, the feature extraction unit may be trained from training samples.

In some embodiments, the initial feature extraction unit may be trained based on a number of training samples with identifications to iteratively update parameters of the initial feature extraction unit to obtain the feature extraction unit. Specifically, a training sample with a mark is input into the initial feature extraction unit, and parameters of the initial feature extraction unit are updated through training iteration. In some embodiments, the training samples may include killing scheme samples, and the training identifier may include at least one of crop influence characteristics, soil influence characteristics, and pesticide effect characteristics corresponding to the killing scheme samples.

In some embodiments, the feature extraction unit may be obtained by training through various methods based on training samples. For example, the training may be based on a gradient descent method.

Step 140, determining a plurality of combination schemes based on a plurality of combination modes of the plurality of searching and killing schemes.

The combined scheme refers to a scheme obtained by combining a plurality of searching and killing schemes. In some embodiments, different insect pest types can correspond to different searching and killing schemes, one insect pest type can also correspond to multiple different searching and killing schemes, and different searching and killing schemes can be combined, so that the combined scheme is used when insect pest searching and killing is carried out on the farm, and searching and killing of multiple insect pest types are realized. For example, determining the types of insect pests on the current farm to comprise wheat spiders, wherein the wheat spiders are applied by applying 40% dimethoate missible oil in the field, wherein the application amount of the dimethoate missible oil is 50-60 ml/mu, and the dimethoate missible oil is uniformly sprayed by adding 50 kilograms of water; the current farm insect pest type also comprises armyworm, and the armyworm searching and killing scheme is that 77.5 percent of dichlorvos missible oil is applied to the armyworm 3-4 instar larvae in the field in the peak period, the application dosage is 50 ml/mu, and 50 kilograms of water is added for uniform spraying; the current farm insect pest types also comprise rice planthoppers, and the rice planthopper searching and killing scheme is that 25% of fly-catching SC is applied in the field, the application dosage is 20 g/mu, and 40-50 kg of water is added for uniform spraying; the current farm insect pest types also comprise rice thrips, and the rice thrips searching and killing scheme is that 10% of cleaning agent is applied in the field, the application dosage is 20-30 g/mu, and 30 kg of water is added for even spraying; in order to simultaneously check and kill various insect pests on the farm, various checking and killing schemes corresponding to various insect pest types can be combined, and for example, the combination scheme can be obtained as (a checking and killing scheme of wheat spiders + a checking and killing scheme of armyworms), (a checking and killing scheme of armyworms + a checking and killing scheme of rice planthoppers), (a checking and killing scheme of wheat spiders + a checking and killing scheme of rice planthoppers + a checking and killing scheme of rice thrips), and the like.

In some embodiments, multiple combination schemes may be determined based on multiple combinations of multiple killing schemes.

In some embodiments, first, various combinations of various killing schemes may be determined. For example, the red spider control scheme includes 6 kinds (a1-a6) shown in table 1, and for example, the grub control scheme includes 4 kinds (B1-B4) shown in tables 2-3.

TABLE 1

Red spider control scheme	Prescription for medication	Dilution factor	(Mode)
				A1	1% pesticide (insecticide)	2500-3000	Spray mist
A2	73% propargite emulsifiable concentrate	2000-3000	Spray mist
				A3	5% Nixolang emulsifiable concentrate	1500-2500	Spray mist
A4	20% acarus emulsifiable concentrate	1000-2000	Spray mist
				A5	10% liuyangmycin emulsifiable concentrate	2000-3000	Spray mist
A6	7.5% concentration pesticide-miticide emulsion	750-1000	Spray mist

TABLE 2

Grub control scheme	Prescription for medication	Dilution factor	(Mode)
				B1	90% crystal trichlorfon	600	Root of irrigated land
B2	40% phoxim	500	Root of irrigated land

TABLE 3

If red spider insect pests and white grub insect pests appear in a certain vegetable garden, 24 combinations of A1B1, A2B1, A3B1, A4B1, A5B1, A6B1, A1B2, A2B2, A3B2, A4B2, A5B2, A6B2, A1B3, A2B3, A3B3, A4B3, A5B3, A6B3, A1B4, A2B4, A3B4, A4B4, A5B4 and A6B4 can be obtained according to the permutation and combination.

Secondly, preset conditions are determined based on the farm information, and a plurality of combinations meeting the preset conditions in the plurality of combinations are used as a plurality of combination schemes. The preset condition is a preset condition for screening the combination recipe. In some embodiments, the predetermined condition may include a ratio of the leaf area index of the crop to the amount applied in the combined screening and killing protocol being greater than a threshold (in some embodiments, the amount applied is greater, the leaf area index is less, and the crop is more damaged), the amount applied is not excessive, the leaf area index is greater than a threshold, the crop is less damaged, and the like, or any combination thereof. In some embodiments, the preset condition may be determined based on farm information. For example, based on the size of the leaf area index in the farm, a smaller threshold value may correspond to the ratio of the leaf area index to the application amount of the hunting protocol combination if the leaf area index is greater than the threshold value, and a larger threshold value may correspond to the ratio of the leaf area index to the application amount of the hunting protocol combination if the leaf area index is less than the threshold value. Wherein the blade area index is the ratio of the average blade surface area to the unit area.

In some embodiments, a plurality of combinations satisfying a preset condition may be used as a plurality of combination schemes to be selected. For example, it is calculated that the ratio of the leaf area index to the application amount of each of 6 combinations (e.g., A1B1, … … A6B1) including B1 is not greater than the threshold value; and the ratios of the rest 18 combination modes (A1B2, … … A6B2, A1B3, … … A6B3, A1B4 and … … A6B4) are all larger than a threshold value and meet a preset condition, so that the 18 combination modes are used as a plurality of combination schemes to be selected.

By screening and combining the checking and killing schemes under preset conditions, the scheme with large side effects (such as damaged crops, soil pollution and the like) can be eliminated, the calculated amount is reduced, and the efficiency is improved.

