CN115775042B - Crop heavy metal enrichment risk prediction method and system based on Bayesian theory - Google Patents
Crop heavy metal enrichment risk prediction method and system based on Bayesian theory Download PDFInfo
- Publication number
- CN115775042B CN115775042B CN202211410742.2A CN202211410742A CN115775042B CN 115775042 B CN115775042 B CN 115775042B CN 202211410742 A CN202211410742 A CN 202211410742A CN 115775042 B CN115775042 B CN 115775042B
- Authority
- CN
- China
- Prior art keywords
- heavy metal
- soil
- model
- crops
- optimizing
- 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
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 discloses a method and a system for predicting heavy metal enrichment risk of crops based on Bayesian theory, which relate to the technical field of precision agriculture and comprise the following steps: acquiring information of basic physicochemical properties of regional soil, heavy metal content of soil and heavy metal content of crops, and constructing an observation data set; establishing a classification regression tree model, setting pruning rules, and obtaining an optimized classification regression tree model; extracting key influencing variables in the heavy metal enrichment process of crops, and analyzing a dynamic interaction process; constructing and optimizing a Bayesian prediction model based on the mixed random variables to obtain an optimized Bayesian prediction model; and performing multi-scene simulation based on the optimized Bayesian prediction model, and optimizing the heavy metal pollution prevention and control countermeasures in the region. The method can reveal the dynamic interaction characteristics of a plurality of soil factors in the heavy metal enrichment process of crops, predict the heavy metal enrichment trend and the exceeding risk of the crops, develop multi-scene simulation and optimize the current pollution source management and control measures.
Description
Technical Field
The invention relates to the technical field of precise agriculture, in particular to a method and a system for predicting heavy metal enrichment risk of crops based on Bayesian theory.
Background
Heavy metals (cadmium, lead, mercury and the like) are nonessential elements of organisms, have high toxicity and high mobility, are easy to enrich in crops (rice, wheat, vegetables and the like), and are harmful to human health through food chains. The long-term consumption of crops with excessive heavy metals can cause various diseases such as hypertension, osteoporosis, renal failure and the like. Revealing the heavy metal enrichment trend of crops and evaluating the heavy metal enrichment risk are key steps for guaranteeing grain safety and improving the health state of soil.
After entering farmland soil, exogenous heavy metals are absorbed by crops in the form of hydrated ions, complex inorganic matters or organic compounds through the processes of adsorption, exchange, dissolution, precipitation and the like. These processes are affected by factors such as soil pH, soil redox potential, soil organic matter content, soil texture and co-existing anions and cations, which exhibit dynamic interactions in the crop heavy metal accumulation process. Because the relationship has the characteristics of low strength, long period, complex influence factors and the like, the dynamic interaction relationship of different soil factors is difficult to accurately quantify by conventional experimental observation, and the relationship is a key and difficult problem for preventing and controlling heavy metal pollution of current crops.
At present, research on heavy metal enrichment risk of crops is concentrated on index evaluation methods such as an enrichment factor method, a potential ecological risk evaluation method and a biological evaluation method, the evaluation methods are applied to the evaluation current situation, the safety risk of agricultural products cannot be predicted, and effective suggestions are provided; meanwhile, the evaluation method ignores consideration of a plurality of soil physicochemical property variables, and the application range of the evaluation result is smaller. Model research on the heavy metal enrichment process of crops at home and abroad is carried out earlier, and an empirical model based on a multiple regression analysis method and a mechanism model based on process simulation are taken as main materials, wherein: the empirical model (Extended Freundlich equation, C-Q transfer equation, etc.) is simple to calculate, but the selected variables are simpler and lack universality; the mechanism model (free ion activity model FIAM, biological ligand model BLM and the like) can effectively explain the environmental behavior of heavy metals in farmland environment, but has high data requirements, more parameters, large modeling difficulty and difficult popularization and application; in addition, parameters used by the experience model and the mechanism model which are promoted in foreign countries are often based on greenhouse experiments or district experiments, the soil factors are few, the differences between the parameters and the actual conditions of farmlands are large, and certain limitation and great uncertainty exist in the actual application.
In recent years, great progress has been made in large data extraction, parsing and transformation models, wherein a classification regression tree model can interpret a dependent variable with one or more independent variables, and both independent and dependent variables can be continuous or discrete data. In view of the fact that the method can explore non-additional and nonlinear relations, reveal hidden structures in a complex data matrix and decompose a data set into a plurality of logic trees, the method has the advantages of being visual, high in robustness, easy to understand results and the like; the Bayesian inference model can fully utilize the existing data, effectively solve the nonlinear relation of pollutants in the environmental multi-medium transmission through priori information, and has strong applicability and easy popularization. However, the two methods are applied to the aspects of logistics management, financial investment risk and weather prediction, and how to improve, couple and optimize the two methods and apply the two methods to the field of agricultural safety production is still a bottleneck problem to be solved in the current model application research.
In summary, how to develop a new model research method based on the existing research results, couple the variation characteristics of a plurality of environmental factors, reveal the influence degree of different soil factor interaction processes on the accumulation of heavy metals in crops, predict the heavy metal enrichment risk of crops, and provide corresponding field management optimization measures, which has great significance for preventing and treating heavy metal pollution in farmlands, and is a technical problem to be solved by those skilled in the art.
Disclosure of Invention
In view of the above, the invention provides a method and a system for predicting the heavy metal enrichment risk of crops based on the Bayesian theory, which are used for constructing a crop heavy metal enrichment risk prediction model based on a classification regression tree and the Bayesian theory, revealing dynamic interaction characteristics of a plurality of soil factors in the heavy metal enrichment process of crops, predicting the heavy metal enrichment trend and the exceeding risk of crops, and optimizing the current pollution source management and control measures so as to reduce the exceeding risk of heavy metals of crops to the maximum extent.
In order to achieve the above object, the present invention provides the following technical solutions:
a method for predicting heavy metal enrichment risk of crops based on Bayesian theory comprises the following steps:
acquiring the basic physicochemical properties of regional soil and the content information of heavy metals in soil and crops, and constructing an observation data set;
establishing a classification regression tree model, setting pruning rules, and obtaining an optimized classification regression tree model;
based on the optimized classification regression tree model, extracting key influencing variables in the heavy metal enrichment process of crops, and analyzing a dynamic interaction process;
constructing and optimizing a Bayesian prediction model based on the mixed random variables to obtain an optimized Bayesian prediction model;
and performing multi-scene simulation based on the optimized Bayesian prediction model, and optimizing the heavy metal pollution prevention and control countermeasures in the region.
