CN115796558B - Tobacco planting area intelligent management system based on deep neural network - Google Patents
Tobacco planting area intelligent management system based on deep neural network Download PDFInfo
- Publication number
- CN115796558B CN115796558B CN202310052828.0A CN202310052828A CN115796558B CN 115796558 B CN115796558 B CN 115796558B CN 202310052828 A CN202310052828 A CN 202310052828A CN 115796558 B CN115796558 B CN 115796558B
- Authority
- CN
- China
- Prior art keywords
- planting
- tobacco
- degree
- acquiring
- land
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 235000002637 Nicotiana tabacum Nutrition 0.000 title claims abstract description 137
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 22
- 244000061176 Nicotiana tabacum Species 0.000 title description 5
- 241000208125 Nicotiana Species 0.000 claims abstract description 132
- 238000012549 training Methods 0.000 claims abstract description 16
- 238000007726 management method Methods 0.000 claims description 31
- 238000000034 method Methods 0.000 claims description 31
- 238000009826 distribution Methods 0.000 claims description 28
- 230000008859 change Effects 0.000 claims description 21
- 238000004364 calculation method Methods 0.000 claims description 14
- 238000013139 quantization Methods 0.000 claims description 7
- 238000000354 decomposition reaction Methods 0.000 claims description 3
- 241001632422 Radiola linoides Species 0.000 claims 1
- 230000008901 benefit Effects 0.000 abstract description 17
- 238000004519 manufacturing process Methods 0.000 abstract description 12
- 238000012545 processing Methods 0.000 abstract description 3
- 230000011218 segmentation Effects 0.000 description 9
- 230000008569 process Effects 0.000 description 7
- 239000002689 soil Substances 0.000 description 7
- 238000005457 optimization Methods 0.000 description 4
- 238000012937 correction Methods 0.000 description 3
- 239000000463 material Substances 0.000 description 3
- IJGRMHOSHXDMSA-UHFFFAOYSA-N Atomic nitrogen Chemical compound N#N IJGRMHOSHXDMSA-UHFFFAOYSA-N 0.000 description 2
- 241001632427 Radiola Species 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000004720 fertilization Effects 0.000 description 2
- 239000002245 particle Substances 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 2
- OAICVXFJPJFONN-UHFFFAOYSA-N Phosphorus Chemical compound [P] OAICVXFJPJFONN-UHFFFAOYSA-N 0.000 description 1
- ZLMJMSJWJFRBEC-UHFFFAOYSA-N Potassium Chemical compound [K] ZLMJMSJWJFRBEC-UHFFFAOYSA-N 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 229910052757 nitrogen Inorganic materials 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 235000021049 nutrient content Nutrition 0.000 description 1
- 235000015097 nutrients Nutrition 0.000 description 1
- 229910052698 phosphorus Inorganic materials 0.000 description 1
- 239000011574 phosphorus Substances 0.000 description 1
- 239000011591 potassium Substances 0.000 description 1
- 229910052700 potassium Inorganic materials 0.000 description 1
- 239000004016 soil organic matter Substances 0.000 description 1
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Abstract
The invention relates to the field of data processing for management purposes, and provides an intelligent tobacco planting area management system based on a deep neural network, which comprises the following steps: acquiring historical reference data of a tobacco planting field; acquiring an influence weight value of each influence factor according to the correlation between each influence factor in the historical reference data and the tobacco yield, and obtaining a tobacco yield prediction model after training; constructing a first objective function according to the prediction model, acquiring a first price degree according to the weight value difference expression of different influence factors of various planting plots, acquiring a second cost degree according to the discreteness of the planting areas of various planting plots and the planting density variation degree, further acquiring a comprehensive cost degree, and correcting the first objective function to acquire a second objective function; and setting constraint conditions according to the second objective function to obtain an optimal planting allocation mode. The invention aims to solve the problem of planting management that the economic objective is met but other influencing factors cannot be considered for planting and the production benefit cannot be maximized.
Description
Technical Field
The invention relates to the field of data processing for management purposes, in particular to an intelligent tobacco planting area management system based on a deep neural network.
Background
Tobacco belongs to special cash crops, along with the development of scientific technology, the tobacco planting industry gradually realizes digital management transformation, so that the tobacco planting production benefit is continuously increased, the purpose of increasing yield and efficiency is achieved, and the tobacco planting planning is an important means for increasing the economic benefit. In the existing tobacco planting planning management process, farmers mainly perform blind planting on planting plots with limited areas according to subjective willingness of the farmers and influence of market economy, and on the premise of meeting economic targets, other planting influence factors are considered, so that the management purpose of maximizing production benefits is difficult to achieve; if higher production benefit is achieved and other influencing factors are considered, an expert is often required to examine and plan on the spot, a great deal of manpower and material resources are wasted, and therefore, the digital intelligent management planting planning transformation is required; in the digital intelligent management planting planning, various data of a large amount of manpower and material resources are not wasted any more to inspect the planting field in the field, a prediction model is built by combining historical reference data, the optimal planting planning can be completed, and the problem that the production benefit is not maximized due to the fact that the existing planning is used for blind planting is solved.