In some embodiments, the combination scheme may be determined in other ways, for example, by eliminating a frequently used or recently used hunting scheme in consideration of drug resistance and pesticide residue, and the present specification is not limited thereto.

In some embodiments, the combination of the killing schemes with the best effect (for example, the combination of the killing schemes with the best effect is good for a plurality of insect pests on the farm, and meanwhile, the negative effects on the crop growth such as growth inhibition, leaf surface burn and the like are small, and the negative effects on the soil such as increase of soil alkalinity, soil agglomeration and the like are small) can be selected, and the problem that the killing effect is influenced between different killing medicines and the waste of resources caused by common use of all the killing schemes for insect pest killing can be avoided by selecting the combination scheme with the best effect.

And 150, obtaining corresponding fusion characteristics based on the at least one characteristic included in each combination scheme.

The fusion characteristics are obtained by fusing at least one characteristic (such as at least one characteristic of crop influence characteristics, soil influence characteristics and pesticide effect characteristics) corresponding to each searching and killing scheme in the combined scheme. The fusion features both preserve the combination scheme detail features and can provide the global features of the combination scheme.

In some embodiments, the fused feature may be obtained by a feature fusion unit. Specifically, the feature fusion unit may be a neural network unit, and the feature combinations corresponding to the combined searching and killing scheme are input to the feature fusion unit, so that the fused features can be obtained. For example, the feature fusion unit may be a Deep Neural Network (DNN). For another example, the feature fusion unit can be a two-layer convolutional neural network unit, and the structure can better fuse multiple groups of fertilizer requirement features, so that under-fitting caused by a single-layer neural network unit and over-fitting caused by multiple layers of neural network units are avoided. In some embodiments, the corresponding fusion features may be obtained by means of stitching, superposition, weighted fusion, and the like.

And 160, obtaining at least one of crop growth potential influence scores and soil quality condition influence scores corresponding to the multiple combination schemes through a prediction model based on the fusion characteristics and the farm information.

A predictive model is a model used to predict the effect and/or influence of a combinatorial approach. In some embodiments, the input of the prediction model is fusion characteristics and farm information corresponding to the combination scheme, and the output of the prediction model is at least one of a crop growth potential influence score and a soil health condition influence score corresponding to the combination scheme.

In some embodiments, the predictive model may be a machine learning model. In some embodiments, the machine learning model may include, but is not limited to, a neural network model (e.g., CNN model, DNN model, RNN model), a support vector machine model, a lambdarak model, a GBDT + LR model, and the like.

The crop growth impact score is an amount by which the crop growth is assessed for the impact of the hunting program. Crop growth can include increased plant height, increased leaf area, increased fruit diameter, and the like. In some embodiments, if the killing scheme has a positive effect on crop growth, e.g., a steady increase in plant height or above a threshold, the crop growth effect score is positive. On the contrary, if the checking and killing scheme has negative influence on the growth vigor of the crops, the crop growth influence score is negative. The larger the positive influence of the checking and killing scheme on the growth vigor of the crops is, the higher the score of the growth vigor influence of the crops is; the greater the negative impact (or the smaller the positive impact) of the check-kill scheme on the growth of the crop, the lower the crop growth impact score.

The soil goodness impact score is the amount of impact of the evaluation of the test and kill protocol on soil quality. The soil condition may include fertility, contamination, biological activity, etc. of the soil. In some embodiments, if the kill protocol has a positive effect on soil goodness, e.g., reduced pesticide residue, the soil goodness effect score is positive. On the contrary, if the checking and killing scheme has negative influence on the soil good condition, the soil good condition influence score is a negative value. The larger the positive influence of the checking and killing scheme on the soil excellent condition is, the higher the influence score of the soil excellent condition is; the greater the negative impact (or the lesser the positive impact) of the survey program on soil goodness, the lower the soil goodness impact score.

In some embodiments, the predictive score for implementing the combined survey plan on the farm can be obtained by a predictive model. Specifically, the fusion characteristics and the farm information of the combined searching and killing scheme are input into a prediction model, and the prediction model outputs at least one of a crop growth potential influence score and a soil excellent condition influence score corresponding to the combined searching and killing scheme.

In some embodiments, the initial prediction model may be trained based on a number of training samples with identifications to iteratively update parameters of the initial prediction model to arrive at the prediction model. In some embodiments, the training samples may include combination scheme samples corresponding to the fusion features and farm information samples, and the sample identification may include at least one of crop growth impact scores and soil health impact scores corresponding to the combination scheme samples and the farm information. Specifically, a training sample with a mark is input into an initial prediction model, and parameters of the initial prediction model are updated through training iteration. In some embodiments, the parameters of the initial predictive model may be iteratively updated by training through various methods based on the training samples. For example, the training may be based on a gradient descent method. In some embodiments, a loss function may be set to assist in training of the initial prediction model until the loss function or prediction model converges, wherein the loss function may be related to a crop impact characteristic, a soil impact characteristic, a pharmacodynamic characteristic.

Loss functions related to crop influence characteristics, soil influence characteristics and pesticide effect characteristics are adopted in the training of the prediction model, and the accuracy of the prediction model is improved.

Step 170, determining a target combination scheme according to the at least one score.

In some embodiments, at least one score for each of the plurality of candidate combination schemes may be predicted separately, and a target combination scheme may be selected based on the scores. For example, the combination scheme with the highest total score or the highest weighted total score is selected as the target combination scheme. For another example, the combination scheme with the highest average score is selected as the target combination scheme. In some embodiments, when there are a plurality of combination schemes with highest total score in parallel, one or more of the plurality of combination schemes may be randomly selected as the target combination scheme.

