CN115963243A

CN115963243A - Soil nitrogen and phosphorus nutrient loss prevention and control method and system based on nutrient dynamic monitoring

Info

Publication number: CN115963243A
Application number: CN202310031014.9A
Authority: CN
Inventors: 张雅蓉; 朱华清; 杨叶华; 熊涵; 李渝; 刘彦伶; 黄兴成
Original assignee: Guizhou Soil And Fertilizer Research Institute Guizhou Ecological Agricultural Engineering Technology Research Center Guizhou Agricultural Resources And Environment Research Institute
Current assignee: Guizhou Soil And Fertilizer Research Institute Guizhou Ecological Agricultural Engineering Technology Research Center Guizhou Agricultural Resources And Environment Research Institute
Priority date: 2023-01-10
Filing date: 2023-01-10
Publication date: 2023-04-14

Abstract

The invention discloses a soil nitrogen and phosphorus nutrient loss prevention and control method and system based on nutrient dynamic monitoring, which are used for acquiring soil physicochemical properties, topographic information and surface runoff information of a target area, analyzing and acquiring spatial distribution of soil nitrogen and phosphorus nutrients according to the soil physicochemical properties, matching meteorological conditions of the target area to acquire a time sequence change sequence of soil nitrogen and phosphorus nutrient content, acquiring nutrient loss characteristics, constructing a nutrient loss prediction model for predicting the soil nitrogen and phosphorus loss after preset time, and judging main influence factors of nutrient loss in a corresponding prediction time period; and determining a control scheme suitable for the target area according to the main influence factors through a related knowledge graph of nutrient loss control. According to the invention, the nutrient loss rule of the target area is mastered by dynamically monitoring the nutrients of the target area, and a targeted soil nitrogen and phosphorus nutrient loss scheme is formulated, so that the comprehensive cost of soil treatment is reduced, the control efficiency of soil nitrogen and phosphorus nutrient loss is increased, and the comprehensive benefit of the control scheme is optimal.

Description

Soil nitrogen and phosphorus nutrient loss prevention and control method and system based on nutrient dynamic monitoring

Technical Field

The invention relates to the technical field of soil nutrient monitoring, in particular to a soil nitrogen and phosphorus nutrient loss prevention and control method and system based on nutrient dynamic monitoring.

Background

The urgent demand of agricultural production on crop yield leads to the large input of resources such as water and fertilizer, the loss of nitrogen and phosphorus nutrients in farmland is greatly influenced by a plurality of factors such as rainfall, fertilization, terrain, soil property, vegetation coverage and farming mode, and the increase of the use amount of the fertilizer is obviously slowed down along with the continuous promotion of the reduction and efficiency enhancement work of the fertilizer. However, the application amount of the fertilizer is difficult to reduce in a short period, and simultaneously, due to the historical accumulation of a large amount of applied fertilizer, a large amount of adsorbed and retained nitrogen and phosphorus nutrients exist in farmlands, and the part of the nutrients which are difficult to activate can hardly be absorbed by crops, and are often lost through various ways such as loss and leaching loss. Although researchers do a lot of research on improving the utilization rate of nitrogen and phosphorus nutrients, reducing the loss of nitrogen and phosphorus nutrients in farmlands and the like, under the condition of high background value of soil nutrients, the loss amount of nitrogen and phosphorus nutrients applied to farmlands every year is still very large.

In recent years, agricultural non-point source pollution caused by the planting industry becomes one of important sources of water body pollutants in various regions and watersheds, excessive nitrogen and phosphorus nutrient loss of farmlands causes water body eutrophication, seriously harms the ecological environment, and most of the existing methods for blocking nitrogen and phosphorus nutrient loss are methods of additionally arranging ridge direction interception, optimizing fertilization and the like, but the implementation of prevention and control technology has great limitation due to factors such as land utilization, terrain and the like. Therefore, a specific soil nitrogen and phosphorus nutrient loss prevention and control scheme is formulated based on the conditions of climatic conditions, land utilization types, slopes and the like of the farmland locations, and the stable improvement of the soil quality is an urgent problem which cannot be solved at the present stage.

Disclosure of Invention

In order to solve at least one technical problem, the invention provides a method and a system for controlling the loss of nitrogen and phosphorus nutrients in soil based on dynamic nutrient monitoring.

The invention provides a soil nitrogen and phosphorus nutrient loss prevention and control method based on nutrient dynamic monitoring, which comprises the following steps:

acquiring soil physicochemical properties, topographic information and surface runoff information of a target area, analyzing and acquiring spatial distribution of nitrogen and phosphorus nutrients of soil according to the soil physicochemical properties, and performing visual representation on the spatial distribution of the nitrogen and phosphorus nutrients of the soil by combining the topographic information;

acquiring a time sequence change sequence of the nitrogen and phosphorus nutrient content of the soil according to the meteorological conditions of the soil nitrogen and phosphorus nutrient spatial distribution matching target area, acquiring nutrient loss characteristics through the time sequence change sequence, constructing a nutrient loss prediction model, and training by utilizing the nutrient loss characteristics;

acquiring the predicted loss amount of nitrogen and phosphorus of soil after preset time of a target area based on a nutrient loss prediction model, and judging main influence factors of nutrient loss in a corresponding prediction time period;

determining a resistance control method suitable for the target area according to main influence factors through a related knowledge graph of nutrient loss resistance control, and adjusting parameters of the resistance control method according to meteorological conditions, land utilization types, surface runoff information and topographic information of the target area to generate a resistance control scheme of the target area.