The technical effect that above-mentioned technical scheme reaches is: constructing a crop heavy metal enrichment risk prediction model based on a classification regression tree and a Bayesian theory, revealing dynamic interaction characteristics of a plurality of soil factors in the process of enriching the heavy metals of the crops, and predicting the heavy metal enrichment trend and the exceeding risk of the crops; and according to the current situation and the target of regional agricultural development, the constructed model is applied to carry out multi-scene simulation, so that the current pollution source management and control measures are optimized, and the risk of exceeding the standard of heavy metals of crops is reduced to the greatest extent.
Optionally, constructing the observation data set includes the steps of:
sampling in the field to obtain soil and crop samples in a research area;
measuring the obtained soil and crop samples in a research area in a laboratory to obtain the basic physicochemical properties of the soil and the heavy metal content of the soil and the crops;
and establishing a data set of soil basic physicochemical properties and heavy metal contents of different samples in a one-to-one correspondence manner as an observation data set.
Optionally, the method for obtaining the optimized classification regression tree model specifically includes the following steps:
normalizing the observation data set, defining the heavy metal coefficient of crops as a dependent variable, and defining the basic physicochemical property of soil as an independent variable;
establishing a classification regression tree model, and analyzing a heavy metal enrichment process of crops;
and setting pruning rules, and optimizing model parameters of the classification regression tree model to obtain an optimized classification regression tree model.
The technical effect that above-mentioned technical scheme reaches is: screening and training the observation data set through the classification regression tree model, initially establishing a logic number-shaped data structure, setting pruning rules, optimizing model parameters and improving model prediction efficiency.
Optionally, the parsing dynamic interaction process specifically includes the following steps:
acquiring key variables affecting the heavy metal enrichment process of crops based on the optimized classification regression tree model;
acquiring a key soil factor change trend and quantifying an interaction relation based on the key variables which are obtained by analysis and influence the heavy metal enrichment process of crops;
and acquiring a dynamic regulation threshold value of the key soil factor based on the key variables and the interaction relation which are obtained by analysis and influence the heavy metal enrichment process of the crops.
The technical effect that above-mentioned technical scheme reaches is: and sorting the variables by using a classification regression tree model, pruning rules and iterative calculation, so that the number, the change trend and the influence degree of key influence variables in the heavy metal enrichment process of crops can be obtained.
Optionally, obtaining an optimized bayesian prediction model specifically includes the following steps:
based on the dynamic interaction process obtained by analysis, defining the crop heavy metal standard exceeding rate as a dependent variable and defining the key soil physicochemical property as an independent variable;
establishing a Bayes prediction model based on a mixed random variable according to a priori probability density function conforming to normal distribution;
predicting the crop heavy metal exceeding risk by using a Bayesian prediction model, optimizing a priori probability density function of the Bayesian prediction model, and comparing the difference between an observed value and a predicted value;
sampling the simulation result for a plurality of times by adopting a Monte Carlo simulation method, optimizing the input parameters of the Bayesian prediction model, and obtaining an optimized Bayesian prediction model.
The technical effect that above-mentioned technical scheme reaches is: according to dynamic interaction characteristics of a plurality of soil factors in the crop heavy metal enrichment process, a Bayesian prediction model based on a mixed random variable is constructed, the model is optimized, and the model prediction precision is improved.
Optionally, the method for optimizing the countermeasures for preventing and controlling the heavy metal pollution in the area specifically comprises the following steps:
performing risk assessment on different soil key factors by using an optimized Bayes prediction model, and predicting the crop heavy metal exceeding rate in the current soil environment;
carrying out regional multi-scene simulation to obtain the crop heavy metal exceeding rate change trend under different regional environments and agricultural measures;
based on the change trend and the target of the heavy metal exceeding rate of crops under different regional environments and agricultural measures, the current pollution source management and control measures and agricultural management measures are optimized.
The technical effect that above-mentioned technical scheme reaches is: according to the current situation and the target of regional agricultural development, the current pollution source management and control measures and agricultural management measures are optimized, the heavy metal enrichment risk of crops is reduced, and the grain safety production is ensured.
The invention also discloses a crop heavy metal enrichment risk prediction system based on Bayesian theory, which comprises: the system comprises a construction module, a first model building and optimizing module, an analyzing module, a second model building and optimizing module and a measure optimizing module; wherein, the liquid crystal display device comprises a liquid crystal display device,
the construction module is used for acquiring the basic physicochemical properties of the regional soil and the heavy metal content information of the soil and crops and constructing an observation data set;
the first model building and optimizing module is used for building a classification regression tree model, setting pruning rules and obtaining an optimized classification regression tree model;
the analysis module is used for extracting key influencing variables in the heavy metal enrichment process of crops based on the optimized classification regression tree model and analyzing the dynamic interaction process;
the second model building and optimizing module is used for building a Bayesian prediction model based on the mixed random variables and optimizing the Bayesian prediction model to obtain an optimized Bayesian prediction model;
and the measure optimization module is used for performing multi-scene simulation based on the optimized Bayesian prediction model and optimizing the heavy metal pollution prevention and control countermeasures in the region.
Optionally, the first model building and optimizing module includes: the system comprises a first defining sub-module, a first establishing sub-module and a first optimizing sub-module; wherein, the liquid crystal display device comprises a liquid crystal display device,
the first definition submodule is used for standardizing the observation data set, defining the heavy metal coefficient of crops as a dependent variable and defining the basic physicochemical property of soil as an independent variable;
the first building sub-module is used for building a classification regression tree model and analyzing the heavy metal enrichment process of crops;
the first optimizing sub-module is used for setting pruning rules, optimizing model parameters of the classification regression tree model, and obtaining an optimized classification regression tree model.