The following description will be given of the process of determining a target combination plan, taking a certain wheat-growing farm as an example. Based on the farm information of the farm, 2 insect pests of armyworm and wheat spider are obtained through the insect pest determination model. Table look-up shows that the mythimna separata control schemes include 3 (C1-C3) shown in table 4, and the wheat spider control schemes include 4 (D1-D4) shown in table 5, thereby obtaining various combinations of the hunting schemes (e.g., C1D1, … … C3D 4).

TABLE 4

Armyworm control scheme	Prescription for medication	(Mode)	Timing of
				C1	77.5% dichlorvos emulsifiable concentrate	Spray mist	Peak period of 3-4 instar larva
C2	77.5% dichlorvos emulsifiable concentrate	Spreading and applying	In the evening
				C3	90% trichlorfon crystal	Spray mist	Peak 3 instar larva

TABLE 5

The crop influence characteristics, the soil influence characteristics and the pesticide effect characteristics of each combination mode (for example, C1D1, … … C3D4) are extracted through a characteristic extraction unit, and then the fusion characteristics of each combination mode are obtained. The fusion characteristics of each combination mode (for example, C1D1, … … C3D4) and the farm information of the wheat farm are input into a prediction model, and a crop growth condition influence score and a soil health condition influence score of each combination mode are obtained. The combination with the highest total score (e.g., C1D3) was taken as the target combination scheme.

It should be noted that the above description of the process 100 is for illustration and description only, and does not limit the scope of the application of the present disclosure. Various modifications and alterations to process 100 will become apparent to those skilled in the art in light of the present description. However, such modifications and variations are intended to be within the scope of the present description.

FIG. 2 is an exemplary flow diagram of a method of adjusting model parameters of a predictive model according to some embodiments described herein.

And step 210, obtaining farm information change values before and after the objective combination scheme is adopted for checking and killing based on the multiple groups of crop data and the multiple groups of environmental data of the multiple sampling points.

The farm information change value is a change value or a change rate of the farm information (for example, the growth vigor of crops, the soil condition, the environmental humidity, the environmental PH value, and the like) after and before the killing. For example, the average height variation value of crops in the farm before and after killing, the variation value of the integrity of leaves of the crops, the variation value of pH value, the variation value of humidity and the like are checked.

The farm information change value may be a positive value, 0, or a negative value. Positive values indicate good direction of change, effective killing, controllable side effects, etc. A0 or negative value indicates poor direction of change, invalid check and kill, large side effect and the like. For example, in the seedling stage and the growing stage of the crop, the change value of the average height of the crop after the check and kill and the average height before the check and kill exceeds a threshold value, which represents that the crop grows well, and the change value of the crop height can be a positive value; on the contrary, the average height change of the crops before and after killing does not exceed the threshold value or the change is within the error range, which represents that the crops grow badly, and the height change value of the crops can be 0. For another example, the change value of the integrity of the leaves of the crops in the farm field before and after the crop is killed is checked, the integrity of the leaves of the crops after the crop is killed is increased, has no change or changes within an error range, the pest killing effect is good, and the change value of the integrity of the leaves can be a positive value; on the contrary, the completeness of the leaves is reduced after checking and killing, the pest killing effect is not good, and the change value of the completeness of the leaves can be a negative value. For another example, the PH value of the environment before and after killing is checked, and for another example, the humidity value of the environment before and after killing is checked. After the investigation and killing, the environment pH value and the humidity may be changed, which may cause that the pH value and the humidity after the investigation and killing are not the optimal pH value and the humidity for the growth of the crops, when the pH value and the humidity in the farm after the investigation and killing are in the optimal range, the change value of the environment pH value and the humidity may be 0, and when the pH value and the humidity in the farm after the investigation and killing are out of the optimal range, the change value of the environment pH value and the humidity may be negative.

In some embodiments, the data may be analyzed directly or indirectly based on the collected data to obtain farm information change values.

In some embodiments, the farm information variation value may be obtained by comparing farm information before and after the killing acquired at a plurality of sampling points in the farm. For example, the farm information change value is obtained by comparing the crop image and the soil information before and after the crop is killed.

In some embodiments, the farm information change value may be obtained by other means, for example, by an unmanned aerial vehicle. Specifically, the unmanned aerial vehicle acquires farm images before and after searching and killing respectively, and through carrying out image recognition on the farm images before and after searching and killing, the complete blades and the incomplete blades of farm crops before and after searching and killing are counted, and according to a counting result, a crop blade integrity change value is acquired.

Also for example, by sensors, detection instruments, and the like. Specifically, an infrared sensor is arranged near the crops to obtain the height change value of the crops; acquiring an environment humidity change value through a humidity sensor; and acquiring the change value of the environmental pH value through a pH automatic detector.

For more on-farm information, see fig. 3 and its associated description.

Step 220, obtaining an evaluation result of the target combination scheme through an evaluation model based on the farm information change value;

the evaluation model is a model for evaluating the actual effect of the killing. The actual effects of checking and killing include: insecticidal effect, crop growth effect, environmental effect, etc. In some embodiments, the evaluation model may comprise a linear regression model. In some embodiments, the evaluation model may include other models, for example, a machine learning model.

In some embodiments, the input to the evaluation model may be the farm information variation values before and after field searching with the target combination scheme. In some embodiments, the output of the evaluation model may be an evaluation score of the target combination scheme, for example, the evaluation score may be a numerical value within the [ -10,10] interval. The killing effect is achieved if the evaluation score is positive, and the higher the evaluation score is, the better the actual killing effect is. If the evaluation score is 0 or negative, the killing is ineffective or even worse than before the killing.