In the scheme, the soil physical and chemical properties, the topographic information and the surface runoff information of a target area are obtained, the spatial distribution of nitrogen and phosphorus nutrients in soil is obtained according to the analysis of the soil physical and chemical properties, and the spatial distribution of the nitrogen and phosphorus nutrients in the soil is visually represented by combining the topographic information, specifically:

acquiring soil physicochemical properties through soil monitoring information of a target area, and extracting soil nitrogen and phosphorus nutrient contents of monitoring points from the soil physicochemical properties of different monitoring points;

dividing a target area, acquiring vegetation coverage information, topographic information and surface runoff information in each sub-area, generating a characteristic value of each sub-area, and calculating the similarity of each sub-area based on the characteristic value;

if the similarity of any two subregions is larger than a preset similarity threshold, classifying the two subregions into the same type region, marking the subregions which do not contain monitoring points in the same type region, and obtaining the average value of the nitrogen and phosphorus nutrient contents of the soil in the same type region as the nitrogen and phosphorus nutrient contents of the soil of the marking subregion;

and generating a three-dimensional terrain model according to the terrain information of the target area, and visually marking the content of nitrogen and phosphorus nutrients in the soil of each subarea according to the depth of the soil layer to obtain the spatial distribution of the nitrogen and phosphorus nutrients in the soil.

In the scheme, a time sequence change sequence of the nitrogen and phosphorus nutrient content of the soil is obtained according to meteorological conditions of a soil nitrogen and phosphorus nutrient space distribution matching target area, nutrient loss characteristics are obtained through the time sequence change sequence, a nutrient loss prediction model is constructed, and the nutrient loss characteristics are utilized for training, and the method specifically comprises the following steps:

determining the initial soil nitrogen and phosphorus nutrient content of a target area by combining the spatial distribution of the soil nitrogen and phosphorus nutrients with the fertilizing amount and the fertilizing timestamp of the target area, and acquiring various existing forms of the soil nitrogen and phosphorus nutrients in the target area according to the initial soil nitrogen and phosphorus nutrient content;

acquiring the change of the runoff rate of the earth surface according to the meteorological conditions of a target area, acquiring the content change of various existing forms of the nitrogen and phosphorus nutrients of the soil according to the change of the runoff rate of the earth surface, and acquiring the existing form with the largest content change in the various existing forms of the nitrogen and phosphorus nutrients of the soil;

calculating the difference value of the initial nitrogen and phosphorus nutrient content of the soil corresponding to the existing form with the maximum change of the nitrogen and phosphorus nutrient content of the soil and the nitrogen and phosphorus nutrient content of the surface runoff, and generating a time sequence change sequence of the nitrogen and phosphorus nutrient content of the soil according to the difference value and the timestamps;

establishing a nutrient loss prediction model by combining a neural network model optimized by a genetic algorithm with an LSTM model, learning and extracting the change rules of the nitrogen and phosphorus nutrient content of soil, meteorological conditions, vegetation coverage information and topographic information in a target area by the neural network optimized by the genetic algorithm, and generating a time sequence characteristic sequence of the change of the nitrogen and phosphorus nutrient content of the soil;

and inputting the time sequence characteristic sequence of the content change of the nitrogen and phosphorus nutrients of the soil into an LSTM model to analyze and learn the time sequence, and predicting the loss of the nitrogen and phosphorus nutrients of the soil after preset time.

In the scheme, main influence factors corresponding to nutrient loss in the prediction time period are judged, and the main influence factors are specifically as follows:

analyzing and extracting influence factors of nitrogen and phosphorus nutrient loss based on historical nitrogen and phosphorus nutrient loss conditions of the target area in combination with big data;

evaluating the significance difference of the influence factors on the reaction of the nitrogen and phosphorus nutrient loss to obtain corresponding probability values, and obtaining the importance indexes of the influence factors according to the probability values and the data sample number of the historical nitrogen and phosphorus nutrient loss conditions;

generating normal distribution through the importance index of the influence factors to obtain the weight value of the influence factors, and judging the deviation rate of the predicted loss of the nitrogen and phosphorus nutrients of the soil after preset time and historical synchronization data;

if the deviation ratio is larger than a preset deviation ratio threshold value, acquiring an influence factor with the maximum deviation between corresponding prediction data and historical reference data in all influence factors of a target area in prediction time as a main influence factor;

and if the deviation rate is not greater than a preset deviation rate threshold value, selecting main influence factors according to the sorting of the weighted values.