Optionally, the second model building and optimizing module includes: the second defining sub-module, the second establishing sub-module and the second optimizing sub-module; wherein, the liquid crystal display device comprises a liquid crystal display device,
the second definition submodule is used for defining the crop heavy metal exceeding rate as a dependent variable and defining the key soil physicochemical property as an independent variable based on the dynamic interaction process obtained by analysis;
the second building sub-module is used for building a Bayesian prediction model based on the mixed random variable according to the prior probability density function conforming to the normal distribution;
and the second optimizing sub-module predicts the crop heavy metal out-of-standard risk by using the Bayesian prediction model, optimizes the prior probability density function of the Bayesian prediction model, compares the difference between the observed value and the predicted value, samples the simulation result for a plurality of times by adopting a Monte Carlo simulation method, optimizes the input parameters of the Bayesian prediction model and acquires the optimized Bayesian prediction model.
Compared with the prior art, the invention discloses a method and a system for predicting the heavy metal enrichment risk of crops based on the Bayesian theory, which have the following beneficial effects:
(1) According to the method, a crop heavy metal enrichment risk prediction model based on a classification regression tree and a Bayesian theory is constructed, dynamic interaction characteristics of a plurality of soil factors in the process of enriching the heavy metals of the crops are revealed, and the heavy metal enrichment trend and the exceeding risk of the crops are predicted; in addition, according to the current situation and the target of regional agricultural development, the constructed model is applied to carry out multi-scene simulation, the current pollution source management and control measures are optimized, the risk of exceeding the standard of heavy metals of crops is reduced to the maximum extent, and the grain safety production is ensured;
(2) According to the method, the observation data set is screened and trained through the classification regression tree model, a logical data structure is initially established, pruning rules are set, model parameters are optimized, model prediction efficiency is improved, the number, variation trend and influence degree of key influence variables in the heavy metal enrichment process of crops are obtained, a Bayesian prediction model based on mixed random variables is constructed, the model is optimized, and model prediction precision is improved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings that are required to be used in the embodiments or the description of the prior art will be briefly described below, and it is obvious that the drawings in the following description are only embodiments of the present invention, and that other drawings can be obtained according to the provided drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of a method for predicting heavy metal enrichment risk of crops based on Bayesian theory;
FIG. 2 is an analytical diagram of a regional soil Cd pollution process;
fig. 3 (a), 3 (b) and 3 (c) are schematic diagrams of prior probability distribution and posterior probability distribution of rice Cd content, soil amorphous manganese content and soil pH as key parameters obtained by random sampling respectively;
fig. 4 (a), fig. 4 (b) and fig. 4 (c) are schematic diagrams of the probability of exceeding the standard of the rice Cd according to the critical parameters of the rice Cd content, the soil amorphous manganese content and the soil pH, which are predicted by the bayesian risk prediction model;
FIGS. 5 (a), 5 (b) and 5 (c) are schematic diagrams showing the risk of exceeding Cd standard of rice by 10%, 30% and 50% respectively;
fig. 6 is a schematic structural diagram of a bayesian theory-based crop heavy metal enrichment risk prediction system.
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.
The embodiment of the invention discloses a crop heavy metal enrichment risk prediction method based on Bayesian theory, which is shown in fig. 1 and comprises the following steps:
acquiring the basic physicochemical properties of regional soil and the content information of heavy metals in soil and crops, and constructing an observation data set;
establishing a classification regression tree model, setting pruning rules, and obtaining an optimized classification regression tree model;
based on the optimized classification regression tree model, extracting key influencing variables in the heavy metal enrichment process of crops, and analyzing a dynamic interaction process;
constructing and optimizing a Bayesian prediction model based on the mixed random variables to obtain an optimized Bayesian prediction model;
and performing multi-scene simulation based on the optimized Bayesian prediction model, and optimizing the heavy metal pollution prevention and control countermeasures in the region.
Rice cadmium (Cd) pollution is currently the most troublesome grain safety problem, but rice Cd pollution risk influencing factors are complex, and a scientific means for accurately evaluating the pollution risk influencing factors is lacking. Next, the process shown in fig. 1 will be described in detail with the rice cadmium pollution risk management problem as an example.
First, an observation dataset is constructed, comprising the steps of: sampling in the field to obtain soil and crop samples in a research area; measuring the obtained soil and crop samples in a research area in a laboratory to obtain the basic physicochemical properties of the soil and the heavy metal content of the soil and the crops; and establishing a data set of soil basic physicochemical properties and heavy metal contents of different samples in a one-to-one correspondence manner as an observation data set.
Further, an optimized classification regression tree model is obtained, and the method specifically comprises the following steps: normalizing the observation data set, defining the heavy metal coefficient of crops as a dependent variable, and defining the basic physicochemical property of soil as an independent variable; establishing a classification regression tree model, and analyzing a heavy metal enrichment process of crops; and setting pruning rules, optimizing model parameters of the classification regression tree model, obtaining an optimized classification regression tree model, and improving model prediction efficiency.
Specifically, firstly establishing a data set A of soil physicochemical properties and heavy metal contents of different samples in a one-to-one correspondence manner, then defining a crop heavy metal coefficient (BCF) as a dependent variable, defining soil physicochemical properties (V) as independent variables, and calculating a formula (1):
wherein: BCF represents rice Cd enrichment factor (dimensionless), C represents constant, n represents sample size, beta i (i=1, 2,3,) n represents a fitting parameter; v (V) i Indicating physical and chemical properties of soil, includingSoil pH, clay content (%), cation exchange capacity (cmolkg) -1 ) Organic matter (gkg) -1 ) Alkaline hydrolysis of nitrogen (mgkg) -1 ) Amorphous Fe (mgkg) -1 ) Amorphous Mn (mgkg) -1 ) Phosphorus in soil (mgkg) -1 ) Zinc in soil (mgkg) -1 ) Isovaries.
And establishing a classification regression tree model through cyclic analysis training based on the data set A. The method comprises the following specific steps: (1) Firstly, starting from a root node, finding out all possible dividing conditions of independent variables which can be used as classification variables, randomly selecting a certain characteristic variable from a sample set for each non-leaf node to train a training set, dividing the training set into two training subsets according to results and using the two training subsets as a new node, and then continuously repeating the steps, and finally using the generated leaf node as a classification or prediction value; (2) The classification process has no stop condition, so that a very huge tree model can be obtained, and the model usually has very high fitting degree, but the prediction accuracy is poor, and the popularization and the application are difficult, so that the nodes of a complex tree are required to be pruned through the pruning process, and the complexity of the tree structure is controlled.