Disclosure of Invention
The invention provides an intelligent tobacco planting area management system based on a deep neural network, which aims to solve the existing planting management problem that the economic objective is met but other influencing factors cannot be considered for planting and the production benefit cannot be maximized, and adopts the following technical scheme:
one embodiment of the invention provides an intelligent tobacco planting area management system based on a deep neural network, which comprises:
the data acquisition module is used for acquiring historical reference data of the tobacco planting field, wherein the historical reference data comprises calendar data of each influence factor and calendar tobacco yield; acquiring data of each influence factor of each planting land in a tobacco planting field;
the prediction model module is used for acquiring an influence weight value of each influence factor according to the historical data of each influence factor in the historical reference data and the tobacco yield of each year, constructing a tobacco yield prediction model by using the deep neural network, and acquiring a tobacco yield prediction model after training according to the historical reference data and the influence weight values of each influence factor;
an objective function module: acquiring a first objective function according to a tobacco yield prediction model, and acquiring a first price degree of each planting land according to data differences of each influence factor of each planting land in a tobacco planting field and each influence factor of other planting lands and comparison results of influence weight values of each influence factor and a first preset threshold;
acquiring a implantable region in each planted land, acquiring the discreteness of the implantable region of each planted land according to the average value of the distance and the area expression of the implantable region in each planted land, acquiring a planting density trend curve of each implantable region according to the change relation between different seed costs and areas in each implantable region, acquiring the planting density change degree of each planted land according to the trend difference expression of each implantable region and other implantable regions in each planted land, and acquiring the second cost degree of each planted land according to the discreteness and the planting density change degree of each planted land;
acquiring the comprehensive cost degree of each planting land according to the first cost degree and the second cost degree of each planting land, and correcting the first objective function according to the comprehensive cost degree to obtain a second objective function;
planting allocation management module: and setting constraint conditions according to the second objective function, and obtaining the optimal planting allocation management method.
Optionally, the method for obtaining the influence weight value of each influence factor includes the following specific steps:
wherein ,
represent the first
The degree of influence of the individual influencing factors,
representing commonality in historical reference data
Reference data for a number of years,
represent the first
The influencing factors are at
The history of the year is used to determine,
represent the first
The mean of the individual influencing factors in the historical reference data,
represent the first
The annual tobacco yield is determined by the ratio of the tobacco to the tobacco,
representing the mean value of tobacco yield in the historical reference data;
and taking the normalized value of the influence degree of each influence factor as the influence weight value of each influence factor.
Optionally, the method for obtaining the first objective function according to the tobacco yield prediction model includes the following specific steps:
wherein ,
i.e. the predicted total yield of tobacco,
indicating that the tobacco plants are common in the field
The number of the planting plots is three,
represent the first
The cost of seeds distributed by each planting field,
representing a tobacco yield prediction model;
the function of the obtained predicted total tobacco yield is the first objective function.
Optionally, the method for obtaining the first price degree of each planting land comprises the following specific steps:
wherein ,
represent the first
The first degree of price of the individual plots,
indicating the number of influencing factors having influencing weight values smaller than a first preset threshold,
indicating the number of influencing factors with influencing weight values larger than or equal to a first preset threshold value,
and
respectively represent the first
And (b)
The first influence weight value in each planting land is smaller than a first preset threshold value
The data of the individual influencing factors are provided,
and
respectively represent the first
And (b)
The first influence weight value in each planting land block is larger than or equal to a first preset threshold value
The data of the individual influencing factors are provided,
represents the number of planting plots in the tobacco planting field,
representing absolute values.
Optionally, the method for obtaining the discreteness of each planting area comprises the following specific steps:
wherein ,
represent the first
The discreteness of the implantable areas of the individual planting plots,
represent the first
The average value of Euclidean distance between all connected domains formed by the implantable areas in the planting plots,
the number of plantable regions in the representation,
represent the first
The first of the planting plots
The area of the communicating region formed by each implantable region,
represent the first
The average area value of the connected domain formed by all the plantable areas in the planting plots; the method for acquiring the mean value of the Euclidean distance between the connected domains comprises the following steps: and calculating Euclidean distances of the centers of any two connected domain areas, and acquiring the average value of all acquired Euclidean distances in the planting land.
Optionally, the obtaining the planting density trend curve of each implantable area includes the following specific methods:
and constructing a planting density change curve for each implantable region in the same planting land by taking the abscissa as different seed cost values and the ordinate as a planting density value, wherein the planting density value is the ratio of the seed cost to the area of a communicating region formed by the implantable region, and obtaining a planting density trend curve of each implantable region by using the planting density change curve of each implantable region through an STL time sequence decomposition algorithm.
Optionally, the method for obtaining the planting density variation degree of each planting land comprises the following specific steps:
wherein ,
represent the first
The degree of change of the planting density of each planting land block,
represent the first
The number of plantable areas in a single planting field,
represent the first
The first of the planting plots
The trend distribution of each implantable area is different from the trend distribution of all other implantable areas by a mean value;
the calculation method of the trend distribution difference comprises the following steps: and acquiring the absolute values of the longitudinal coordinate difference values of corresponding curve points in the two curves under each seed cost in the planting density trend curves of the two planting areas, and taking the average value of the absolute values of the longitudinal coordinate difference values under all seed costs as the trend distribution difference of the two planting areas.