In the scheme, a resistance control method suitable for a target area is determined according to main influence factors through a related knowledge graph of nutrient loss resistance control, and the method specifically comprises the following steps:

acquiring a relevant knowledge map of nutrient loss control by utilizing big data retrieval, constructing a knowledge base as a data source according to the relevant knowledge map of nutrient loss control, and establishing a scoring matrix of a land and nutrient loss control method based on the data source;

acquiring a historical nutrient loss prevention and control method used by a target area level according to the applicability degree of the grading matrix representing the nutrient loss prevention and control method of the land, and setting the initial weight of a corresponding knowledge node in a knowledge base according to the grading matrix of the historical nutrient loss prevention and control method of the target area;

calculating the similarity between the target area and the land in the knowledge graph by using the main influence factors of the area where the land is located in the knowledge graph and the scoring matrix, and determining a target node of the target area in the knowledge graph according to the similarity;

acquiring a triple with a target node as a head node in a knowledge graph to generate a neighbor node set, and performing aggregation representation on node information in the triple set through a graph convolution neural network;

an attention mechanism is introduced, initial weights are combined, vectorization representation after target node dynamic aggregation is obtained according to an aggregation mechanism, a plurality of feature vectorization representations of the target nodes are obtained according to attention information of different levels and are fused, and target node vectorization representation with neighbor node description is generated;

meanwhile, acquiring the vectorization representation of the nutrient loss prevention and control method nodes according to the knowledge graph, calculating the inner product of the vectorization representation of the target nodes and the vectorization representation of the nutrient loss prevention and control method nodes, and sequencing according to the inner product calculation result;

and obtaining a preset number of nutrient loss prevention and control methods through the sequencing result to serve as nutrient loss prevention and control methods suitable for the target area.

In the scheme, parameters of the resistance control method are adjusted according to the meteorological conditions, the land utilization type, the surface runoff information and the topographic information of the target area to generate a resistance control scheme of the target area, which specifically comprises the following steps:

acquiring a nutrient loss prevention and control method suitable for a target area, generating a description feature set of the target area according to meteorological conditions, land utilization types, surface runoff information and topographic information of the target area, and constructing a feature description matrix based on feature values of all description features in the description feature set;

matching each nutrient loss control method according to the characteristic description matrix to perform data retrieval, and performing correlation calculation in a retrieval space through the characteristic description matrix to obtain retrieval data of which the correlation meets a preset standard under each nutrient loss control label;

acquiring a high-frequency parameter interval corresponding to retrieval data under each nutrient loss prevention and control label through data statistics, and generating an average parameter based on parameter data in the high-frequency parameter interval to serve as parameter information of the nutrient loss prevention and control method;

and distributing parameter information under each nutrient loss prevention and control method, and screening and combining the nutrient loss prevention and control methods after the parameters are determined according to the method feasibility of the target area to generate a nutrient loss prevention and control scheme of the target area.

The invention also provides a system for controlling the loss of nitrogen and phosphorus nutrients in soil based on dynamic nutrient monitoring, which comprises: the method comprises a storage and a processor, wherein the storage comprises a program of a soil nitrogen and phosphorus nutrient loss prevention and control method based on nutrient dynamic monitoring, and when the program of the soil nitrogen and phosphorus nutrient loss prevention and control method based on nutrient dynamic monitoring is executed by the processor, the following steps are realized:

acquiring a time sequence change sequence of the nitrogen and phosphorus nutrient content of the soil according to the spatial distribution of the nitrogen and phosphorus nutrients of the soil and the meteorological conditions of a target area, acquiring nutrient loss characteristics through the time sequence change sequence, constructing a nutrient loss prediction model, and training by using the nutrient loss characteristics;

The invention discloses a soil nitrogen and phosphorus nutrient loss prevention and control method and system based on nutrient dynamic monitoring, which are used for acquiring soil physicochemical properties, topographic information and surface runoff information of a target area, analyzing and acquiring spatial distribution of soil nitrogen and phosphorus nutrients according to the soil physicochemical properties, matching meteorological conditions of the target area to acquire a time sequence change sequence of soil nitrogen and phosphorus nutrient content, acquiring nutrient loss characteristics, constructing a nutrient loss prediction model for predicting the soil nitrogen and phosphorus loss after preset time, and judging main influence factors of nutrient loss in a corresponding prediction time period; and determining a control scheme suitable for the target area according to the main influence factors through a related knowledge graph of nutrient loss control. According to the invention, the nutrient loss rule of the target area is dynamically monitored and mastered by the nutrient dynamic monitoring system, and a specific soil nitrogen and phosphorus nutrient loss scheme is formulated, so that the comprehensive cost of soil treatment is reduced, the control inhibition efficiency of soil nitrogen and phosphorus nutrient loss is increased, and the comprehensive benefit of the control inhibition scheme is optimal.

Drawings

FIG. 1 shows a flow chart of a method for controlling the loss of nitrogen and phosphorus nutrients in soil based on dynamic nutrient monitoring according to the invention;

FIG. 2 shows a flow chart of a method for predicting nitrogen and phosphorus nutrient loss by constructing a nutrient loss prediction model in the invention;

FIG. 3 is a flow chart of a method for determining a blocking method applicable to a target area according to main influence factors through a knowledge graph in the present invention;

FIG. 4 shows a block diagram of a soil nitrogen and phosphorus nutrient loss prevention and control system based on nutrient dynamic monitoring.

Detailed Description

In order that the above objects, features and advantages of the present invention can be more clearly understood, a more particular description of the invention, taken in conjunction with the accompanying drawings and detailed description, is set forth below. It should be noted that the embodiments and features of the embodiments of the present application may be combined with each other without conflict.

In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention, however, the present invention may be practiced otherwise than as specifically described herein and, therefore, the scope of the present invention is not limited by the specific embodiments disclosed below.