In some embodiments, the initial evaluation model may be trained based on a large number of training samples with identifications to iteratively update parameters of the initial evaluation model to obtain an evaluation model, so as to realize evaluation of all killing schemes. In some embodiments, the initial evaluation model may be trained based on farm information change value samples corresponding to a plurality of killing schemes or combination schemes, and the sample identification may be an evaluation score corresponding to the farm information change value sample. In some embodiments, an initial evaluation model may be trained based on farm information change value samples corresponding to a certain searching and killing scheme or combination scheme, and the sample identifier may be an evaluation score corresponding to the farm information change value samples, so as to implement evaluation of the searching and killing scheme or combination scheme, that is, for multiple searching and killing schemes or combination schemes, corresponding multiple evaluation models may be trained, and each evaluation model corresponds to one searching and killing scheme or combination scheme.

And 230, adjusting model parameters of the prediction model based on the evaluation result.

In some embodiments, model parameters of the predictive model may be adjusted based on the evaluation.

In some embodiments, the model parameters of the predictive model may include a plurality of scoring weights for a plurality of combination schemes. In some embodiments, the scoring weight corresponding to a combination scheme (e.g., a target combination scheme) may be adjusted based on the evaluation score of the combination scheme.

The scoring weights are parameters used in the predictive model to characterize the importance of the combinatorial approach. In some embodiments, the output of the prediction model is the product of the scoring weight and the prediction score, for example, the scoring weight a corresponding to a certain combination scheme and the prediction score of the combination scheme is S, and the combination scheme (scheme information/parameters such as a drug formula) is input into the prediction model, and the obtained output is a × S. In some embodiments, the scoring weight may default to 1, and may be adjusted to the equivalent of 0.8, 1.1.

In some embodiments, the scoring weight corresponding to a combination scheme may be adjusted based on the evaluation score of the combination scheme. For example, based on the average value of the multiple historical evaluation scores, the corresponding scoring weight of the target combination scheme is adjusted. Specifically, if the average value is low, the scoring weight corresponding to the combination scheme is reduced, and if the average value is high, the scoring weight corresponding to the combination scheme is increased.

In some embodiments, model parameters of the predictive model may also be adjusted based on the evaluation results by other methods. For example, for a combination scenario in which the evaluation score is negative, the prediction model outputs a negative value directly for the combination scenario.

Based on the evaluation score reflecting the actual searching and killing effect, the weight of the target combination scheme is adjusted, the prediction score is corrected, the searching and killing scheme score of the actual effect can be higher, and the searching and killing scheme with the better actual effect can be obtained.

It should be noted that the above description related to the flow 200 is only for illustration and description, and does not limit the applicable scope of the present specification. Various modifications and alterations to flow 200 will be apparent to those skilled in the art in light of this description. However, such modifications and variations are intended to be within the scope of the present description.

Fig. 3 is a schematic view of an application scenario of a farm pest killing system according to some embodiments of the present disclosure.

As shown in fig. 3, an application scenario 300 of the farm pest screening and killing system (hereinafter referred to as "application scenario 300") may include a server 310, a processor 312, a farm 320, a monitoring device 330, a storage device 340, a network 350, and a drone 360.

In some embodiments, the server 310 may be used to process information and/or data related to the application scenario 300, for example, determine a kill scheme, evaluate the effectiveness of a kill. In some embodiments, server 310 may be a single server or a group of servers. The set of servers can be centralized or distributed (e.g., server 310 can be a distributed system). In some embodiments, server 310 may be local or remote. For example, the server 310 may access information and/or data stored in the monitoring apparatus 330, the drone 360, and/or the storage device 340 via the network 350. As another example, the server 310 may be directly connected to the monitoring apparatus 330, the drone 360, and/or the storage device 340 to access stored information and/or data. In some embodiments, the server 310 may be implemented on a cloud platform or provided in a virtual manner. By way of example only, the cloud platform may include a private cloud, a public cloud, a hybrid cloud, a community cloud, a distributed cloud, an internal cloud, a multi-tiered cloud, and the like, or any combination thereof.

In some embodiments, server 310 may include a processor 312. The processor 312 may process information and/or data related to the application scenario 300 to perform one or more functions described herein. For example, processor 312 may obtain farm information and determine a plurality of pest types based on the farm information. For another example, the processor 312 may determine a plurality of searching and killing schemes corresponding to a plurality of pest types, and obtain crop influence characteristics, soil influence characteristics, and pesticide effect characteristics corresponding to the plurality of searching and killing schemes. In some embodiments, processor 312 may include one or more processing engines (e.g., a single chip processing engine or a multi-chip processing engine). Merely by way of example, the processor 312 may include a Central Processing Unit (CPU), an Application Specific Integrated Circuit (ASIC), an application specific instruction set processor (ASIP), a Graphics Processing Unit (GPU), a Physical Processing Unit (PPU), a Digital Signal Processor (DSP), a Field Programmable Gate Array (FPGA), a Programmable Logic Device (PLD), a controller, a microcontroller unit, a Reduced Instruction Set Computer (RISC), a microprocessor, or the like, or any combination thereof.

Farm 320 refers to a field where crops are planted. Such as rice fields, wheat fields, vegetable greenhouses, orchards, tea gardens, tobacco plantations, and the like. In some embodiments, farm 320 may include a natural habitat of wild plants. In some embodiments, the farm 320 may comprise a multi-purpose farm, such as a crab paddy field, a picking garden, or the like.

The sampling point refers to the point where the sample was taken. In some embodiments, multiple sampling points may be provided in at least one area of the farm. For example, the greenhouse is randomly arranged, one is arranged per mu, and two ends of each greenhouse are respectively arranged. In some embodiments, a plurality of sampling points may be provided in at least one area of the farm according to a preset rule. For example, the sampling points are set by a diagonal point arrangement method, a quincunx point arrangement method, a checkerboard point arrangement method, a snake point arrangement method, or the like.

In some embodiments, one or more monitoring devices 330 may be provided at each sampling point.