Optionally, the method for correcting the first objective function according to the comprehensive cost degree to obtain the second objective function includes the following specific steps:
wherein ,
indicating the predicted total tobacco yield after correction,
indicating that the tobacco plants are common in the field
The number of the planting plots is three,
represent the first
The cost of seeds distributed by each planting field,
a tobacco yield prediction model is represented and is used for the prediction of tobacco yield,
represent the first
The comprehensive cost degree of each planting land block,
indicating a cost level quantization hyper-parameter.
Compared with the prior art, the invention has the beneficial effects that:
(1) In the digital intelligent management planting planning, various data of a planting field are not required to be investigated in the field by wasting a large amount of manpower and material resources, a prediction model and an objective function are established by combining historical reference data, so that the optimal planting planning and management can be finished, the management problem that the production benefit cannot be maximized due to the blind planting in the existing planning is solved, and meanwhile, the prediction can be finished without consuming more financial resources.
(2) The method comprises the steps of acquiring influence weight values of different influence factors influencing tobacco yield through historical reference data, so that a tobacco yield prediction model after training is more in line with inherent attribute characteristics of each planting land, further, a prediction result of the tobacco yield prediction model is more accurate, and a planning basis is provided for optimization distribution in the future; meanwhile, the influence weight values obtained according to different influence factors provide a calculation basis for quantifying the comprehensive cost degree of each planting land block later, so that the quantified comprehensive cost degree is more accurate.
(3) Quantifying the comprehensive cost degree of tobacco planting among different plots by the different weights of influence factors among different plots and the different corresponding implantable areas; the comprehensive cost degree of tobacco planting is used for representing inherent attribute factors of the current planting land and costs of the distribution of the implantable areas for planting a certain amount of tobacco, and then the constructed objective function is adjusted, so that the maximum planting yield is achieved, the consumed costs are smaller, and the obtained optimal distribution is more reasonable.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions of the prior art, the drawings which are used in the description of the embodiments or the prior art will be briefly described, it being obvious that the drawings in the description below are only some embodiments of the invention, and that other drawings can be obtained according to these drawings without inventive faculty for a person skilled in the art.
Fig. 1 is a block diagram of a tobacco planting area intelligent management system based on a deep neural network according to an embodiment of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Referring to fig. 1, a block diagram of an intelligent tobacco planting area management system based on a deep neural network according to an embodiment of the present invention is shown, where the system includes:
and the data acquisition module S101 acquires historical reference data of the tobacco planting field.
The aim of the embodiment is to predict the tobacco yield according to each influencing factor and complete an optimal planting allocation management method so as to obtain the maximum production benefit; by digitally and intelligently managing the planting planning, the field investigation is not needed, the tobacco yield is predicted by the relation of the data layers, and the purpose of planting planning and management with maximized production benefits is achieved.
Firstly, acquiring historical reference data according to the prediction of tobacco yield according to each influencing factor, wherein the historical reference data comprises related data of each influencing factor in the past year, seed cost and tobacco yield in the past year and images of tobacco planting fields; wherein various influencing factors comprise climate, soil nitrogen, phosphorus and potassium content, soil nutrient content, soil organic matter content, soil water content and the like; the tobacco planting field contains a plurality of tobacco planting plots, and this embodiment adopts unmanned aerial vehicle to shoot and acquires the image of every planting plot in order to plan planting area more rationally to splice after denoising processing and form a complete tobacco planting field image.
In the process of planning the tobacco planting area, under the environment of different plots, the soil nutrients and the soil water content of the plots are different, the number of the preset planted tobacco seeds is different, the planting density is different, and the tobacco yield of the corresponding different plots is also different; therefore, in order to better achieve the maximization of economic benefit on the basis of limited production cost of various planting plots, the objective function of economic benefit is maximized in all plots by constructing the objective function of economic benefit for each planting plot, so that the purpose of reasonably planning the planting area of tobacco is achieved; in the process of constructing the objective function, in the embodiment, the tobacco yield is taken as the quantization characteristic of economic benefit, the seed cost is taken as the quantization characteristic of the planting distribution mode for analysis, and in the specific implementation process, an implementer can adjust the economic benefit and the quantization characteristic of the planting distribution mode according to the situation.
And the prediction model module S102 acquires the influence weight value of each influence factor according to the correlation between each influence factor in the historical reference data and the tobacco yield, and obtains a tobacco yield prediction model after training.
It should be noted that, besides the most fundamental seed cost, the factors affecting the tobacco yield are affected by conditions such as climate and soil, and the influence degree of each affecting factor on the tobacco yield is different, so that according to the data of each affecting factor and the tobacco yield in the historical reference data, the correlation between each affecting factor and the tobacco yield is obtained, and a basis is provided for building a tobacco yield prediction model.
Specifically, according to the data relation between each year influence factor in the historical reference data and the tobacco yield of the corresponding year, obtaining the tobacco yield of each influence factorDegree of influence, by
By way of example, the degree of influence of the factors on the tobacco yield is obtained
The calculation method of (1) is as follows:
wherein ,
representing commonality in historical reference data
Reference data for a number of years,
represent the first
The influencing factors are at
The history of the year is used to determine,
represent the first
The mean of the individual influencing factors in the historical reference data,
represent the first
The annual tobacco yield is determined by the ratio of the tobacco to the tobacco,
representing the mean value of tobacco yield in the historical reference data; obtaining the influence degree of all influence factors on tobacco yield according to the method, and passing each influence degree through sThe oftmax function is normalized, and the normalized result is recorded as an influence weight value, namely
Degree of influence of individual influencing factors
The influence weight value obtained by the normalization result of (2) is recorded as
。
At this time, the calculation formula of the influence degree is essentially a calculation of the correlation, and the smaller the difference between the numerator and the denominator is, the closer the influence factor and the change trend of the tobacco yield in each year are, the larger the correlation is, and the larger the influence degree of the influence factor on the tobacco yield is.
Further, in order to more reasonably plan the tobacco planting area, a tobacco yield prediction model needs to be constructed to quantify the relationship between the influencing factors and the tobacco yield, and the process of constructing and training the tobacco yield prediction model is as follows:
constructing a prediction neural network, and adopting the existing BP neural network;
randomly initializing parameters of a neural network;
taking historical reference data of a plurality of tobacco planting fields as samples, taking the current-year tobacco yield corresponding to each tobacco planting field as a label, and forming a training data set by all samples and the labels;
inputting each annual influence factor data in the historical reference data and seed cost into a neural network, and outputting the data as predicted tobacco yield corresponding to the current year;
the purpose of the network is to make predictions, so the loss function uses a root-mean-square error function;
and training and predicting the neural network by using the training data set according to the loss function by using a random gradient descent algorithm to enable the neural network to converge, so as to obtain a tobacco yield prediction model after training.
The influence weight value of each influence factor is obtained according to the historical reference data, and a tobacco yield prediction model after training is obtained, so that the predicted tobacco yield can be more compatible with each influence factor, and a planning basis is provided for optimization distribution in the future.
And the objective function module S103 is used for constructing a first objective function according to the prediction model, obtaining a first price degree according to the weight value difference expression of different influence factors of various planting plots, obtaining a second cost degree according to the discreteness and planting density change degree of the planting regions of various planting plots, further obtaining a comprehensive cost degree, and correcting the first objective function to obtain the second objective function.
It should be noted that, the trained tobacco yield prediction model is already obtained in the block S102, and is used to construct the first objective function, so as to ensure that the maximum production benefit is achieved at a limited cost, and further, the optimal planting allocation mode is completed.
Specifically, a first objective function is firstly constructed according to a prediction model, and the calculation method comprises the following steps:
wherein ,
i.e. the predicted total yield of tobacco,
indicating that the tobacco plants are common in the field
The number of the planting plots is three,
represent the first
Tobacco yield predicted by individual plots based on seed cost,
represent the first
The cost of seeds distributed by each planting field,
representing a tobacco yield prediction model; it should be noted that, since each planting plot has different seed costs, other influencing factors are fixed in the planting year, that is, the predicted yield can be changed only by adjusting the seed costs, so that the tobacco yield of each planting plot is predicted by using the prediction model on the change of the seed costs; at this time, the cost of seeds allocated per plot maximizes the predicted yield, i.e., the predicted total tobacco yield.
It should be further noted that the first objective function obtained at this time is allocated only according to the tobacco yield; however, in the actual tobacco planting process, the planting cost needs to be considered, for example, the planting is performed in what planting mode, the tobacco yield is maximized on the basis of the minimum labor cost, and meanwhile, not all areas on the planting land are suitable for planting due to the inherent position distribution difference of the planting land, for example, the planting land contains field roads, drainage channels and the like; therefore, the first objective function needs to be corrected through the difference expression of the influence weight values of different influence factors of various planting plots, the discreteness of the planting areas of various planting plots and the planting density variation degree, so that a more reasonable planting distribution mode can be obtained.
Firstly, according to the difference expression of influence weight values of different influence factors of various planting plots, obtaining a first price degree of each planting plot; various influencing factors of various planting plots are different, for example, if the soil quality of a certain planting plot is higher, then less planting cost is needed to obtain higher tobacco yield, and the planting cost comprises planting labor such as cultivation and the like; that is, in a certain planting plot, the larger the positively-related influence factor is, the higher tobacco yield can be obtained with less planting cost, and conversely, the larger the negatively-related influence factor is, the higher tobacco yield can be obtained with more planting cost, and a first preset threshold value is given
For determining whether the influencing factors are positive or negative, in this embodiment
To perform the calculation.