FIG. 1 shows a flow chart of a method for controlling the loss of nitrogen and phosphorus nutrients in soil based on dynamic nutrient monitoring.

As shown in fig. 1, a first aspect of the present invention provides a method for controlling soil nitrogen and phosphorus nutrient loss based on dynamic nutrient monitoring, including:

s102, acquiring soil physicochemical properties, topographic information and surface runoff information of a target area, analyzing and acquiring spatial distribution of nitrogen and phosphorus nutrients of soil according to the soil physicochemical properties, and performing visual representation on the spatial distribution of the nitrogen and phosphorus nutrients of the soil by combining topographic information;

s104, acquiring a time sequence change sequence of the nitrogen and phosphorus nutrient content of the soil according to the meteorological conditions of the soil nitrogen and phosphorus nutrient spatial distribution matching target area, acquiring nutrient loss characteristics through the time sequence change sequence, constructing a nutrient loss prediction model, and training by using the nutrient loss characteristics;

s106, acquiring the predicted loss amount of nitrogen and phosphorus of soil after preset time of a target area based on a nutrient loss prediction model, and judging main influence factors of nutrient loss in a corresponding prediction time period;

s108, determining a resistance control method suitable for the target area according to the main influence factors through the relevant knowledge graph of nutrient loss resistance control, and adjusting parameters of the resistance control method according to meteorological conditions, land utilization types, surface runoff information and topographic information of the target area to generate a resistance control scheme of the target area.

The method includes the steps that soil physical and chemical properties are obtained through soil monitoring information of a target area, soil nitrogen and phosphorus nutrient contents of monitoring points are extracted from the soil physical and chemical properties of different monitoring points, and the soil physical and chemical properties include one or more of soil nutrient content, soil organic matter content, soil salt content, soil heavy metal pollutant content, soil structural information, soil temperature information, soil humidity information and soil pH information; dividing a target area, acquiring vegetation coverage information, topographic information and surface runoff information in each sub-area, generating a characteristic value of each sub-area, and calculating the similarity of any two sub-areas in the target area based on the characteristic value; if the similarity of any two subregions is larger than a preset similarity threshold, classifying the two subregions into the same type region, marking the subregions which do not contain monitoring points in the same type region, and acquiring the average value of the nitrogen and phosphorus nutrient contents of the soil detected by the monitoring points in the type region as the nitrogen and phosphorus nutrient contents of the soil of the marking subregion; and generating a three-dimensional terrain model according to the terrain information of the target area, and visually marking the content of nitrogen and phosphorus nutrients in the soil of each subarea according to the depth of the soil layer to obtain the spatial distribution of the nitrogen and phosphorus nutrients in the soil.

FIG. 2 shows a flow chart of a method for predicting loss of nitrogen and phosphorus nutrients by constructing a nutrient loss prediction model in the invention.

According to the embodiment of the invention, a time sequence change sequence of the nitrogen and phosphorus nutrient content of the soil is obtained according to the meteorological condition that the spatial distribution of the nitrogen and phosphorus nutrients of the soil is matched with a target area, the nutrient loss characteristic is obtained through the time sequence change sequence, a nutrient loss prediction model is constructed, and the nutrient loss characteristic is used for training, and the method specifically comprises the following steps:

s202, determining the initial content of nitrogen and phosphorus nutrients in soil in a target area by combining the spatial distribution of the nitrogen and phosphorus nutrients in the soil with the fertilizing amount and the fertilizing timestamp in the target area, and acquiring various existing forms of the nitrogen and phosphorus nutrients in the soil in the target area according to the initial content of the nitrogen and phosphorus nutrients in the soil;

s204, acquiring the surface runoff change according to the meteorological conditions of a target area, acquiring the content change of various existing forms of soil nitrogen and phosphorus nutrients according to the surface runoff change, and acquiring the existing form with the largest content change in the various existing forms of the soil nitrogen and phosphorus nutrients;

s206, calculating the difference value of the initial soil nitrogen and phosphorus nutrient content corresponding to the existing form with the maximum change of the soil nitrogen and phosphorus nutrient content and the nitrogen and phosphorus nutrient content in surface runoff, and generating a time sequence change sequence of the soil nitrogen and phosphorus nutrient content according to the difference value and all timestamps;

s208, a nutrient loss prediction model is constructed by combining the neural network model optimized through the genetic algorithm with the LSTM model, and the change rule of the nitrogen and phosphorus nutrient content of soil in the target area, the meteorological condition, the vegetation coverage information and the topographic information is learned and extracted through the neural network optimized through the genetic algorithm to generate a time sequence characteristic sequence of the change of the nitrogen and phosphorus nutrient content of the soil;

s210, inputting the time sequence characteristic sequence of the content change of the nitrogen and phosphorus nutrients in the soil into an LSTM model to analyze and learn the time sequence, and predicting the loss of the nitrogen and phosphorus nutrients in the soil after preset time.