Monitoring device 330 refers to a device that monitors a farm and/or a crop in the farm. In some embodiments, the monitoring device 330 may include sensors, cameras, positioning components, storage components, and the like. Wherein the positioning component can position the monitoring device 330, in some embodiments, the monitoring device 330 can be installed in a farm-fixed location. For example, the southwest corner of the farm, 2 nd plant, No. 2. In some embodiments, the monitoring device 330 may be mounted on livestock and/or machinery operating in a farm, for example, on the ears of a farm cattle, on an agricultural tractor. In some embodiments, the monitoring device 330 may be located at a sampling point.

In some embodiments, the placement density of the monitoring devices 230 (the placement density of the area where the sampling site is located) may be determined based on farm information and pest conditions. For example, the placement density is high when the types of pests are many; the placement density is high when the crops in the farm are in the critical growth periods such as the seedling period, the flowering and fruit setting period and the like. In some embodiments, the packing density a may be determined by the following equation: and a is (1+ insect pest type number is 0.5+ b is 0.5) m, wherein the growth period is a key period b is 1, otherwise b is 0, and m is a conventional mounting density empirical value.

In some embodiments, each monitoring device has a unique identity, and a record corresponding to the unique identity and the position of the sampling point is stored on the monitoring device, and the record may be in the form of a two-dimensional code, a barcode, an NFC tag, or the like. In some embodiments, the recorded content includes: the unique identification of the monitoring device, the sampling point position, the sampling time, the sampling point environmental data, the sampling point crop data and the like.

In some embodiments, the monitoring device 330 may monitor environmental data of the sampling point. Such as air temperature, air humidity, air pressure, ultraviolet intensity, light, total solar radiation, wind speed, wind direction, wind power, rainfall, and/or weather data. For example, soil data such as soil temperature, soil moisture, soil salinity (conductivity), and soil PH. In some embodiments, the monitoring device 330 may monitor the crop data at the sampling point. Such as crop images, crop nutrient data, canopy coverage data, and the like.

In some embodiments, corresponding sets of crop data and sets of environmental data may be obtained at a plurality of sampling points via the monitoring device. In some embodiments, the sets of crop data and the sets of environmental data for the plurality of sampling points may be collected by the drone, for example, by the drone scanning records (e.g., two-dimensional codes, etc.) saved in the monitoring device at the plurality of sampling points. In some embodiments, the farm information may be obtained by performing statistical analysis on the sets of crop data and the sets of environmental data based on weights corresponding to the sampling points, where the weights are related to the density of the sampling points in the area where the sampling points are located. For example, the greater the density of sample points, the less the weight.

In some embodiments, staging farm information for at least 1 time stage may be obtained by the monitoring device over a period of time during which pest infestation with the target combination program is performed. In some embodiments, at least one of a staged crop growth impact score, a staged soil goodness impact score for a target combination scenario may be determined by a predictive model based on the fused features of the target combination scenario and the staged farm information. In some embodiments, a density of sampling points for the at least one region at a next time phase and/or an adjustment to a sampling trajectory for the at least one region may be determined based on the at least one phasic score for the current phase. For example, if the crop growth influence score of the current stage is low, a key monitoring area (an area with poor crop growth condition) is determined, and the density of sampling points in the area is increased. For another example, if the crop growth influence score at the current stage is low, a key monitoring area (an area with poor crop growth condition) is determined, and other sampling points (not included in the initial sampling trajectory) in the area are added into the sampling trajectory.

The pest searching and killing effect is evaluated in stages, the problems in the searching and killing period can be found in time, the deviation can be corrected in time by mainly observing the problematic area, the searching and killing efficiency is improved, and the searching and killing effect is ensured.

The storage device 340 may be used to store data and/or instructions related to the application scenario 300. In some embodiments, the storage device 340 may store data obtained/obtained from the monitoring apparatus 330 and/or the drone 360. In some embodiments, storage device 340 may store data and/or instructions that server 310 uses to perform or use to perform the exemplary methods described in this application. In some embodiments, storage device 340 may include mass storage, removable storage, volatile read-write memory, read-only memory (ROM), and the like, or any combination thereof. Exemplary mass storage devices may include magnetic disks, optical disks, solid state disks, and the like. Exemplary removable memory may include flash drives, floppy disks, optical disks, memory cards, compact disks, magnetic tape, and the like. Exemplary volatile read and write memories can include Random Access Memory (RAM). Exemplary RAM may include Dynamic Random Access Memory (DRAM), Double Data Rate Synchronous Dynamic Random Access Memory (DDRSDRAM), Static Random Access Memory (SRAM), thyristor random access memory (T-RAM), zero capacitance random access memory (Z-RAM), and the like. Exemplary read-only memories may include model read-only memory (MROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM), digital versatile disc read-only memory, and the like. In some embodiments, storage device 340 may be implemented on a cloud platform. By way of example only, the cloud platform may include a private cloud, a public cloud, a hybrid cloud, a community cloud, a distributed cloud, an internal cloud, a multi-tiered cloud, and the like, or any combination thereof. In some embodiments, the storage device 340 may be connected to the network 350 to communicate with one or more components of the application scenario 300 (e.g., the server 310, the monitoring apparatus 330, the drone 360). One or more components of the application scenario 300 may access data or instructions stored in the storage device 340 via the network 350. In some embodiments, the storage device 340 may be directly connected to or in communication with one or more components of the application scenario 300 (e.g., the server 310, the monitoring apparatus 330, the drone 360). In some embodiments, storage device 340 may be part of server 310. In some embodiments, storage device 340 may be a separate memory.