In order to ensure the classification accuracy of the model, the embodiment uses the GINI value as the basis of node splitting, and the smaller the GINI value, the higher the sample purity is, the better the classification effect is, and the calculation process is shown in the formula (2):
wherein I (A) represents the purity of the data set A, m represents the class, l is the total number of classes, K m Is the proportion belonging to class m in the observation point.
In order to ensure the prediction accuracy of the model, when the model is used for regression, the root mean square error of the input variable is used as the basis of the split node, the smaller the square error is, the higher the sample purity is, the better the prediction effect is, and the calculation equation is shown in the formula (3):
wherein RMSE represents root mean square error, n is sample size, y pi Representing the predicted value of the target variable, y i Representing the target variable observations.
In order to ensure the operation efficiency of the model, in this embodiment, the minimum loss function is used to cut the regression tree, so as to avoid excessive fitting, and the calculation equation is shown in formula (4):
where α represents a balance constant that measures the model runtime and complexity, T represents a node, T t Represents a subtree rooted at internal node t, |N t I represents the subtree T t The number of leaf nodes in (a). R (T) represents a subtree T t The calculation formula is R (t) =r (t) ×p (t), wherein R (t) is the number of the misclassified samples of the node t, and p (t) is all samples falling into the node t. R (T) t ) Representing subtree T t The number of misclassified samples when not pruned is calculated as R (T t ) = Σr (i), i being the subtree T t Is a leaf node of (c).
The pruning process is divided into two parts: (1) First, alpha value is calculated for each non-leaf node of the complete regression tree T, the subtree with the minimum alpha value is pruned by circular calculation until the root node is left, and a series of pruned trees { T }, are generated 0 ,T 1 ,T 2 ,...,T n }, T therein 0 Is T, T i+1 To T pair i Pruning is carried out to obtain a pruning process tree; (2) The new pruning set is applied on the basis to recalculate the steps and evaluate the model running time and complexity until the best pruning tree is found as an output result.
Further, the analysis dynamic interaction process specifically comprises the following steps: acquiring key variables affecting the heavy metal enrichment process of crops based on the optimized classification regression tree model; acquiring a key soil factor change trend and quantifying an interaction relation based on the key variables which are obtained by analysis and influence the heavy metal enrichment process of crops; and acquiring a dynamic regulation threshold value of the key soil factor based on the key variables and the interaction relation which are obtained by analysis and influence the heavy metal enrichment process of the crops.
Further, an optimized Bayesian prediction model is obtained, which specifically comprises the following steps: based on the dynamic interaction process obtained by analysis, defining the crop heavy metal standard exceeding rate as a dependent variable and defining the key soil physicochemical property as an independent variable; establishing a Bayes prediction model based on a mixed random variable according to a priori probability density function conforming to normal distribution; predicting the crop heavy metal exceeding risk by using a Bayesian prediction model, optimizing a priori probability density function of the Bayesian prediction model, and comparing the difference between an observed value and a predicted value; sampling the simulation result for a plurality of times by adopting a Monte Carlo simulation method, optimizing the input parameters of the Bayesian prediction model, obtaining an optimized Bayesian prediction model, and improving the model prediction precision.
Specifically, aiming at the output result of the classification regression tree, a Bayes prediction model based on the mixed random variable is further constructed. The Bayesian model pays attention to the collection, mining and processing of priori information, and is applied to statistical inference, so that the quality of the statistical inference can be effectively improved, and the calculation equation is shown in a formula (5):
where P (A|B) represents the conditional probability of occurrence of random event A in the event of occurrence of random event B (i.e., the posterior probability of occurrence of random event A), P (A) i ) And P (A) j ) For the prior probability of two key variables occurring at random event a, P (b|a i ) And P (B|A) j ) To be in variable A i And A j The conditional probability of random event B occurring after random event a occurs under influence.
According to the Bayes theory, a Bayes prediction model for predicting the risk of exceeding the standard of the rice Cd is constructed, and the specific steps are as follows: (1) Depending on the type of distribution of the data, it is appropriateTo make it normally distributed; in this embodiment, the data is distributed in a biased state, so that the data is logarithmically converted to be distributed in a normal state; (2) Randomly taking n pairs of soil-rice paired samples as priori data, and taking the rest samples as posterior data; dividing n pairs of samples of prior data into up-to-standard samples n according to whether Cd content of rice exceeds standard 1 And superscalar samples (n-n) 1 ) And taking the Cd content of the rice as the prior distribution of whether the Cd content of the rice exceeds the standard; (3) Establishing a normal distribution density function aiming at the prior data of the key variables, and calculating the average value and standard deviation of the Cd content of the soil under the condition of exceeding standard and reaching standard; (4) And establishing a Bayesian risk prediction model according to the distribution density function of the prior data. The core equation is shown in equation (6).
Wherein: p is Bayes risk probability of exceeding Cd content of rice; n is n 1 /n、n 2 And n is the standard rate and the standard exceeding rate of Cd content of rice in the prior data respectively; mu (mu) 1 Sum sigma 1 Respectively obtaining an average value and a standard deviation of a soil factor V under the condition that the Cd content of rice in prior data meets the standard; mu (mu) 2 、σ 2 Respectively obtaining an average value and a standard deviation of a soil factor V under the condition that the Cd content of rice in prior data exceeds the standard; x is the observed value of the soil factor V in posterior data.
Introducing parameter eta a 、η b And eta c Simplifying the parameters required by the formula (6) (for optimization equations, see the formula (7), the formula (8) and the formula (9)), the rice Cd superscalar risk Bayesian prediction risk model can be further optimized into the formula (10).
The model prediction result can be obtained by a coefficient (R 2 ) For verification, see equation (11).
In which y pi Representing the predicted value of the target variable, y i The observed value of the target variable (the target variable in this embodiment is the rice Cd superscalar).
In the embodiment, two key variables (amorphous manganese in soil and soil pH) affecting the Cd enrichment of rice are obtained through a classification regression tree model, and an optimal tree structure (formulas 1-4) is further obtained through pruning, so that the dynamic interaction relation and the regulation and control threshold value of key soil factors in the Cd enrichment process of rice are revealed (see FIG. 2). Based on the above, a Bayes prediction model based on mixed random variables is constructed, the operation process is shown in formulas (5) - (10), the model parameters are shown in table 1, and the obtained optimal model is shown in formulas (12) - (14).
In the above, P (Cd) rice )、P(Mn ox ) And P (pH) are respectivelyAnd the Bayesian risk probability of the Cd content of the rice when the Cd content of the rice, the amorphous manganese of the soil and the pH of the soil are the key parameters.
Table 1 Bayes Cd out-of-standard risk prediction model parameters based on mixed random variables
Further, optimizing the countermeasures for preventing and treating heavy metal pollution in the area specifically comprises the following steps: performing risk assessment on different soil key factors by using an optimized Bayes prediction model, and predicting the crop heavy metal exceeding rate in the current soil environment; carrying out regional multi-scene simulation to obtain the crop heavy metal exceeding rate change trend under different regional environments and agricultural measures; based on the change trend and the target of the heavy metal exceeding rate of crops under different regional environments and agricultural measures, the current pollution source management and control measures and agricultural management measures are optimized.
At present, the occurrence place of the 'cadmium rice wind wave' event is selected as a research area, and the regional rice field Cd pollution profile is obtained through field sampling investigation and laboratory analysis of 683 rice fields in the research area, so as to build a database. As a result, it was found that the Cd content of the regional soil was 0.443.+ -. 0.248mgkg -1 The Cd content of the rice is 0.467+/-0.329 mgDWkg -1 The rice exceeds standard rate>0.2mgDWkg -1 ) 82.1%. The average value of the Cd enrichment factor (rice Cd/soil Cd) of the regional rice is 1.19, and exceeds a safety threshold value>1) The probability of (2) is 50.4%. Based on the constructed database, a classification regression tree model is established and pruning is carried out, the model prediction efficiency is improved, and the heavy metal enrichment process of crops is analyzed, as shown in figure 2.
The regression tree analysis result shows that the pruned model comprises 13 nodes and 7 terminal nodes, the interpretation rate of the model on the Cd enrichment factor of rice is up to 56.9%, and the model prediction precision is good. Model shows soil amorphous manganese (Mn ox ) And the soil pH (SoilpH) is 2 soil factors which influence the change of the Cd enrichment factor of the rice most critically, other soil factors influence the Cd enrichment process of the rice through indirect interaction with the 2 key soil factors, and the 2 key soil factors are in dynamic interaction in the Cd enrichment process of the rice. Among them, the most important key factor affecting the Cd enrichment factor change of rice is amorphous manganese (Mn ox ) Has influence on 683 investigation plots, and the regulation and control threshold values are 73.7mg kg respectively -1 (Node 1-3) and 179mgkg -1 (Node 6-7). After eliminating the influence of amorphous manganese in the soil, the soil pH (SoilpH) is a secondary key factor affecting the variation of Cd enrichment factor of rice. The soil pH has an effect on 486 survey plots, and the regulation and control thresholds are 5.50 (Node 4-5), 4.65 (Node 10-11) and 5.21 (Node 12-13) respectively. Soil Organic Matter (OM) is also more pronounced for rice Cd enrichment, but to a lesser extent than soil amorphous manganese and soil pH. The field experiment carried out in the research area further proves that the regression tree model analysis result, compared with the positive and negative correlation results obtained by the traditional correlation analysis, the data information mined by the model is more abundant, the content level, the number of sample areas, the regulation threshold value and the interaction characteristics among different variables are related, and the result is more accurate and reliable.
Based on the analysis result of the regression classification tree, carrying out data conversion and calculation, carrying out random sampling on key parameters by combining with a MonteCarlo simulation method, obtaining a priori density function and a posterior density function, and constructing a Bayesian prediction model based on mixed random variables. The result of the prior and posterior probability density functions of the key soil factors accumulated by the rice Cd in the affected area is shown in FIG. 3. The results of fig. 3 (a) -3 (c) show that the prior probability density function and the posterior probability density function of 3 key variables of rice Cd content, soil amorphous manganese and soil pH are consistent in distribution characteristics, so that the prior probability density function and the posterior probability density function can be used for predicting the out-of-standard probability of rice Cd, and the closer the data distribution is, the smaller the prediction relative deviation is. The results further illustrate the exact extraction of key variables by the classification regression tree and the efficient classification and extraction of existing data by the bayesian model.
And (4) carrying out rice Cd heavy metal exceeding rate prediction work aiming at different soil key factors in a research area by applying an optimized Bayesian model, wherein the simulation times are set to 1000 times, and the equation prediction results are good and pass through the significance test (figure 4). As can be seen from fig. 4 (a), when the content of the rice Cd is taken as a key parameter, the predicted risk probability (P) of exceeding the standard of the rice Cd is 76.7% ± 7.43%, which is basically consistent with the observed value (76.6% ± 6.47%) of exceeding the standard of the rice Cd, and the predicted values of 95.8% are all within the 95% confidence interval; as can be seen from fig. 4 (b), when the amorphous manganese content of the soil is taken as a key parameter, the predicted risk probability (P) of the rice Cd exceeding the standard is 72.0% ± 10.5%, which is slightly lower than the observed value (73.2% ± 8.85%) of the rice Cd exceeding the standard of the group, but the difference between the two is small (1.2%), and the predicted values of 94.1% are all within the 95% confidence interval; as can be seen from fig. 4 (c), when the pH of the soil is the key parameter, the predicted risk probability (P) of exceeding the standard of rice Cd is 76.5% ± 6.60%, which is slightly lower than the observed value (76.7% ± 5.60%) of exceeding the standard of rice Cd in the group, but the difference between the two is small (0.2%), and the predicted value of 91.8% is within the 95% confidence interval. The result shows that the constructed Bayesian model has higher prediction accuracy, the more accurate the parameter selection (the distribution characteristics of the prior probability density function and the posterior probability density function are closer), the more the sampling times, the more stable the model performance and the smaller the deviation.