It should be noted that, the optimal parameter value is determined for the neural network model through the genetic algorithm, and the accuracy precision of the model is used as the fitness of the genetic algorithm; initializing chromosome population parameters, adopting binary coding, calculating individual fitness, selecting through iterative training, crossing, and mutating to the maximum iteration times to obtain chromosome individuals corresponding to the highest accuracy and precision; decoding chromosome individuals corresponding to the highest accuracy to obtain the number of layers of a neural network model and the number of neurons in each layer, extracting a nutrient loss characteristic sequence through the neural network model optimized by a genetic algorithm, carrying out standardization processing, inputting the nutrient loss characteristic sequence into an LSTM network, controlling a transmission state by the LSTM network mainly through a forgetting gate, a memory gate and an output gate, updating the state and activating a corresponding mapping relation by a gate structure in a specific mode, sequentially inputting each characteristic vector into the LSTM network, combining the output of a first neuron and the output of a last neuron as an output vector, and outputting the predicted loss of nitrogen and phosphorus nutrients in soil after preset time through a full connection layer

The main influence factors of nutrient loss in the corresponding prediction time period are judged, and the influence factors of nitrogen and phosphorus nutrient loss are extracted based on the historical nitrogen and phosphorus nutrient loss condition of the target area and big data analysis; evaluating the significance difference of the influence factors on the reaction of the nitrogen and phosphorus nutrient loss to obtain corresponding probability values, and obtaining the importance indexes of the influence factors according to the probability values and the data sample number of the historical nitrogen and phosphorus nutrient loss conditions; generating normal distribution through the importance index of the influence factors to obtain the weight value of the influence factors, and judging the deviation rate of the predicted loss of the nitrogen and phosphorus nutrients of the soil after preset time and historical synchronization data; if the deviation ratio is larger than a preset deviation ratio threshold value, acquiring an influence factor with the maximum deviation between corresponding prediction data and historical reference data in all influence factors of a target area in prediction time as a main influence factor; and if the deviation rate is not greater than a preset deviation rate threshold value, selecting main influence factors according to the sorting of the weighted values.

FIG. 3 is a flow chart of a method for determining a blocking control method suitable for a target area according to main influence factors through a knowledge graph.

According to the embodiment of the invention, the resistance control method suitable for the target area is determined according to main influence factors through the related knowledge graph of nutrient loss resistance control, and specifically comprises the following steps:

s302, retrieving and acquiring a relevant knowledge graph of nutrient loss prevention and control by utilizing big data, constructing a knowledge base as a data source according to the relevant knowledge graph of nutrient loss prevention and control, and establishing a scoring matrix of a land and nutrient loss prevention and control method based on the data source;

s304, obtaining a historical nutrient loss prevention and control method used by a target area level according to the applicability degree of the grading matrix representing the soil to the nutrient loss prevention and control method, and setting the initial weight of a corresponding knowledge node in a knowledge base according to the grading matrix of the target area to the historical nutrient loss prevention and control method;

s306, calculating the similarity between the target area and the land in the knowledge graph by using the main influence factors of the area where the land is located in the knowledge graph and the scoring matrix, and determining a target node of the target area in the knowledge graph according to the similarity;

s308, acquiring triples with target nodes as head nodes in the knowledge graph to generate a neighbor node set, and performing aggregation representation on node information in the triples through a graph convolution neural network;

s310, an attention mechanism is introduced, initial weights are combined, vectorization representation after target node dynamic aggregation is obtained according to an aggregation mechanism, a plurality of feature vectorization representations of the target nodes are obtained according to attention information of different levels and are fused, and target node vectorization representation with neighbor node description is generated;

s312, simultaneously, acquiring nutrient loss resistance control method node vectorization representation according to the knowledge graph, calculating the inner product of the vectorization representation of the target node and the nutrient loss resistance control method node vectorization representation, and sequencing according to the inner product calculation result;

and S314, obtaining a preset number of nutrient loss prevention and control methods through the sequencing result to serve as nutrient loss prevention and control methods suitable for the target area.

It should be noted that the target node N is used as the triplet of the head node to generate the neighbor node set N _n Is defined as N _n The method comprises the following steps of = (n, g, m), wherein n is a head node in a knowledge graph, g is a node relation, m is a tail node in the knowledge graph, a target node is selected from land nodes in the knowledge graph, an attention mechanism is introduced, an initial weight is combined, and vectorization representation after dynamic aggregation of the target node is obtained according to an aggregation mechanism, and the method specifically comprises the following steps:

wherein, the first and the second end of the pipe are connected with each other,

representing vectorization representation after dynamic aggregation of target nodes, LRelu representing aggregation function, w and b representing aggregation weight and bias respectively, adjusting convolution according to different attention weights of neighbor nodes to the target nodes through attention mechanism setting, R representing initial weight matrix, e _m Representing the initial vectorization representation of a single tail node, namely a nutrient loss resistance control method node, wherein T represents matrix transposition, and k represents the number of convolution layers; the attention weight of the attention mechanism is obtained through a nonlinear activation function Relu training weight matrix parameter and an attention parameter vector;

obtaining a plurality of characteristic vectorization representations of the target node according to the attention information of different levels, fusing the characteristic vectorization representations to generate vectorization representations of the target node after characteristic fusion, generating vectorization representations of the nutrient loss prevention and control method nodes in the same way, and obtaining an inner product calculation result, wherein the inner product f is calculated as follows:

vectorized representation of a target node after feature fusion is represented, based on the results of the feature fusion>

And expressing vectorization expression of the nutrient loss prevention and control method node after feature fusion.