Specifically, by the first
Taking a plurality of planting plots as an example, obtaining the first price degree of the planting plots
The calculation method of (1) is as follows:
wherein ,
indicating that the impact weight value is less than
Is used for the number of influencing factors of (a),
indicating that the influence weight value is greater than or equal to
Is used for the number of influencing factors of (a),
and
respectively represent the first
And (b)
The influence weight value in each planting land is smaller than
Is the first of (2)
The data of the individual influencing factors are provided,
and
respectively represent the first
And (b)
The influence weight value in each planting land is greater than or equal to
Is the first of (2)
The data of the individual influencing factors are provided,
represents the number of planting plots in the tobacco planting field,
representing absolute value; the denominator 1 is added to avoid the case where the denominator is 0, and the denominator 1 is added to keep the same with the denominator.
At this time, the first
The smaller the difference between the positive correlation influence factors in the individual planting plots and the positive correlation influence factors in the average value of all the planting plots, the smaller the positive influence in the planting plots, and the higher tobacco yield can be obtained only by correspondingly needing more planting cost, and the higher the first price degree is; the larger the difference between the negative correlation influence factors and the corresponding average values is, the larger the negative influence in the planting land is, and the higher the tobacco yield can be obtained only with more planting cost, and the higher the first price degree is; the first price degree of each planting land is obtained according to the method.
Further, semantic segmentation is carried out on images of various planting plots in the tobacco planting field, and the construction and training processes of the semantic segmentation are as follows:
constructing a semantic segmentation network, and adopting the existing DNN network structure;
randomly initializing parameters of a semantic segmentation network;
taking a tobacco planting land block image in the Internet as a training data set, marking a planting area in a planting land block as 1, and marking a non-planting area as 0;
inputting each tobacco planting land block image in the training data set into a semantic segmentation network, wherein the output result is a mask corresponding to the input image, the planting area is 1, and the non-planting area is 0;
the network aims at classifying, and the loss function adopts a cross entropy loss function;
and training the semantic segmentation network by using the training data set according to the loss function by using a random gradient descent algorithm to enable the semantic segmentation network to be converged, so as to obtain a trained semantic segmentation network.
Inputting various planting land block images acquired by an unmanned aerial vehicle into a trained semantic segmentation network, and acquiring a planting area in each planting land block; at this time, the more the areas which can be planted in a certain planting land are dispersed, the more labor cost is required to be consumed in the planting land, for example, more labor and time are required for manual watering and fertilization, and the planting cost of the corresponding planting land is higher.
Specifically, by the first
Taking a plurality of planting plots as an example, obtaining the discreteness of the planting plots in the planting regions
The calculation method of (1) is as follows:
wherein ,
represent the first
An average value of Euclidean distances among all connected domains formed by the implantable areas in the planting plots; the method for acquiring the mean value of the Euclidean distance between the connected domains comprises the following steps: calculating Euclidean distances of the centers of any two connected domain areas, and acquiring the average value of all the acquired Euclidean distances in the planting land, wherein the area center is the average value of the transverse coordinates and the longitudinal coordinates of all the points in the connected domain;
represent the first
The number of plantable areas in a single planting field,
represent the first
The first of the planting plots
The area of the communicating region formed by each implantable region,
represent the first
The average area value of the connected domain formed by all the plantable areas in the planting plots; 0.5 represents a reference coefficient, and in this embodiment, the Euclidean distance mean value and the area difference between connected domains are considered to be equally important for discrete calculation.
At the moment, the larger the Euclidean distance average value among all the connected domains formed by the implantable areas is, the more dispersed the implantable areas in the planted land are, and the larger the corresponding discreteness is; the larger the area variance of each connected domain formed by the implantable region, the larger the area variance, the more labor cost the implantable region needs to consume, and the larger the corresponding discreteness.
It should be further noted that, if the difference of the correlation between the planting density and the seed cost of each planting-capable area is larger in the same planting plot, that is, the degree of variation of the planting density is larger, more labor cost is required in the same planting plot, and the planting cost of the corresponding planting plot is larger.
Specifically, firstly, taking an abscissa as different seed cost values and an ordinate as a planting density value, wherein the planting density value is the ratio of the seed cost to the area of a communicating region formed by the implantable regions, and constructing a planting density change curve for each implantable region in the same planting land; it should be noted that, in this embodiment, the difference and the variation performance of the correlation are considered, so the variation degree of the planting density can be reflected by the difference of the curve trend performance; the planting density change curve of each implantable region is obtained by using an STL time sequence decomposition algorithm, the planting density trend curve of each implantable region is obtained, and the planting density change degree of the planted land is obtained through the trend distribution difference of each implantable region and other implantable regions, so as to obtain the planting density change degree of the planted land
For example, the planting density of each planting field is changed
The calculation method of (1) is as follows:
wherein ,
represent the first
The number of plantable areas in a single planting field,
represent the first
The first of the planting plots
The trend distribution of each implantable area is different from the trend distribution of all other implantable areas by a mean value; the calculation method of the trend distribution difference comprises the following steps: acquiring the absolute values of the longitudinal coordinate difference values of corresponding curve points in two curves under each seed cost in the planting density trend curves of the two planting areas, and calculating the average value of the absolute values of the longitudinal coordinate difference values under all seed costs to obtain trend distribution differences of the two planting areas; at this time, the greater the trend distribution difference between the respective plantable regions, the greater the degree of change in the planting density of the planted plots.