And carrying out regional multi-scene simulation on the basis. Taking lifting of soil amorphous manganese as an example, referring to fig. 5 (a), the observed value of Cd exceeding standard rate of the rice in the region of 10% of the soil amorphous manganese is 63.9% ± 4.97%, the predicted value (P) of Cd exceeding standard rate of the rice predicted by the model under the situation is 64.2% ± 5.78%, and the difference is smaller (0.3%) compared with the observed value; referring to fig. 5 (b), the observed value of the Cd exceeding standard rate of the area rice in the amorphous manganese of 30% of the soil is improved to be 49.4% ± 5.67%, the predicted value (P) of the Cd exceeding standard rate of the area rice predicted by the model in the situation is 49.5% ± 6.98%, and the difference is smaller (0.1%) compared with the observed value; referring to fig. 5 (c), the observed value of the over-standard rate of the rice Cd in the area of 50% of soil amorphous manganese is 34.9% ± 3.01%, the predicted value (P) of the over-standard rate of the rice Cd predicted by the model in the scene is 35.1% ± 3.71%, and the difference is small (0.2%) compared with the observed value. According to the model prediction result, the excessive rate of rice Cd can be effectively reduced by improving the amorphous manganese in the soil, and compared with the excessive rate of the regional rice Cd (73.2% +/-8.85%), the maximum reduction can reach 38.1%, so that the effect of improving the amorphous manganese in the soil by 50% is most obvious. The model predictions were further confirmed in subsequent field experiments.
In summary, the optimized Bayes model is applied to develop the prediction work of the rice Cd heavy metal exceeding rate (formulas 12-14) aiming at different soil key factors of a research area, the simulation times are set to 1000 times, the formulas pass through the significance test and are stable in performance (see fig. 3 (a) -3 (c) and fig. 4 (a) -4 (c)), and the prediction deviation can be further controlled within 5% by coupling key variables (see fig. 4). By applying the model to perform multi-scene simulation, the method finds that the Cd standard exceeding rate of rice can be effectively reduced by improving the amorphous manganese in the soil, and the reducing amplitude of the Cd standard exceeding rate of the rice can reach nearly 40% when the amorphous manganese in the soil is improved by 50% (see fig. 5 (a) -5 (c)).
Corresponding to the method shown in fig. 1, the embodiment of the invention also provides a system for predicting the heavy metal enrichment risk of crops based on bayesian theory, which is used for realizing the method shown in fig. 1, and the system for predicting the heavy metal enrichment risk of crops based on bayesian theory provided by the embodiment of the invention can be applied to a computer terminal or various mobile devices, and has a structure schematic diagram shown in fig. 6, and comprises: the system comprises a construction module, a first model building and optimizing module, an analyzing module, a second model building and optimizing module and a measure optimizing module; wherein, the liquid crystal display device comprises a liquid crystal display device,
the construction module is used for acquiring the basic physicochemical properties of the regional soil and the heavy metal content information of the soil and crops and constructing an observation data set;
the first model building and optimizing module is used for building a classification regression tree model, setting pruning rules and obtaining an optimized classification regression tree model;
the analysis module is used for extracting key influencing variables in the heavy metal enrichment process of crops based on the optimized classification regression tree model and analyzing the dynamic interaction process;
the second model building and optimizing module is used for building a Bayesian prediction model based on the mixed random variables and optimizing the Bayesian prediction model to obtain an optimized Bayesian prediction model;
and the measure optimization module is used for performing multi-scene simulation based on the optimized Bayesian prediction model and optimizing the heavy metal pollution prevention and control countermeasures in the region.
Further, the first model building and optimizing module includes: the system comprises a first defining sub-module, a first establishing sub-module and a first optimizing sub-module; wherein, the liquid crystal display device comprises a liquid crystal display device,
the first definition submodule is used for standardizing the observation data set, defining the heavy metal coefficient of crops as a dependent variable and defining the basic physicochemical property of soil as an independent variable;
the first building sub-module is used for building a classification regression tree model and analyzing the heavy metal enrichment process of crops;
the first optimizing sub-module is used for setting pruning rules, optimizing model parameters of the classification regression tree model, and obtaining an optimized classification regression tree model.
Further, the second model building and optimizing module includes: the second defining sub-module, the second establishing sub-module and the second optimizing sub-module; wherein, the liquid crystal display device comprises a liquid crystal display device,
the second definition submodule is used for defining the crop heavy metal exceeding rate as a dependent variable and defining the key soil physicochemical property as an independent variable based on the dynamic interaction process obtained by analysis;
the second building sub-module is used for building a Bayesian prediction model based on the mixed random variable according to the prior probability density function conforming to the normal distribution;
and the second optimizing sub-module predicts the crop heavy metal out-of-standard risk by using the Bayesian prediction model, optimizes the prior probability density function of the Bayesian prediction model, compares the difference between the observed value and the predicted value, samples the simulation result for a plurality of times by adopting a Monte Carlo simulation method, optimizes the input parameters of the Bayesian prediction model and acquires the optimized Bayesian prediction model.
In summary, the crop heavy metal enrichment risk prediction model based on the classification regression tree and the Bayesian theory provided by the embodiment has wide applicability and stable operation, can accurately predict the crop heavy metal exceeding risk, reveals the influence degree of different soil factor interaction processes on crop heavy metal accumulation, provides corresponding field management optimization measures, and provides powerful data support for regional pollution control, grain safety production and decision suggestion.
In the present specification, each embodiment is described in a progressive manner, and each embodiment is mainly described in a different point from other embodiments, and identical and similar parts between the embodiments are all enough to refer to each other. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant points refer to the description of the method section.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims (7)
1. The crop heavy metal enrichment risk prediction method based on the Bayesian theory is characterized by comprising the following steps of:
acquiring information of basic physicochemical properties of regional soil, heavy metal content of soil and heavy metal content of crops, and constructing an observation data set;
establishing a classification regression tree model, setting pruning rules, and obtaining an optimized classification regression tree model;
based on the optimized classification regression tree model, extracting key influencing variables in the heavy metal enrichment process of crops, and analyzing a dynamic interaction process;
constructing and optimizing a Bayesian prediction model based on the mixed random variables to obtain an optimized Bayesian prediction model;
performing multi-scene simulation based on the optimized Bayesian prediction model, and optimizing heavy metal pollution prevention and control countermeasures in the region;
the optimized Bayesian prediction model is obtained, which comprises the following steps:
based on the dynamic interaction process obtained by analysis, defining the crop heavy metal standard exceeding rate as a dependent variable and defining the key soil physicochemical property as an independent variable;
establishing a Bayes prediction model based on a mixed random variable according to a priori probability density function conforming to normal distribution;
predicting the crop heavy metal exceeding risk by using a Bayesian prediction model, optimizing a priori probability density function of the Bayesian prediction model, and comparing the difference between an observed value and a predicted value;
sampling the simulation result for a plurality of times by adopting a Monte Carlo simulation method, optimizing the input parameters of the Bayesian prediction model, and obtaining an optimized Bayesian prediction model.
2. The method for predicting the heavy metal enrichment risk of crops based on the Bayesian theory according to claim 1, wherein the step of constructing an observation data set comprises the following steps:
sampling in the field to obtain soil and crop samples in a research area;
measuring the obtained soil and crop samples in a research area in a laboratory to obtain the basic physicochemical properties of the soil, the content of heavy metals in the soil and the content of heavy metals in the crops;
and establishing a data set of soil basic physicochemical properties and heavy metal contents of different samples in a one-to-one correspondence manner as an observation data set.
3. The method for predicting the heavy metal enrichment risk of crops based on the Bayesian theory according to claim 1, wherein the method for predicting the heavy metal enrichment risk of crops based on the Bayesian theory is characterized by obtaining an optimized classification regression tree model, and specifically comprises the following steps:
normalizing the observation data set, defining the heavy metal coefficient of crops as a dependent variable, and defining the basic physicochemical property of soil as an independent variable;
establishing a classification regression tree model, and analyzing a heavy metal enrichment process of crops;
and setting pruning rules, and optimizing model parameters of the classification regression tree model to obtain an optimized classification regression tree model.
4. The method for predicting the heavy metal enrichment risk of crops based on the Bayesian theory according to claim 1, wherein the analysis dynamic interaction process specifically comprises the following steps:
acquiring key variables affecting the heavy metal enrichment process of crops based on the optimized classification regression tree model;
acquiring a key soil factor change trend and quantifying an interaction relation based on the key variables which are obtained by analysis and influence the heavy metal enrichment process of crops;
and acquiring a dynamic regulation threshold value of the key soil factor based on the key variables and the interaction relation which are obtained by analysis and influence the heavy metal enrichment process of the crops.
5. The method for predicting the heavy metal enrichment risk of crops based on the Bayesian theory according to claim 1, wherein the method for optimizing the countermeasures for preventing and controlling the heavy metal pollution in the area is characterized by comprising the following steps:
performing risk assessment on different soil key factors by using an optimized Bayes prediction model, and predicting the crop heavy metal exceeding rate in the current soil environment;
carrying out regional multi-scene simulation to obtain the crop heavy metal exceeding rate change trend under different regional environments and agricultural measures;
based on the change trend and the target of the heavy metal exceeding rate of crops under different regional environments and agricultural measures, the current pollution source management and control measures and agricultural management measures are optimized.
6. Crop heavy metal enrichment risk prediction system based on Bayesian theory, which is characterized by comprising: the system comprises a construction module, a first model building and optimizing module, an analyzing module, a second model building and optimizing module and a measure optimizing module; wherein, the liquid crystal display device comprises a liquid crystal display device,
the construction module is used for acquiring information of basic physicochemical properties of regional soil, heavy metal content of soil and heavy metal content of crops and constructing an observation data set;
the first model building and optimizing module is used for building a classification regression tree model, setting pruning rules and obtaining an optimized classification regression tree model;
the analysis module is used for extracting key influencing variables in the heavy metal enrichment process of crops based on the optimized classification regression tree model and analyzing the dynamic interaction process;
the second model building and optimizing module is used for building a Bayesian prediction model based on the mixed random variables and optimizing the Bayesian prediction model to obtain an optimized Bayesian prediction model;
the measure optimization module is used for performing multi-scene simulation based on the optimized Bayesian prediction model and optimizing the heavy metal pollution prevention and control countermeasures in the region;
the second model building and optimizing module comprises: the second defining sub-module, the second establishing sub-module and the second optimizing sub-module; wherein, the liquid crystal display device comprises a liquid crystal display device,
the second definition submodule is used for defining the crop heavy metal exceeding rate as a dependent variable and defining the key soil physicochemical property as an independent variable based on the dynamic interaction process obtained by analysis;
the second building sub-module is used for building a Bayesian prediction model based on the mixed random variable according to the prior probability density function conforming to the normal distribution;
and the second optimizing sub-module predicts the crop heavy metal out-of-standard risk by using the Bayesian prediction model, optimizes the prior probability density function of the Bayesian prediction model, compares the difference between the observed value and the predicted value, samples the simulation result for a plurality of times by adopting a Monte Carlo simulation method, optimizes the input parameters of the Bayesian prediction model and acquires the optimized Bayesian prediction model.
7. The system for predicting heavy metal enrichment risk of crops based on bayesian theory according to claim 6, wherein the first model building and optimizing module comprises: the system comprises a first defining sub-module, a first establishing sub-module and a first optimizing sub-module; wherein, the liquid crystal display device comprises a liquid crystal display device,
the first definition submodule is used for standardizing the observation data set, defining the heavy metal coefficient of crops as a dependent variable and defining the basic physicochemical property of soil as an independent variable;
the first building sub-module is used for building a classification regression tree model and analyzing the heavy metal enrichment process of crops;
the first optimizing sub-module is used for setting pruning rules, optimizing model parameters of the classification regression tree model, and obtaining an optimized classification regression tree model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211410742.2A CN115775042B (en) | 2022-11-11 | 2022-11-11 | Crop heavy metal enrichment risk prediction method and system based on Bayesian theory |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211410742.2A CN115775042B (en) | 2022-11-11 | 2022-11-11 | Crop heavy metal enrichment risk prediction method and system based on Bayesian theory |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115775042A CN115775042A (en) | 2023-03-10 |
| CN115775042B true CN115775042B (en) | 2023-05-05 |
Family
ID=85388942
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211410742.2A Active CN115775042B (en) | 2022-11-11 | 2022-11-11 | Crop heavy metal enrichment risk prediction method and system based on Bayesian theory |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115775042B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116384624B (en) * | 2023-03-13 | 2023-09-05 | 中国科学院生态环境研究中心 | Method and system for determining optimal soil tillage depth of region for deep tillage measure |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108549620A (en) * | 2018-03-07 | 2018-09-18 | 广东省生态环境技术研究所 | A kind of method of estimation of Study on Availability Control of Heavy Metals in Soil, system and device |
| CN108764515A (en) * | 2018-04-04 | 2018-11-06 | 河海大学 | A kind of reservoir operation Application of risk decision method of Coupled Numerical meteorological model DATA PROCESSING IN ENSEMBLE PREDICTION SYSTEM |
| CN112949202A (en) * | 2021-03-19 | 2021-06-11 | 交通运输部科学研究院 | Bayesian network-based rockburst probability prediction method |
| CN113435707A (en) * | 2021-06-03 | 2021-09-24 | 大连钜智信息科技有限公司 | Soil testing and formulated fertilization method based on deep learning and weighted multi-factor evaluation |
| CN113642790A (en) * | 2021-08-12 | 2021-11-12 | 北京工业大学 | Regional soil background value prediction method based on support vector machine |
| CN114238858A (en) * | 2021-12-15 | 2022-03-25 | 中国科学院生态环境研究中心 | Method and system for reducing accumulation value of heavy metals in crops |
| CN114239278A (en) * | 2021-12-17 | 2022-03-25 | 中国科学院生态环境研究中心 | Method for constructing time-space simulation model of soil heavy metal accumulation process |
-
2022
- 2022-11-11 CN CN202211410742.2A patent/CN115775042B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108549620A (en) * | 2018-03-07 | 2018-09-18 | 广东省生态环境技术研究所 | A kind of method of estimation of Study on Availability Control of Heavy Metals in Soil, system and device |
| CN108764515A (en) * | 2018-04-04 | 2018-11-06 | 河海大学 | A kind of reservoir operation Application of risk decision method of Coupled Numerical meteorological model DATA PROCESSING IN ENSEMBLE PREDICTION SYSTEM |
| CN112949202A (en) * | 2021-03-19 | 2021-06-11 | 交通运输部科学研究院 | Bayesian network-based rockburst probability prediction method |
| CN113435707A (en) * | 2021-06-03 | 2021-09-24 | 大连钜智信息科技有限公司 | Soil testing and formulated fertilization method based on deep learning and weighted multi-factor evaluation |
| CN113642790A (en) * | 2021-08-12 | 2021-11-12 | 北京工业大学 | Regional soil background value prediction method based on support vector machine |
| CN114238858A (en) * | 2021-12-15 | 2022-03-25 | 中国科学院生态环境研究中心 | Method and system for reducing accumulation value of heavy metals in crops |
| CN114239278A (en) * | 2021-12-17 | 2022-03-25 | 中国科学院生态环境研究中心 | Method for constructing time-space simulation model of soil heavy metal accumulation process |
Non-Patent Citations (5)
| Title |
|---|
| Bayesian inference of metal oxide ultrathin film structure based on crystal truncation rod measurements;Anada Masato 等;《Journal of Applied Crystallography》;第50卷(第6期);1611-1616 * |
| Publicly available QSPR models for environmental media persistence;Lunghini Filippo 等;《SAR and QSAR in Environmental Research》;第31卷(第7期);493-510 * |
| 基于数字图像处理技术的苹果树叶片氮含量检测研究;张磊;《中国优秀硕士学位论文全文数据库农业科技辑》(第01期);D048-131 * |
| 大宝山矿区流域重金属生态风险评估预警和修复对策;卜中原;《中国优秀硕士学位论文全文数据库工程科技Ⅰ辑》(第02期);B021-337 * |
| 飞播马尾松林树冠特征及单木冠幅模型研究;杨阳;《中国优秀硕士学位论文全文数据库农业科技辑》(第06期);D049-78 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115775042A (en) | 2023-03-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Xu et al. | Research and application of a hybrid model based on dynamic fuzzy synthetic evaluation for establishing air quality forecasting and early warning system: A case study in China | |
| US20220082545A1 (en) | Total Nitrogen Intelligent Detection Method Based on Multi-objective Optimized Fuzzy Neural Network | |
| Wei et al. | Comprehensive evaluation model for water environment carrying capacity based on VPOSRM framework: A case study in Wuhan, China | |
| US20180029900A1 (en) | A Method for Effluent Total Nitrogen-based on a Recurrent Self-organizing RBF Neural Network | |
| CN110782658B (en) | Traffic prediction method based on LightGBM algorithm | |
| Xiong et al. | Pollution reduction effect of the digital transformation of heavy metal enterprises under the agglomeration effect | |
| CN106991437A (en) | The method and system of sewage quality data are predicted based on random forest | |
| CN111489036A (en) | Resident load prediction method and device based on electrical appliance load characteristics and deep learning | |
| CN114626512A (en) | High-temperature disaster forecasting method based on directed graph neural network | |
| CN103886218A (en) | Lake and reservoir algal bloom predicating method based on multielement nonstationary time series analysis and neural network and support vector machine compensation | |
| CN110648014A (en) | Regional wind power prediction method and system based on space-time quantile regression | |
| CN115775042B (en) | Crop heavy metal enrichment risk prediction method and system based on Bayesian theory | |
| CN115829120A (en) | Water quality prediction early warning system based on machine learning method | |
| Wang et al. | A novel water quality mechanism modeling and eutrophication risk assessment method of lakes and reservoirs | |
| CN112785450A (en) | Soil environment quality partitioning method and system | |
| CN113159456A (en) | Water quality prediction method, device, electronic device, and storage medium | |
| CN103793604A (en) | Sewage treatment soft measuring method based on RVM | |
| Zhang et al. | Study on water quality prediction of urban reservoir by coupled CEEMDAN decomposition and LSTM neural network model | |
| Aldrees et al. | Evolutionary and ensemble machine learning predictive models for evaluation of water quality | |
| CN116777608A (en) | Agricultural financial risk supervision system, method and storage medium based on big data | |
| Yang et al. | An ensemble self-learning framework combined with dynamic model selection and divide-conquer strategies for carbon emissions trading price forecasting | |
| Li et al. | Prediction of composting maturity and identification of critical parameters for green waste compost using machine learning | |
| Yang et al. | A multi-factor forecasting model for carbon emissions based on decomposition and swarm intelligence optimization | |
| Peng et al. | Monitoring of wastewater treatment process based on multi-stage variational autoencoder | |
| Lv et al. | Enhancing effluent quality prediction in wastewater treatment plants through the integration of factor analysis and machine learning |
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 |