It should be noted that the nutrient loss prevention and control method suitable for the target area is obtained, a description feature set of the target area is generated according to the meteorological conditions, the land utilization type, the surface runoff information and the topographic information of the target area, and a feature description matrix is constructed on the basis of feature values of all description features in the description feature set; matching each nutrient loss control method according to the characteristic description matrix to perform data retrieval, and performing correlation calculation in a retrieval space according to the characteristic description matrix by using a Pearson correlation coefficient to obtain retrieval data of which the correlation meets a preset standard under each nutrient loss control label; obtaining high-frequency parameter intervals corresponding to retrieval data under each nutrient loss control label through data statistics, and generating average parameters based on parameter data in the high-frequency parameter intervals to serve as parameter information of the nutrient loss control method; and distributing parameter information under each nutrient loss prevention and control method, and screening and combining the nutrient loss prevention and control methods after the parameters are determined according to the method feasibility of the target area to generate a nutrient loss prevention and control scheme of the target area, wherein the assessment of the feasibility is comprehensively judged based on indexes such as implementation cost, implementation difficulty and the existing nutrient loss prevention and control facility condition of the target area.

According to the embodiment of the invention, an exclusive data set is established according to the change characteristics of each item of basic parameter information of the target area, and the method specifically comprises the following steps:

monitoring nitrogen and phosphorus nutrient loss and meteorological conditions, vegetation coverage types, topographic information and surface runoff self-centering of a target area, and generating a dynamic environment characteristic set in a monitoring stage by combining time characteristics;

acquiring nutrient loss prevention and control schemes corresponding to various environmental characteristics based on the matching of the dynamic environmental characteristic set and the nutrient loss prevention and control scheme of the target area, and acquiring the use frequency of the same nutrient loss prevention and control method under different environmental characteristics according to data analysis;

acquiring preference characteristics of a target area for nutrient loss control according to the use frequency of a nutrient loss control method, updating the initial weight of a corresponding knowledge node in a knowledge base according to the preference characteristics, and constructing a proprietary database according to a dynamic environment characteristic set of the target area and the preference characteristics of the nutrient loss control

When the nutrient loss prevention and control is carried out on the target area, acquiring the characteristic deviation of the current environmental characteristics of the target area and the characteristic data in the dynamic environmental characteristic set, and selecting a nutrient loss prevention and control method according to the characteristic deviation and the preference characteristics;

when the nutrient loss prevention and control method generated by the current characteristic environment of the target environment does not accord with the preference characteristic, marking and storing the corresponding environment characteristic;

in addition, feedback data in the actual implementation process of the nutrient loss prevention and control scheme of the target area are obtained, and the environment characteristics marked and stored by the feedback data set are utilized to update the data of the exclusive database, so that the preference characteristics are optimized.

The invention also provides a system 4 for controlling the soil nitrogen and phosphorus nutrient loss based on nutrient dynamic monitoring, which comprises: the system comprises a memory 41 and a processor 42, wherein the memory comprises a program of a soil nitrogen and phosphorus nutrient loss prevention and control method based on nutrient dynamic monitoring, and when the program of the soil nitrogen and phosphorus nutrient loss prevention and control method based on nutrient dynamic monitoring is executed by the processor, the following steps are realized:

The method includes the steps that soil physical and chemical properties are obtained through soil monitoring information of a target area, soil nitrogen and phosphorus nutrient contents of monitoring points are extracted from the soil physical and chemical properties of different monitoring points, and the soil physical and chemical properties include one or more of soil nutrient content, soil organic matter content, soil salt content, soil heavy metal pollutant content, soil structural information, soil temperature information, soil humidity information and soil pH information; dividing a target area, acquiring vegetation coverage information, topographic information and surface runoff information in each subarea, generating characteristic values of each subarea, and calculating the similarity of any two subareas in the target area based on the characteristic values; if the similarity of any two sub-areas is larger than a preset similarity threshold, classifying the two sub-areas into the same type of area, marking the sub-areas which do not contain monitoring points in the same type of area, and acquiring the mean value of the nitrogen and phosphorus nutrient contents of the soil detected by the monitoring points in the type of area as the nitrogen and phosphorus nutrient contents of the soil marked by the sub-areas; and generating a three-dimensional terrain model according to the terrain information of the target area, and visually marking the content of nitrogen and phosphorus nutrients in the soil of each subarea according to the depth of the soil layer to obtain the spatial distribution of the nitrogen and phosphorus nutrients in the soil.

determining the initial content of nitrogen and phosphorus nutrients in soil in a target area by combining the spatial distribution of the nitrogen and phosphorus nutrients in the soil with the fertilizing amount and the fertilizing timestamp of the target area, and acquiring various existing forms of the nitrogen and phosphorus nutrients in the soil in the target area according to the initial content of the nitrogen and phosphorus nutrients in the soil;

acquiring earth surface runoff change according to meteorological conditions of a target area, acquiring content change of various existing forms of the nitrogen and phosphorus nutrients of the soil according to the earth surface runoff change, and acquiring the existing form with the largest content change in the various existing forms of the nitrogen and phosphorus nutrients of the soil;

establishing a nutrient loss prediction model by combining a neural network model optimized by a genetic algorithm with an LSTM model, learning and extracting the change rules of the nitrogen and phosphorus nutrient content of soil in a target area, meteorological conditions, vegetation coverage information and topographic information by the neural network optimized by the genetic algorithm, and generating a time sequence characteristic sequence of the change of the nitrogen and phosphorus nutrient content of the soil;

and inputting the time sequence characteristic sequence of the content change of the nitrogen and phosphorus nutrients in the soil into an LSTM model to analyze and learn the time sequence, and predicting the loss of the nitrogen and phosphorus nutrients in the soil after a preset time.