The network 350 may facilitate the exchange of information and/or data. In some embodiments, one or more components of the application scenario 300 (e.g., the server 310, the monitoring apparatus 330, the storage 340, the drone 360) may send information and/or data to other components of the application scenario 300 via the network 350. For example, the drone 360 may send farm information to the server 310 via the network 350. In some embodiments, the network 350 may be a wired network or a wireless network, or the like, or any combination thereof. By way of example only, network 350 may include a cable network, a wireline network, a fiber optic network, a telecommunications network, an intranet, the Internet, a Local Area Network (LAN), a Wide Area Network (WAN), a Wireless Local Area Network (WLAN), a Metropolitan Area Network (MAN), a Public Switched Telephone Network (PSTN), a Bluetooth network, a ZigBee network, a Near Field Communication (NFC) network, a global system for mobile communications (GSM) network, a Code Division Multiple Access (CDMA) network, a Time Division Multiple Access (TDMA) network, a General Packet Radio Service (GPRS) network, an enhanced data rates for GSM evolution (EDGE) network, a Wideband Code Division Multiple Access (WCDMA) network, a High Speed Downlink Packet Access (HSDPA) network, a Long Term Evolution (LTE) network, a User Datagram Protocol (UDP) network, a transmission control protocol/Internet protocol (TCP/IP) network, a Short Message Service (SMS) network, a Wireless Application Protocol (WAP) network, Ultra-wideband (UWB) networks, infrared, and the like, or any combination thereof. In some embodiments, the application scenario 300 may include one or more network access points. Such as base stations and/or wireless access points 350-1, 350-2, …, one or more components of the application scenario 300 may connect to the network 350 to exchange data and/or information.

Drone 360 refers to an unmanned aircraft, either an unmanned aircraft operated with a radio remote control device and self-contained program control, or autonomously operated by an onboard computer, either completely or intermittently. In some embodiments, the drone may include a scanning component, a positioning component, and the like. In some embodiments, flight parameters of the drone may be set, such as altitude, speed, perspective, trajectory, and so forth. In some embodiments, the mode of flight of the drone may be set, e.g., point-to-point, point-around, hover, multi-point, follow, etc. For example, the pesticide spraying situation is acquired in real time by following an airplane spraying pesticide at a preset distance. For example, the operation condition in the farm can be obtained in real time by following irrigation equipment, a tractor and the like.

In some embodiments, the drone 360 may scan the records in the monitoring device 330 to obtain data for the sampling points. In some embodiments, the drone 360 may scan all or a portion of the sampling points. Utilize unmanned aerial vehicle to acquire farm information, practiced thrift a large amount of manpowers to it is more reliable effective.

It should be noted that the application scenario 300 is provided for illustrative purposes only and is not intended to limit the scope of the present application. It will be apparent to those skilled in the art that various modifications and variations can be made in light of the description of the present application. For example, the application scenario 300 may also include a database. As another example, the application scenario 300 may implement similar or different functionality on other devices. However, such changes and modifications do not depart from the scope of the present application.

In some embodiments, farm pest killing system 400 may include an acquisition module 410, a determination module 420, a feature extraction module 430, a combination module 440, a feature fusion module 450, a scoring module 460, and a determination module 470.

The acquisition module can be used for acquiring farm information, and the farm information comprises crop information and environment information.

In some embodiments, the obtaining module is to: setting a plurality of sampling points in at least one area of the farm; obtaining multiple groups of corresponding crop data and multiple groups of corresponding environmental data at multiple sampling points through a monitoring device; the method comprises the steps that multiple groups of crop data and multiple groups of environment data of multiple sampling points are collected through the unmanned aerial vehicle, statistical analysis is conducted on the multiple groups of crop data and the multiple groups of environment data based on the corresponding weights of the sampling points, farm information is obtained, and the weights are related to the density of the sampling points in the area where the sampling points are located.

In some embodiments, the obtaining module is to: acquiring stage farm information of at least 1 time stage through a monitoring device in a period of time for pest searching and killing by adopting a target combination scheme; determining at least one stage score in stage crop growth potential influence scores and stage soil excellent condition influence scores of a target combination scheme through a prediction model based on the fusion characteristics and the stage farm information; based on the at least one stage score of the current time stage, determining a density of sampling points of the at least one region of the next time stage and/or adjusting a sampling trajectory of the at least one region.

In some embodiments, the obtaining module is to: obtaining farm information change values before and after the checking and killing by adopting a target combination scheme based on a plurality of groups of crop data and a plurality of groups of environmental data of a plurality of sampling points; obtaining an evaluation result of the target combination scheme through an evaluation model based on the farm information change value; based on the evaluation result, model parameters of the prediction model are adjusted.

In some embodiments, the model parameters of the prediction model include a plurality of scoring weights corresponding to a plurality of combination schemes, and the obtaining module is configured to: and adjusting the scoring weight corresponding to the target combination scheme based on the evaluation result.

The judging module can be used for determining pest situations based on the farm information, wherein the pest situations comprise multiple pest types.

In some embodiments, the determining module is to: and determining the corresponding multiple insect pest types through an insect pest determination model based on the farm information.

The characteristic extraction module can be used for determining a plurality of searching and killing schemes corresponding to a plurality of insect pest types and acquiring at least one characteristic of crop influence characteristics, soil influence characteristics and pesticide effect characteristics corresponding to the plurality of searching and killing schemes.

The combination module can be used for determining a plurality of combination schemes based on a plurality of combination modes of a plurality of searching and killing schemes.

In some embodiments, the combination module is for: obtaining various combinations of various searching and killing schemes according to the permutation and combination; determining a preset condition based on the farm information; and taking a plurality of combinations meeting preset conditions in the plurality of combinations as a plurality of combination schemes.

And the feature fusion module can be used for obtaining corresponding fusion features based on at least one feature included in each combination scheme.

And the scoring module can be used for obtaining at least one of crop growth potential influence scores and soil quality condition influence scores corresponding to the multiple combination schemes through a prediction model based on the fusion characteristics and the farm information.

A determination module may be configured to determine a target combination scheme based on the at least one score.

It should be noted that the above description of the farm pest killing system and its modules is for convenience of description only, and should not limit the scope of the present disclosure to the illustrated embodiments. It will be appreciated by those skilled in the art that, given the teachings of the present system, any combination of modules or sub-system configurations may be used to connect to other modules without departing from such teachings. In some embodiments, the obtaining module 410, the determining module 420, the feature extracting module 430, the combining module 440, the feature fusing module 450, the scoring module 460, and the determining module 470 disclosed in fig. 4 may be different modules in a system, or may be a module that implements the functions of two or more modules described above. For example, each module may share one memory module, and each module may have its own memory module. Such variations are within the scope of the present disclosure.

The embodiment of the specification also provides a farm pest searching and killing device which comprises at least one processor and at least one memory; the at least one memory is for storing computer instructions; the at least one processor is configured to execute at least some of the computer instructions to implement a method for farm pest extermination.

The beneficial effects that may be brought by the embodiments of the present description include, but are not limited to: (1) by screening and combining the checking and killing schemes under preset conditions, the scheme with large side effects (such as damaged crops, soil pollution and the like) can be eliminated, the calculated amount is reduced, and the efficiency is improved. (2) Loss functions related to crop influence characteristics, soil influence characteristics and pesticide effect characteristics are adopted in the training of the prediction model, and the accuracy of the prediction model is improved. (3) Based on the evaluation score reflecting the actual searching and killing effect, the weight of the target combination scheme is adjusted, the prediction score is corrected, the searching and killing scheme score of the actual effect can be higher, and the searching and killing scheme with the better actual effect can be obtained. (4) The pest searching and killing effect is evaluated in stages, the problems in the searching and killing period can be found in time, the deviation can be corrected in time by mainly observing the problematic area, the searching and killing efficiency is improved, and the searching and killing effect is ensured.

It is to be noted that different embodiments may produce different advantages, and in different embodiments, any one or combination of the above advantages may be produced, or any other advantages may be obtained.

Having thus described the basic concept, it will be apparent to those skilled in the art that the foregoing detailed disclosure is to be regarded as illustrative only and not as limiting the present specification. Various modifications, improvements and adaptations to the present description may occur to those skilled in the art, although not explicitly described herein. Such modifications, improvements and adaptations are proposed in the present specification and thus fall within the spirit and scope of the exemplary embodiments of the present specification.

Also, the description uses specific words to describe embodiments of the description. Reference throughout this specification to "one embodiment," "an embodiment," and/or "some embodiments" means that a particular feature, structure, or characteristic described in connection with at least one embodiment of the specification is included. Therefore, it is emphasized and should be appreciated that two or more references to "an embodiment" or "one embodiment" or "an alternative embodiment" in various places throughout this specification are not necessarily all referring to the same embodiment. Furthermore, some features, structures, or characteristics of one or more embodiments of the specification may be combined as appropriate.

Moreover, those skilled in the art will appreciate that aspects of the present description may be illustrated and described in terms of several patentable species or situations, including any new and useful combination of processes, machines, manufacture, or materials, or any new and useful improvement thereof. Accordingly, aspects of this description may be performed entirely by hardware, entirely by software (including firmware, resident software, micro-code, etc.), or by a combination of hardware and software. The above hardware or software may be referred to as "data block," module, "" engine, "" unit, "" component, "or" system. Furthermore, aspects of the present description may be represented as a computer product, including computer readable program code, embodied in one or more computer readable media.

The computer storage medium may comprise a propagated data signal with the computer program code embodied therewith, for example, on baseband or as part of a carrier wave. The propagated signal may take any of a variety of forms, including electromagnetic, optical, etc., or any suitable combination. A computer storage medium may be any computer-readable medium that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. Program code located on a computer storage medium may be propagated over any suitable medium, including radio, cable, fiber optic cable, RF, or the like, or any combination of the preceding.

Computer program code required for the operation of various portions of this specification may be written in any one or more programming languages, including an object oriented programming language such as Java, Scala, Smalltalk, Eiffel, JADE, Emerald, C + +, C #, VB.NET, Python, and the like, a conventional programming language such as C, Visual Basic, Fortran2003, Perl, COBOL2002, PHP, ABAP, a dynamic programming language such as Python, Ruby, and Groovy, or other programming languages. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or processing device. In the latter scenario, the remote computer may be connected to the user's computer through any network format, such as a Local Area Network (LAN) or a Wide Area Network (WAN), or the connection may be made to an external computer (for example, through the Internet), or in a cloud computing environment, or as a service, such as a software as a service (SaaS).

Additionally, the order in which the elements and sequences of the process are recited in the specification, the use of alphanumeric characters, or other designations, is not intended to limit the order in which the processes and methods of the specification occur, unless otherwise specified in the claims. While various presently contemplated embodiments of the invention have been discussed in the foregoing disclosure by way of example, it is to be understood that such detail is solely for that purpose and that the appended claims are not limited to the disclosed embodiments, but, on the contrary, are intended to cover all modifications and equivalent arrangements that are within the spirit and scope of the embodiments herein. For example, although the system components described above may be implemented by hardware devices, they may also be implemented by software-only solutions, such as installing the described system on an existing server or mobile device.

Similarly, it should be noted that in the preceding description of embodiments of the present specification, various features are sometimes grouped together in a single embodiment, figure, or description thereof for the purpose of streamlining the disclosure aiding in the understanding of one or more of the embodiments. This method of disclosure, however, is not intended to imply that more features than are expressly recited in a claim. Indeed, the embodiments may be characterized as having less than all of the features of a single embodiment disclosed above.

Numerals describing the number of components, attributes, etc. are used in some embodiments, it being understood that such numerals used in the description of the embodiments are modified in some instances by the use of the modifier "about", "approximately" or "substantially". Unless otherwise indicated, "about", "approximately" or "substantially" indicates that the number allows a variation of ± 20%. Accordingly, in some embodiments, the numerical parameters used in the specification and claims are approximations that may vary depending upon the desired properties of the individual embodiments. In some embodiments, the numerical parameter should take into account the specified significant digits and employ a general digit preserving approach. Notwithstanding that the numerical ranges and parameters setting forth the broad scope of the range are approximations, in the specific examples, such numerical values are set forth as precisely as possible within the scope of the application.

For each patent, patent application publication, and other material, such as articles, books, specifications, publications, documents, etc., cited in this specification, the entire contents of each are hereby incorporated by reference into this specification. Except where the application history document does not conform to or conflict with the contents of the present specification, it is to be understood that the application history document, as used herein in the present specification or appended claims, is intended to define the broadest scope of the present specification (whether presently or later in the specification) rather than the broadest scope of the present specification. It is to be understood that the descriptions, definitions and/or uses of terms in the accompanying materials of this specification shall control if they are inconsistent or contrary to the descriptions and/or uses of terms in this specification.

Finally, it should be understood that the embodiments described herein are merely illustrative of the principles of the embodiments of the present disclosure. Other variations are also possible within the scope of the present description. Thus, by way of example, and not limitation, alternative configurations of the embodiments of the specification can be considered consistent with the teachings of the specification. Accordingly, the embodiments of the present description are not limited to only those embodiments explicitly described and depicted herein.

Claims

1. A farm pest searching and killing method comprises the following steps:

acquiring farm information, wherein the farm information comprises crop information and environment information;

determining pest conditions based on the farm information, wherein the pest conditions include a plurality of pest types;

determining a plurality of searching and killing schemes corresponding to a plurality of insect pest types, and acquiring a plurality of characteristics corresponding to each searching and killing scheme in the plurality of searching and killing schemes, wherein the plurality of characteristics comprise crop influence characteristics, soil influence characteristics and pesticide effect characteristics;

determining a plurality of combination schemes based on a plurality of combination modes of the plurality of searching and killing schemes;

obtaining fusion characteristics corresponding to each combination scheme through a characteristic fusion unit based on the multiple characteristics of each searching and killing scheme included in each combination scheme, wherein the characteristic fusion unit comprises a deep neural network;

obtaining at least one of a crop growth potential influence score and a soil health condition influence score corresponding to each combination scheme in the plurality of combination schemes through a prediction model based on the fusion characteristics and the farm information, wherein the prediction model comprises a machine learning model;

determining a target combination scenario among the plurality of combination scenarios based on the at least one score.

2. The method of claim 1, wherein said determining a plurality of combination programs based on a plurality of combinations of said plurality of killing programs comprises:

obtaining the various combinations of the various searching and killing schemes according to permutation and combination;

determining a preset condition based on the farm information;

and taking a plurality of combinations which meet the preset condition in the plurality of combinations as the plurality of combination schemes.

3. The method of claim 1, the determining pest status based on the farm information comprising:

and determining the corresponding multiple insect pest types through an insect pest determination model based on the farm information.

4. The method of claim 1, the obtaining farm information comprising:

setting a plurality of sampling points in at least one area of the farm;

obtaining multiple groups of corresponding crop data and multiple groups of corresponding environmental data at the multiple sampling points through a monitoring device;

the method comprises the steps that multiple groups of crop data and multiple groups of environment data of multiple sampling points are collected through an unmanned aerial vehicle, statistics analysis is conducted on the multiple groups of crop data and the multiple groups of environment data based on corresponding weights of the sampling points, farm information is obtained, and the weights are related to the density of the sampling points in the area where the sampling points are located.

5. The method of claim 4, further comprising:

acquiring stage farm information of at least 1 time stage through a monitoring device in a period of time for pest searching and killing by adopting the target combination scheme;

determining at least one stage score in stage crop growth potential influence scores and stage soil health influence scores of the target combination scheme through the prediction model based on the fusion features and the stage farm information;

determining the density of sampling points and/or adjusting a sampling trajectory of the at least one region for a next time phase based on the at least one phasic score for a current time phase.

6. The method of claim 4, further comprising:

obtaining farm information change values before and after the objective combination scheme is adopted for checking and killing based on the multiple groups of crop data and the multiple groups of environmental data of the multiple sampling points;

obtaining an evaluation result of the target combination scheme through an evaluation model based on the farm information change value;

based on the evaluation result, model parameters of the prediction model are adjusted.

7. The method of claim 6, wherein the model parameters of the predictive model include a plurality of scoring weights corresponding to the plurality of combination scenarios, and wherein adjusting the model parameters of the predictive model based on the evaluation comprises:

and adjusting the scoring weight corresponding to the target combination scheme based on the evaluation result.

8. A farm pest searching and killing system comprises:

the system comprises an acquisition module, a storage module and a management module, wherein the acquisition module is used for acquiring farm information which comprises crop information and environment information;

the judging module is used for determining insect pest situations based on the farm information, wherein the insect pest situations comprise various insect pest types;

the characteristic extraction module is used for determining a plurality of searching and killing schemes corresponding to a plurality of insect pest types and acquiring a plurality of characteristics corresponding to each searching and killing scheme in the plurality of searching and killing schemes, wherein the plurality of characteristics comprise crop influence characteristics, soil influence characteristics and pesticide effect characteristics;

the combination module is used for determining a plurality of combination schemes based on a plurality of combination modes of the plurality of searching and killing schemes;

the feature fusion module is used for obtaining fusion features corresponding to each combination scheme through a feature fusion unit based on the multiple features of each searching and killing scheme included in each combination scheme, and the feature fusion unit comprises a deep neural network;

the scoring module is used for obtaining at least one of a crop growth potential influence score and a soil health condition influence score corresponding to each combination scheme in the plurality of combination schemes through a prediction model based on the fusion characteristics and the farm information, and the prediction model comprises a machine learning model;

a determining module for determining a target combination plan among the plurality of combination plans according to the at least one score.

9. A farm pest killing device, the device comprising at least one processor and at least one memory;

the at least one memory is for storing computer instructions;

the at least one processor is configured to execute at least some of the computer instructions to implement the method of any of claims 1-7.

10. A computer readable storage medium storing computer instructions which, when executed by a processor, implement the method of any one of claims 1-7.