Further, a second cost degree of each planting land is obtained according to the discreteness of the planting area and the planting density variation degree of each planting land, so as to obtain
For example, the second degree of cost is that of the planting land
The calculation method of (1) is as follows:
wherein ,
represent the first
The discreteness of the plantable area of individual planting plots,
represent the first
The degree of change of the planting density of each planting land block; the larger the discreteness of the implantable area is, the more labor cost is needed, and the correspondingThe greater the planting cost, the greater the second cost degree; the greater the planting density variation degree is, the more labor cost is needed, the greater the planting cost is, and the greater the second cost degree is; and obtaining a second cost degree of each planting land according to the method.
The first price degree and the second cost degree of each planted land block are obtained, the product of the first price degree and the second cost degree of each planted land block is obtained, the obtained product of each planted land block is normalized, the obtained result is recorded as the comprehensive cost degree, the first price degree
The comprehensive cost degree of each planting land is recorded as
The method comprises the steps of carrying out a first treatment on the surface of the It should be noted that, because the seed cost is fixed, the seed cost needs to be allocated according to the comprehensive cost degree to obtain the maximum production benefit, and the obtained product of each planting land block is normalized by adopting the softmax method in the embodiment.
Further, the method for correcting the first objective function to obtain the second objective function according to the comprehensive cost degree of various planting plots comprises the following steps:
wherein ,
indicating the predicted total tobacco yield after correction,
indicating that the tobacco plants are common in the field
The number of the planting plots is three,
represent the first
Tobacco yield predicted by individual plots based on seed cost,
represent the first
The comprehensive cost of tobacco planting quantization of each planting plot,
represent the first
The cost of seeds distributed by each planting field,
a tobacco yield prediction model is represented and is used for the prediction of tobacco yield,
represent the first
The comprehensive cost degree of each planting land block,
the quantization super-parameter representing the cost degree can be quantized according to the specific implementation situation of an implementer, for example, according to the labor cost and the fertilization cost, and the super-parameter value set in the embodiment is 100.
And the planting distribution management module S104 sets constraint conditions according to the second objective function and obtains the optimal planting distribution management method.
In this case, the corrected predicted total tobacco yield in the second objective function
Can be regarded as the first
Seed cost for individual planting plot allocation
Wherein
Other parameters being models or values already determined, the present embodiment requires obtaining a model or value such that
The maximum seed cost distribution mode can be based on
And
to set constraints.
A second objective function for predicting the total tobacco yield has been obtained in block S103, and constraints are set according to the second objective function as follows:
wherein ,
in order to correct the predicted total tobacco yield after correction,
representing the set yield target, requiring the actual yield requirement of the implementer to be set, the specific value is not given in the embodiment;
represent the first
The cost of seeds distributed by each planting field,
representing the cost of the existing seed to be dispensed.
To achieve a second objective function result, i.e. a modified predicted total tobacco yield
Maximizing, according to a particle swarm optimization algorithm, obtaining an optimal seed cost distribution mode of various planting plots under a second objective function in a set constraint condition, wherein the particle swarm optimization algorithm is the prior art and is not repeated.
So far, the optimal seed cost distribution mode of various planting plots is obtained, namely the optimal planting distribution management method for the tobacco planting fields.
The foregoing description of the preferred embodiments of the invention is not intended to be limiting, but rather is intended to cover all modifications, equivalents, alternatives, and improvements that fall within the spirit and scope of the invention.
Claims (6)
1. Tobacco planting area intelligent management system based on degree of depth neural network, its characterized in that, this system includes:
the data acquisition module is used for acquiring historical reference data of the tobacco planting field, wherein the historical reference data comprises calendar data of each influence factor and calendar tobacco yield; acquiring data of each influence factor of each planting land in a tobacco planting field;
the prediction model module is used for acquiring an influence weight value of each influence factor according to the historical data of each influence factor in the historical reference data and the tobacco yield of each year, constructing a tobacco yield prediction model by using the deep neural network, and acquiring a tobacco yield prediction model after training according to the historical reference data and the influence weight values of each influence factor;
an objective function module: acquiring a first objective function according to a tobacco yield prediction model, and acquiring a first price degree of each planting land according to data differences of each influence factor of each planting land in a tobacco planting field and each influence factor of other planting lands and comparison results of influence weight values of each influence factor and a first preset threshold;
acquiring a implantable region in each planted land, acquiring the discreteness of the implantable region of each planted land according to the average value of the distance and the area expression of the implantable region in each planted land, acquiring a planting density trend curve of each implantable region according to the change relation between different seed costs and areas in each implantable region, acquiring the planting density change degree of each planted land according to the trend difference expression of each implantable region and other implantable regions in each planted land, and acquiring the second cost degree of each planted land according to the discreteness and the planting density change degree of each planted land;
acquiring the comprehensive cost degree of each planting land according to the first cost degree and the second cost degree of each planting land, and correcting the first objective function according to the comprehensive cost degree to obtain a second objective function;
planting allocation management module: setting constraint conditions according to the second objective function, and obtaining an optimal planting allocation management method;
the method for correcting the first objective function according to the comprehensive cost degree to obtain the second objective function comprises the following specific steps:
wherein ,
representing the corrected predicted total yield of tobacco, +.>
Indicating that there is common +.>
Planting land parcels, herba polygoni multiflori>
Indicate->
Seed cost allocated to individual planting plots, +.>
Representing a tobacco yield prediction model,/->
Indicate->
Comprehensive cost degree of each planting land block, < +.>
Representing a cost degree quantization hyper-parameter;
the method for acquiring the first objective function according to the tobacco yield prediction model comprises the following specific steps:
wherein ,
i.e. the predicted total yield of tobacco,/-)>
Indicating that there is common +.>
Planting land parcels, herba polygoni multiflori>
Indicate->
Seed cost allocated to individual planting plots, +.>
Representing a tobacco yield prediction model;
the function of the obtained predicted total tobacco yield is the first objective function.
2. The intelligent tobacco planting area management system based on the deep neural network according to claim 1, wherein the acquiring the influence weight value of each influence factor comprises the following specific methods:
wherein ,
indicate->
Degree of influence of individual influencing factors, +.>
Representing common +.>
Reference data of the year, +.>
Indicate->
The influencing factors are at->
Historical data of year, < >>
Indicate->
Mean value of individual influencing factors in historical reference data, < >>
Indicate->
Annual tobacco yield,/->
Representing the mean value of tobacco yield in the historical reference data;
and taking the normalized value of the influence degree of each influence factor as the influence weight value of each influence factor.
3. The intelligent tobacco planting area management system based on the deep neural network according to claim 1, wherein the obtaining the first price degree of each planting land comprises the following specific methods:
wherein ,
indicate->
First degree of price of individual planting plots, < ->
Indicating the number of influencing factors having a influencing weight value smaller than a first preset threshold value,/for>
Indicating the number of influencing factors with influencing weight values larger than or equal to a first preset threshold value, +.>
and />
Respectively represent +.>
Person and->
The +.sup.th of the influence weight value in the individual planting plots is smaller than a first preset threshold>
Data of individual influencing factors->
And
respectively represent +.>
Person and->
The first part of the influence weight value in each planting land is larger than or equal to a first preset threshold value>
Data of individual influencing factors->
Indicates the number of plots in the tobacco field, < > and->
Representing absolute values.
4. The intelligent tobacco planting area management system based on the deep neural network according to claim 1, wherein the method for obtaining the discreteness of each planting land block planting area comprises the following specific steps:
wherein ,
indicate->
Discretization of the implantable area of the individual planting plots,/->
Indicate->
Average value of Euclidean distance between communicating domains formed by the implantable regions in the planting plots>
Indicate->
The number of plantable areas in the individual planting plots, < >>
Indicate->
The first part of the planting land>
Area of the connected domain formed by the plantable region, < >>
Indicate->
The average area value of the connected domain formed by all the plantable areas in the planting plots; the method for acquiring the mean value of the Euclidean distance between the connected domains comprises the following steps: calculating Euclidean distance between centers of any two connected domain areas, and obtaining all obtained Euclidean distances in the planting landThe average value of the distance.
5. The intelligent tobacco planting area management system based on the deep neural network according to claim 1, wherein the obtaining the planting density trend curve of each implantable area comprises the following specific methods:
and constructing a planting density change curve for each implantable region in the same planting land by taking the abscissa as different seed cost values and the ordinate as a planting density value, wherein the planting density value is the ratio of the seed cost to the area of a communicating region formed by the implantable region, and obtaining a planting density trend curve of each implantable region by using the planting density change curve of each implantable region through an STL time sequence decomposition algorithm.
6. The intelligent tobacco planting area management system based on the deep neural network according to claim 1, wherein the method for obtaining the planting density variation degree of each planting land comprises the following specific steps:
wherein ,
indicate->
Degree of change of planting density of individual planting plots, +.>
Indicate->
The number of plantable areas in the individual planting plots, < >>
Indicate->
The first part of the planting land>
The trend distribution of each implantable area is different from the trend distribution of all other implantable areas by a mean value;
the calculation method of the trend distribution difference comprises the following steps: and acquiring the absolute values of the longitudinal coordinate difference values of corresponding curve points in the two curves under each seed cost in the planting density trend curves of the two planting areas, and taking the average value of the absolute values of the longitudinal coordinate difference values under all seed costs as the trend distribution difference of the two planting areas.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310052828.0A CN115796558B (en) | 2023-02-03 | 2023-02-03 | Tobacco planting area intelligent management system based on deep neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310052828.0A CN115796558B (en) | 2023-02-03 | 2023-02-03 | Tobacco planting area intelligent management system based on deep neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115796558A CN115796558A (en) | 2023-03-14 |
| CN115796558B true CN115796558B (en) | 2023-05-05 |
Family
ID=85429589
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310052828.0A Active CN115796558B (en) | 2023-02-03 | 2023-02-03 | Tobacco planting area intelligent management system based on deep neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115796558B (en) |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109892063A (en) * | 2019-03-28 | 2019-06-18 | 华中农业大学 | Utilize the method for the method and evaluation seed quality of high-temperature process prediction seed Seedling emergence rate |
Family Cites Families (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10667456B2 (en) * | 2014-09-12 | 2020-06-02 | The Climate Corporation | Methods and systems for managing agricultural activities |
| SG10201506012SA (en) * | 2015-07-31 | 2017-02-27 | Accenture Global Services Ltd | Inventory, growth, and risk prediction using image processing |
| US11145007B2 (en) * | 2017-08-21 | 2021-10-12 | The Climate Corporation | Digital modeling and tracking of agricultural fields for implementing agricultural field trials |
| CN107657385A (en) * | 2017-09-29 | 2018-02-02 | 西安微媒软件有限公司 | A kind industry data system |
| US20200005401A1 (en) * | 2018-07-02 | 2020-01-02 | The Climate Corporation | Optimal placement and portfolio opportunity targeting |
| MX2021010120A (en) * | 2019-02-21 | 2021-09-23 | Climate Llc | Digital modeling and tracking of agricultural fields for implementing agricultural field trials. |
| CN111226729A (en) * | 2020-02-21 | 2020-06-05 | 中国农业科学院烟草研究所 | Tobacco-rape self-flow irrigation double-cropping planting method in southwest tobacco area |
| CN115205695B (en) * | 2022-04-06 | 2024-03-15 | 布瑞克(苏州)农业互联网股份有限公司 | Method and system for determining planting strategy according to planting data |
-
2023
- 2023-02-03 CN CN202310052828.0A patent/CN115796558B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109892063A (en) * | 2019-03-28 | 2019-06-18 | 华中农业大学 | Utilize the method for the method and evaluation seed quality of high-temperature process prediction seed Seedling emergence rate |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115796558A (en) | 2023-03-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111126662B (en) | Irrigation decision making method, device, server and medium based on big data | |
| CN116307191B (en) | Water resource configuration method, device and equipment based on artificial intelligence algorithm | |
| CN106373022A (en) | BP-GA-based greenhouse crop plantation efficiency condition optimization method and system | |
| CN112352523B (en) | Tea garden water and fertilizer irrigation control method and system based on intelligent decision | |
| CN113159439A (en) | Crop yield prediction method and system, storage medium and electronic equipment | |
| CN117322214A (en) | Crop fertilizer accurate application method and system based on neural network | |
| CN115796558B (en) | Tobacco planting area intelligent management system based on deep neural network | |
| CN116863403B (en) | Crop big data environment monitoring method and device and electronic equipment | |
| CN115759487B (en) | Meteorological risk prediction method for penaeus vannamei boone cultivation fertilizer and water operation window period | |
| CN115879750B (en) | Aquatic seedling environment monitoring management system and method | |
| CN108122137A (en) | A kind of method and system of intelligent equalization agricultural product supply and demand | |
| Chen et al. | A water-saving irrigation decision-making model for greenhouse tomatoes based on genetic optimization TS fuzzy neural network | |
| CN113158565B (en) | Artificial intelligence method, system and electronic equipment for acquiring water-nitrogen strategy | |
| Chandana et al. | Efficient machine learning regression algorithm using naïve Bayes classifier for crop yield prediction and optimal utilization of fertilizer | |
| CN113221446A (en) | Method and system for acquiring water and fertilizer strategies of saline soil, storage medium and electronic equipment | |
| CN116595333B (en) | Soil-climate intelligent rice target yield and nitrogen fertilizer consumption determination method | |
| Bhansali et al. | Machine learning for sustainable agriculture | |
| Rubia et al. | Predictive Soil Analytics Using Data Mining Techniques | |
| Vijaya et al. | Developing AI-powered Systems to Optimize Planting, Irrigation, and Harvest Processes for Increased Agricultural Productivity | |
| CN117575174B (en) | Intelligent agricultural monitoring and management system | |
| CN116432849B (en) | Leaf area index prediction method based on primary productivity data assimilation | |
| Ashika et al. | A review of the contemporary soil monitoring systems in modern farming | |
| Hai et al. | Water resource management using remote sensing and coyote optimization algorithms | |
| O'Donncha et al. | A SWAT-based Reinforcement Learning Framework for Crop Management. | |
| Ahmed et al. | An Improved GA-Based Recommendation System for Soil Fertilization |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CP03 | Change of name, title or address |
Address after: No.99 Denglong Road, Fuzhou Economic and Technological Development Zone, Fujian Province 350015 Patentee after: YANGO University Patentee after: Guangze Hongxiang Intelligent Technology Co.,Ltd. Address before: Floor 6, No. 66, Guangming Avenue Middle Road, Guangze County, Nanping City, Fujian Province, 354199 Patentee before: Guangze Hongxiang Intelligent Technology Co.,Ltd. Patentee before: YANGO University |
|
| CP03 | Change of name, title or address |