The method comprises the steps of judging main influence factors corresponding to nutrient loss in a prediction time period, and analyzing and extracting the influence factors of nitrogen and phosphorus nutrient loss based on historical nitrogen and phosphorus nutrient loss conditions of a target area and big data; evaluating the significance difference of the influence factors on the reaction of the nitrogen and phosphorus nutrient loss to obtain corresponding probability values, and obtaining the importance indexes of the influence factors according to the probability values and the data sample number of the historical nitrogen and phosphorus nutrient loss conditions; generating normal distribution through the importance index of the influence factors to obtain the weight values of the influence factors, and judging the deviation rate of the predicted loss amount of the nitrogen and phosphorus nutrients of the soil after preset time and historical synchronization data; if the deviation ratio is larger than a preset deviation ratio threshold value, acquiring an influence factor with the maximum deviation between corresponding prediction data and historical reference data in all influence factors of a target area in prediction time as a main influence factor; and if the deviation rate is not greater than a preset deviation rate threshold value, selecting main influence factors according to the sorting of the weighted values.

acquiring a relevant knowledge map of nutrient loss prevention and control by utilizing big data retrieval, constructing a knowledge base as a data source according to the relevant knowledge map of nutrient loss prevention and control, and establishing a scoring matrix of a land and nutrient loss prevention and control method based on the data source;

calculating the similarity between a target area and the land in the knowledge graph by using the main influence factors of the area where the land is located in the knowledge graph and the scoring matrix, and determining a target node of the target area in the knowledge graph according to the similarity;

acquiring triples with target nodes as head nodes in a knowledge graph to generate a neighbor node set, and performing aggregation representation on node information in the triplet set through a graph convolution neural network;

meanwhile, acquiring nutrient loss prevention and control method node vectorization representation according to the knowledge graph, calculating the inner product of the vectorization representation of the target node and the nutrient loss prevention and control method node vectorization representation, and sequencing according to the inner product calculation result;

It should be noted that the target node N is used as the triplet of the head node to generate the neighbor node set N _n Is defined as N _n = (n, g, m), n is head node in the knowledge graph, g is node relation, m is tail node in the knowledge graphIdentifying land nodes in the map, selecting target nodes, introducing an attention mechanism and combining initial weights to obtain vectorization representation after dynamic aggregation of the target nodes according to an aggregation mechanism, and specifically comprising the following steps:

wherein the content of the first and second substances,

representing vectorization representation after dynamic aggregation of target nodes, LRelu representing aggregation function, w and b representing aggregation weight and bias respectively, adjusting convolution according to different attention weights of neighbor nodes to the target nodes through attention mechanism setting, R representing initial weight matrix, e _m Representing the initial vectorization representation of a single tail node, namely a nutrient loss resistance control method node, wherein T represents matrix transposition, and k represents the number of convolution layers;

wherein the content of the first and second substances,

vectorized representation of a target node with fused representation features, based on the feature fusion>

And expressing the vectorization expression of the nutrient loss prevention and control method node after the characteristic fusion.

It should be noted that the nutrient loss prevention and control method suitable for the target area is obtained, a description feature set of the target area is generated according to the meteorological conditions, the land utilization type, the surface runoff information and the topographic information of the target area, and a feature description matrix is constructed on the basis of feature values of all description features in the description feature set; matching each nutrient loss control method according to the feature description matrix to perform data retrieval, and performing correlation calculation in a retrieval space according to the feature description matrix by using a Pearson correlation coefficient to obtain retrieval data of which the correlation meets a preset standard under each nutrient loss control label; acquiring a high-frequency parameter interval corresponding to retrieval data under each nutrient loss prevention and control label through data statistics, and generating an average parameter based on parameter data in the high-frequency parameter interval to serve as parameter information of the nutrient loss prevention and control method; and distributing parameter information under each nutrient loss prevention and control method, screening and combining the nutrient loss prevention and control methods after the parameters are determined according to the method feasibility of the target area, and generating a nutrient loss prevention and control scheme of the target area, wherein the evaluation of the feasibility is comprehensively judged based on indexes such as implementation cost, implementation difficulty and the condition of the existing nutrient loss prevention and control facilities of the target area.

In the several embodiments provided in the present application, it should be understood that the disclosed apparatus and method may be implemented in other ways. The above-described device embodiments are merely illustrative, for example, the division of the unit is only a logical functional division, and there may be other division ways in actual implementation, such as: multiple units or components may be combined, or may be integrated into another system, or some features may be omitted, or not implemented. In addition, the coupling, direct coupling or communication connection between the components shown or discussed may be through some interfaces, and the indirect coupling or communication connection between the devices or units may be electrical, mechanical or in other forms.

The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units; can be located in one place or distributed on a plurality of network units; some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.

In addition, all the functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may be separately regarded as one unit, or two or more units may be integrated into one unit; the integrated unit can be realized in a form of hardware, or in a form of hardware plus a software functional unit.

Those of ordinary skill in the art will understand that: all or part of the steps for realizing the method embodiments can be completed by hardware related to program instructions, the program can be stored in a computer readable storage medium, and the program executes the steps comprising the method embodiments when executed; and the aforementioned storage medium includes: a mobile storage device, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and various media capable of storing program codes.

Alternatively, the integrated unit of the present invention may be stored in a computer-readable storage medium if it is implemented in the form of a software functional module and sold or used as a separate product. Based on such understanding, the technical solutions of the embodiments of the present invention may be essentially implemented or a part contributing to the prior art may be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the methods described in the embodiments of the present invention. And the aforementioned storage medium includes: a removable storage device, a ROM, a RAM, a magnetic or optical disk, or various other media that can store program code.

The above description is only for the specific embodiments of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art can easily conceive of the changes or substitutions within the technical scope of the present invention, and all the changes or substitutions should be covered within the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the appended claims.

Claims

1. A soil nitrogen and phosphorus nutrient loss prevention and control method based on nutrient dynamic monitoring is characterized by comprising the following steps:

2. The method for preventing and controlling the loss of the nitrogen and phosphorus nutrients in the soil based on the dynamic nutrient monitoring, according to claim 1, is characterized by obtaining soil physicochemical properties, topographic information and surface runoff information of a target area, analyzing and obtaining the spatial distribution of the nitrogen and phosphorus nutrients in the soil according to the soil physicochemical properties, and performing visual representation on the spatial distribution of the nitrogen and phosphorus nutrients in the soil by combining the topographic information, and specifically comprises the following steps:

3. The method for preventing and controlling the loss of nitrogen and phosphorus nutrients in soil based on dynamic nutrient monitoring of claim 1, wherein a time sequence variation sequence of nitrogen and phosphorus nutrient content in soil is obtained according to a meteorological condition that spatial distribution of nitrogen and phosphorus nutrients in soil is matched with a target area, nutrient loss characteristics are obtained through the time sequence variation sequence, a nutrient loss prediction model is constructed, and training is performed by using the nutrient loss characteristics, and the method specifically comprises the following steps:

calculating the difference value of the initial soil nitrogen and phosphorus nutrient content and the nitrogen and phosphorus nutrient content in surface runoff corresponding to the existence form with the maximum change of the soil nitrogen and phosphorus nutrient content, and generating a time sequence change sequence of the soil nitrogen and phosphorus nutrient content according to the difference value and the timestamps;

4. The method for preventing and controlling the loss of nitrogen and phosphorus nutrients in soil based on nutrient dynamic monitoring as claimed in claim 1, wherein the main influencing factors of nutrient loss in the corresponding prediction time period are judged, and specifically:

and if the deviation rate is not greater than a preset deviation rate threshold value, selecting main influence factors according to the sequence of the weight values.

5. The method for preventing and controlling the loss of nitrogen and phosphorus nutrients in soil based on dynamic nutrient monitoring of claim 1, wherein the method for preventing and controlling the loss of nitrogen and phosphorus nutrients in the soil is determined according to main influence factors through a knowledge graph related to nutrient loss prevention and control, and specifically comprises the following steps:

6. The method for controlling the loss of nitrogen, phosphorus and nutrients in soil based on dynamic nutrient monitoring as claimed in claim 1, wherein parameters of the method for controlling the loss of nitrogen, phosphorus and nutrients in soil are adjusted according to meteorological conditions, land utilization types, surface runoff information and topographic information of a target area to generate a control scheme of the target area, and specifically comprises the following steps:

matching each nutrient loss control method according to the feature description matrix to perform data retrieval, and performing correlation calculation in a retrieval space through the feature description matrix to obtain retrieval data of which the correlation meets a preset standard under each nutrient loss control label;

7. The utility model provides a soil nitrogen phosphorus nutrient loss control system based on nutrient dynamic monitoring which characterized in that, this system includes: the method comprises a storage and a processor, wherein the storage comprises a program of a soil nitrogen and phosphorus nutrient loss prevention and control method based on nutrient dynamic monitoring, and the program of the soil nitrogen and phosphorus nutrient loss prevention and control method based on nutrient dynamic monitoring realizes the following steps when being executed by the processor:

8. The system according to claim 7, wherein a time sequence variation sequence of the nitrogen and phosphorus nutrient content in the soil is obtained according to a meteorological condition that spatial distribution of nitrogen and phosphorus nutrients in the soil is matched with a target area, nutrient loss characteristics are obtained through the time sequence variation sequence, a nutrient loss prediction model is constructed, and training is performed by using the nutrient loss characteristics, specifically:

9. The soil nitrogen and phosphorus nutrient loss control system based on nutrient dynamic monitoring of claim 7, which is characterized in that a control method suitable for a target area is determined according to main influence factors through a nutrient loss control related knowledge graph, and specifically comprises the following steps:

10. The soil nitrogen and phosphorus nutrient loss prevention and control system based on nutrient dynamic monitoring of claim 7, characterized in that parameters of the prevention and control method are adjusted according to meteorological conditions, land utilization types, surface runoff information and topographic information of a target area to generate a prevention and control scheme of the target area, and the prevention and control scheme specifically comprises the following steps: