Performance evaluation method for constructed wetland ecosystem

Abstract

The invention provides a constructed wetland ecosystem performance evaluation method, which comprises the following steps of: 1, determining an optimization index of constructed wetland ecosystem evaluation, and verifying that the optimization index can replace an original index; the step of determining the optimization index for artificial wetland ecosystem evaluation comprises the steps of processing data of an original index, and screening out an index which can best evaluate the performance of the artificial wetland as the optimization index through principal component analysis and correlation analysis; 2, constructing a logistic regression model with input data as the optimization indexes and the evaluation values, and obtaining a simulation equation with the optimization indexes as independent variables and the evaluation values as dependent variables; the accuracy of the simulation equation is verified; 3, calculating a performance estimation value according to the simulation equation which is verified to be accurate; and evaluating the performance of the constructed wetland ecosystem according to the evaluation value. Indexes required for artificial wetland performance evaluation are reduced, and the complexity of artificial wetland performance evaluation is greatly reduced.

Description

Technical Field

[0001] The invention relates to the technical field of environmental protection, and in particular to a method for evaluating the performance of an artificial wetland ecosystem. Background Art

[0002] The performance evaluation of artificial wetland ecosystem is of great significance. It is not only related to the design and operation of artificial wetlands, but also directly affects its benefits in environmental protection and resource utilization.

[0003] The existing related technology includes a publication number of CN101916425A, and the name is: a method for comprehensive evaluation of artificial wetland ecosystem services, which discloses that the artificial wetland ecosystem services are jointly evaluated by combining the hierarchical analysis method with the fuzzy membership function method. In the application of the hierarchical analysis method, expert scoring is first adopted to construct a comparative judgment matrix of the criterion layer and the indicator layer, and the weight vectors of each judgment matrix are calculated by MATLAB software. Secondly, the hierarchical single sorting and consistency test of the criterion layer and the indicator layer are performed by MATLAB software, and the hierarchy is sorted and the consistency is tested. Then, according to the tracking and monitoring of the research object and the characteristics of each indicator of the indicator layer, the original value is obtained, and the fuzzy membership function method is used to construct the fuzzy membership function of each indicator of the indicator layer, and the original value of each indicator of the indicator layer is substituted into its fuzzy membership function to obtain the fuzzy membership of each indicator. The fuzzy membership function method can realize the dimensionless transformation of various data, convert various types of data into values ​​in the interval [0, 1], and then calculate the fuzzy membership of each indicator for the calculation of the comprehensive index of artificial wetland ecosystem services.

[0004] Patents in the prior art use multiple indicators to comprehensively evaluate the ecosystem services of artificial wetlands. The large number of indicators used makes the calculations more complicated. The monitoring indicators of artificial wetlands are not simplified, and the performance of artificial wetlands is not simulated. The ecosystem service index of artificial wetlands is only calculated through the fuzzy membership function method. Summary of the invention

[0005] In view of the deficiencies in the prior art, the present invention aims to provide a method and device for evaluating the performance of an artificial wetland ecosystem, which can solve the existing problems.

[0006] In order to achieve the above object, the technical solution of the present invention is as follows:

[0007] The present invention is realized by the following technical scheme: a method for evaluating the performance of an artificial wetland ecosystem, comprising the following steps:

[0008] Step 1: Determine the optimization index for artificial wetland ecosystem assessment and verify that the optimized index can replace the original index;

[0009] Determining the optimization index for artificial wetland ecosystem assessment includes processing the data of the original index, and selecting the index that best evaluates the performance of the artificial wetland as the optimization index through principal component analysis and correlation analysis;

[0010] Step 2: construct a logistic regression model whose input data is the optimization index and the evaluation value, obtain a simulation equation with the optimization index as the independent variable and the evaluation value as the dependent variable; and verify the accuracy of the simulation equation;

[0011] Step 3: Calculate the performance estimation value based on the simulation equation after verification; and evaluate the performance of the artificial wetland ecosystem based on the evaluation value.

[0012] Furthermore, the principal component analysis and correlation analysis are used to select the index that best evaluates the performance of the artificial wetland as the optimization index; including

[0013] The principal component analysis extracts the main components of the data from the dimensionality reduction, analyzes the weights of the indicators in the main components, and determines the important indicators;

[0014] The correlation analysis uses the calculation of the Spearman rank correlation coefficient to capture the linear relationship and determine the strongest indicator;

[0015] The indicators that include both important indicators and the strongest indicators are used as optimization indicators.

[0016] Further, the verification that the optimized indicator can replace the original indicator includes:

[0017] Determine constructed wetland performance values;

[0018] Construct a gradient boosting tree model between the original index, the optimized index and the constructed wetland performance value, use the training set to train the two models respectively, and predict the test sets of the two trained models to obtain the prediction results;

[0019] The results were validated by comparing the mean square error and coefficient of determination between the original indicator model and the optimized indicator model.

[0020] Further, the determining of the performance value of the artificial wetland includes:

[0021] If the artificial wetland ecosystem contains data on both qualitative and quantitative indicators, the weights are calculated using the analytic hierarchy process, and the performance value of the artificial wetland is determined based on the weighted scoring results of the analytic hierarchy process;

[0022] If the artificial wetland ecosystem only contains data on quantitative indicators, the entropy weight method is used to calculate the weights. Based on the weights, the TOPSIS standardized evaluation matrix is ​​constructed to obtain the values ​​for evaluating the performance of artificial wetland treatment and determine the performance value of the artificial wetland.

[0023] Furthermore, the qualitative indicators include ecological benefits, social benefits, economic benefits and operation and maintenance;

[0024] The quantitative indicators include the concentrations of ammonia nitrogen, nitrate nitrogen, nitrite nitrogen, total nitrogen, total phosphorus and chemical oxygen demand in the inlet and outlet water.

[0025] Furthermore, the construction of a logistic regression model whose input data is the optimization index and the evaluation value, and obtaining a simulation equation with the optimization index as the independent variable and the evaluation value as the dependent variable, includes:

[0026] 2.1. Building a Dataset

[0027] Data matrix X:

[0028]

[0029] In the above formula, n is the optimized index and m is the number of samples;

[0030] Label vector y: represents the weighted score or closeness, ranging from 0 to 1;

[0031]

[0032] 2.2. Divide the data set into training set and test set, 80% of the data is used for training and 20% of the data is used for testing;

[0033] Training set:

[0034]

[0035] Test set:

[0036]

[0037] 2.3. Training Logistic Regression Model

[0038] Using training set X train and train Train the logistic regression model; the goal of the model is to find the parameter β so that the predicted value is consistent with the true value to the greatest extent; the hypothesis function of the logistic regression model is:

[0039]

[0040] is the hypothesis function of the logistic regression model, X n is the nth eigenvalue;

[0041] 2.4 Model Prediction

[0042] Use the parameters β obtained from the training set to test the data set X testMake predictions and get the predicted value of each test sample The prediction formula is:

[0043]

[0044] Furthermore, the verification of the accuracy of the simulation equation includes fitting the simulation value and the true value to verify the accuracy of the simulation equation.

[0045] Further, the accuracy of the verification simulation equation includes fitting the simulation value and the true value to verify the accuracy of the simulation equation. The simulation equation is as follows:

[0046] F=β0+β1X1+β2X2+…+β n X n

[0047] Where β0 is the intercept, X1, X2...X n Represents the optimization index, β1, β2...β n is the parameter corresponding to the optimization index; F is the calculated weighted score or closeness;

[0048] The calculated value of the simulation equation and the original value are linearly fitted to obtain the determination coefficient R 2 If R 2 If it is greater than 0.8, it means that the simulation equation can be used to simplify the calculation of constructed wetland performance evaluation;

[0049]

[0050] In the formula, y i is the original weighted score or closeness value, is the calculated value of the simulation equation, is the mean of the original closeness.

[0051] A device for evaluating the performance of an artificial wetland ecosystem, characterized in that it comprises a processor and a memory; the memory is used to store programs; and the processor executes the program to implement the method described above.

[0052] A computer-readable storage medium stores a program, and the program is executed by a processor to implement the method described above.

[0053] Compared with the prior art, the beneficial effects of the present invention include:

[0054] The present invention combines correlation analysis with the data to reduce the dimension by principal component analysis, extracts the eigenvector with larger coefficient value in the principal component as an important indicator, and analyzes the indicators with stronger correlation with closeness or weighted score in the data from multiple angles, so as to achieve the purpose of calculating the screening indicators instead of the original indicators. And the rationality of the selected optimization indicators is verified by using the gradient boosting tree model, which is different from the previous technology that only uses principal component analysis to reduce the dimension of the data set without considering the influence of the eigenvector coefficient value on the principal component. The correlation analysis used at the same time also further strengthens the reliability and persuasiveness of the screening results of the indicators. In addition, the present invention uses a logistic regression model to construct a linear simulation equation of artificial wetland indicators and closeness or weighted score, which greatly reduces the complexity of evaluating the performance of artificial wetlands, and can obtain results that are more consistent with the original values. The uniqueness of the present invention is that the indicators required for the performance evaluation of artificial wetlands are reduced, which means that the same or even better results can be obtained with less data, and the model calculation results are highly accurate, which greatly reduces the complexity of the performance evaluation of artificial wetlands. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] The disclosure of the present invention is described with reference to the accompanying drawings. It should be understood that the drawings are only for illustrative purposes and are not intended to limit the scope of protection of the present invention. In the drawings, the same reference numerals are used to refer to the same components. Among them:

[0056] Figure 1 It is a schematic diagram of the flow chart of the method for evaluating the performance of artificial wetland ecosystem of the present invention. DETAILED DESCRIPTION

[0057] It is easy to understand that according to the technical solution of the present invention, without changing the essential spirit of the present invention, a person skilled in the art can propose a variety of interchangeable structural modes and implementation modes. Therefore, the following specific implementation modes and drawings are only exemplary descriptions of the technical solution of the present invention, and should not be regarded as the whole of the present invention or as a limitation or restriction to the technical solution of the present invention.

[0058] The present invention provides a method for evaluating the performance of an artificial wetland ecosystem. Figure 1 As shown, the following steps are included:

[0059] Step 1: Determine the optimization index for artificial wetland ecosystem assessment and verify that the optimized index can replace the original index;

[0060] 1.1: Determine the optimization index

[0061] Specifically, the determination of the optimization index for the artificial wetland ecosystem assessment includes processing the data of the original index, and selecting the index that best evaluates the performance of the artificial wetland as the optimization index through principal component analysis and correlation analysis;

[0062] Specifically, we first extract the main components through principal component analysis and dimensionality reduction, and further analyze the weights of the indicators in these principal components to select the most important indicators. Secondly, we perform correlation analysis by calculating the Spearman rank correlation coefficient to capture the nonlinear relationship and find the indicators with the strongest correlation. Finally, we select the indicators included in both the principal component analysis and the correlation analysis as optimization indicators.

[0063] (1.11) Principal component analysis

[0064] Principal component analysis is a common dimensionality reduction technique, which is mainly used to convert high-dimensional data into low-dimensional data while retaining the main information of the data as much as possible. The core idea is to map the original data to a new coordinate system through linear transformation, so that the axes of the new coordinate system (i.e., principal components) are orthogonal to each other, and these principal components are sorted from large to small according to the variance of the data. The first few principal components contain most of the information in the data, so they can be used to approximate the original data to achieve the purpose of dimensionality reduction.

[0065] (1): Each indicator is standardized so that its mean is 0 and its standard deviation is 1. When both qualitative and quantitative data are included, the weighted score is standardized. When only quantitative data are included, the water quality indicator is standardized.

[0066] The normalization formula is:

[0067]

[0068] in:

[0069] Where Y ij is the original value of the i-th sample under the j-th index, is the mean of the jth index, S j is the standard deviation of the jth indicator, and m represents the number of samples. ij is the standardized value of the i-th sample under the j-th index.

[0070] (2): Calculate the covariance matrix of the standardized data. For the standardized data matrix Z, the covariance matrix C is:

[0071]

[0072] In the formula, Z T is the transposed matrix of Z.

[0073] (3): Calculate the eigenvalues ​​and eigenvectors from the covariance matrix C.

[0074] Cv K =λ k v k

[0075] In the formula, λ k is the kth eigenvalue, v k is the corresponding eigenvector; C is the covariance matrix.

[0076] (4): Select the principal component

[0077] The principal components are selected based on the size of the eigenvalues, and the principal components that can explain most of the variance are selected. The eigenvalues ​​are sorted from large to small, and the eigenvectors corresponding to the first k eigenvalues ​​constitute the principal components.

[0078] (5): Construct principal components and project the standardized data onto the selected principal components.

[0079] T=ZV k

[0080] Where T is the new principal component matrix, V k is a matrix consisting of the selected k eigenvectors.

[0081] (6): Analyze the principal components and analyze the load of each principal component (i.e., the coefficient of the eigenvector) to understand the contribution of the original index. The principal component loading matrix P can be obtained from the eigenvector.

[0082] P=V k

[0083] Where P is the principal component loading matrix, where each column corresponds to a principal component and each row corresponds to an original indicator.

[0084] The importance of the original index to the principal component is judged by the size of the coefficient value of the eigenvector. The larger the coefficient value, the greater the influence of the original index on the principal component. On the contrary, the smaller the coefficient value, the less the original index can reflect the characteristics of the principal component. Taking the coefficient value of 0.5 as the screening principle, the index with a coefficient value greater than 0.5 in each principal component is selected as the optimization index 1.

[0085] (1.12) Correlation analysis

[0086] The Spearman rank correlation coefficient is calculated to determine the strength of the relationship between two variables. It is based on the order of the data rather than the original value. The value of the correlation coefficient ranges between -1 and 1. The closer the value is to 1 or -1, the stronger the relationship between the two variables; closer to 0 means a weaker relationship. A positive correlation is a correlation coefficient close to 1, and a negative correlation is close to -1.

[0087] (1): Data sorting

[0088] Sort all the data in the two variables x and y separately and assign a rank (R) to each data. If the data have the same value, assign the same average rank to these values. Rank is the position number after the data is sorted, which is used to evaluate the relative position of the data in the group. The purpose of calculating the rank is to eliminate the dimension and unit differences in the original data and only retain the sorting information of the data, so as to make the analysis more robust. Among them, when it contains both qualitative data and quantitative data, x and y represent the quantitative data and weighted score respectively. When it contains only quantitative data, x and y represent the quantitative data and closeness respectively.

[0089] (2): Calculate the rank difference

[0090] For each pair of data (xi, yi), calculate their rank difference d i =R(x i )-R(y i )

[0091] (3): Calculate the Spearman rank correlation coefficient

[0092]

[0093] Where ρ is the Spearman rank correlation coefficient, and m represents the number of samples. The value range of ρ is [-1, 1].

[0094] The index with a correlation coefficient value ρ greater than 0.5 is the optimization index 2.

[0095] The water quality index that exists in both optimization index 1 and optimization index 2 is selected as the optimization index.

[0096] 1.2. And verify that the optimized indicators can replace the original indicators

[0097] 1.21. Determine the performance value of artificial wetlands;

[0098] Specifically, artificial wetlands often contain qualitative data (including but not limited to ecological benefits, social benefits, economic benefits, and operation and maintenance) and quantitative data (including but not limited to the concentrations of ammonia nitrogen, nitrate nitrogen, nitrite nitrogen, total nitrogen, total phosphorus, and chemical oxygen demand in the inlet and outlet water). The analytic hierarchy process is used to calculate the weights of data containing both qualitative and quantitative data. The performance of artificial wetlands can be judged based on the weighted scoring results of the analytic hierarchy process. If only quantitative data are included, the entropy weight method is used to calculate the weights. Based on the weights, the TOPSIS standardized evaluation matrix is ​​constructed to obtain the value (T) for evaluating the performance of artificial wetlands. The calculated value (T) is named closeness. The closeness ranges from 0 to 1, which can evaluate the excellent performance of artificial wetlands. Closeness can be divided into 5 levels, namely 0-0.2 (Ⅰ), 0.2-0.4 (Ⅱ), 0.4-0.6 (Ⅲ), 0.6-0.8 (Ⅳ), and 0.8-1.0 (Ⅴ). The larger the closeness value, the better the performance of the artificial wetland.

[0099] 1.211: Use the analytic hierarchy process to calculate weights for qualitative and quantitative indicators

[0100] (1) Establishing a hierarchical model

[0101] The target layer is the performance evaluation of artificial wetlands, the criterion layer is water quality, ecological benefits, landscape value, economic benefits and management level, and the indicator layer is specific calculation indicators, including: the concentrations of ammonia nitrogen, nitrate nitrogen, nitrite nitrogen, total nitrogen, total phosphorus and chemical oxygen demand in inlet and outlet water, biodiversity index, vegetation coverage, landscape value, construction cost, operating cost, management level and maintenance frequency.

[0102] (2) Constructing the judgment matrix at the criterion level and the indicator level based on the expert scores

[0103] An expert scoring questionnaire for artificial wetland performance evaluation was designed and sent to experts via email. After obtaining the scoring results (scoring results range from 1 to 100), the average value of each calculation indicator was then calculated to construct a judgment matrix for each level.

[0104] The form of the judgment matrix is:

[0105]

[0106] In the formula, n represents the index, A is the judgment matrix

[0107] (3) Calculate the weight vector

[0108] (3.1) Find the eigenvalue and weight vector

[0109] AW=λW

[0110] Where W = [w1, w2…wn ] T , W is the weight vector, and λ is the corresponding eigenvalue. This formula can be used to calculate the maximum eigenvalue λ max And its corresponding weight vector is the final weight vector.

[0111] (3.2) Normalized weight vector

[0112] In order to make the sum of all weights of the feature vector W equal to 1, normalization is required

[0113]

[0114] In the formula, w j is the normalized weight.

[0115] (4)Consistency test

[0116]

[0117] In the formula, CI is an indicator used to measure the consistency of the judgment matrix, λ max is the maximum eigenvalue of the judgment matrix, n is the number of indicators. CR is the consistency ratio, and RI is the random consistency index, which is determined according to the matrix dimension n. If CR < 0.1, the judgment matrix has acceptable consistency.

[0118] (5) Calculate the overall weight

[0119] w A =w 准则 × 指标

[0120] In the formula, w A is the weight calculated by the hierarchical analysis method, w 准则 is the weight of the criterion layer, w 指标 is the weight of the indicator layer.

[0121] (6) Calculate weighted score

[0122] W ij =S ij ×w′ j

[0123] Where W ij is the weighted score of the i-th evaluation object on the j-th indicator, S ij is the expert score of the i-th evaluation object on the j-th indicator.

[0124] The weighted score ranges from 1 to 100, and the closer the value is to 100, the better the performance of the constructed wetland.

[0125] 1.222: Calculate the weights of quantitative indicators using the entropy weight method

[0126] (1), normalization:

[0127] First, the quantitative data of artificial wetlands need to be normalized. The purpose of this is to compare and analyze indicators of different dimensions and finally map them to [0, 1] for processing. There are calculations of positive and negative indicators in the data. The larger the value of a positive indicator, the better the indicator; on the contrary, the smaller the value of a negative indicator, the better the indicator. The calculation formula is as follows:

[0128] Positive indicators:

[0129]

[0130] Negative Indicators:

[0131]

[0132] Where: X i and Y i are the original value and the normalized value respectively; X min and X max are the minimum and maximum values ​​in the sample.

[0133] (2) Entropy weight method:

[0134] Entropy is a measure of the disorder of a system. The smaller the information entropy of an indicator, the greater the variation of its indicator value, which means that the indicator provides more information, plays a greater role in the comprehensive evaluation, and has a higher weight; vice versa. In practical applications, entropy can be used to calculate the weight of each indicator based on the variation of each indicator value, and a more objective weight value can be obtained.

[0135] (2.1): Using the index data of each partition as a column vector, a matrix containing m evaluation objects and n evaluation indicators is constructed.

[0136] R=(Y ij ) m×n

[0137] In the formula, R is the evaluation matrix, Y ij is the value of the i-th evaluation object on the j-th indicator.

[0138] (2.2): Calculate the characteristic weight of the j evaluation indicators in the i-th evaluation object.

[0139]

[0140] Where P ij is the characteristic weight.

[0141] (2.3): Calculation of indicator entropy value.

[0142]

[0143] In the formula, e j is the entropy value of the jth index, if P ij =0,P ij oeLh ij =0.

[0144] (2.4): Calculation of indicator weights.

[0145]

[0146] In the formula, w j is the weight of j indicators.

[0147] (3) Construct the TOPSIS standardized evaluation matrix:

[0148] (3.1) Construct a standardized evaluation matrix (R):

[0149]

[0150] (3.2) Calculate the positive and negative ideal solutions and the Euclidean distance.

[0151]

[0152] In the formula, R+ is a positive ideal solution, which represents the optimal value of each indicator in the evaluation system, that is, the ideal value of the maximization target. R- is a negative ideal solution, which represents the worst value of each indicator in the evaluation system, that is, the ideal value of the minimization target.

[0153]

[0154] In the formula, D+ is the Euclidean distance between the i-th evaluation object and the positive ideal solution, which indicates the gap between the evaluation object and the optimal solution. D- is the Euclidean distance between the i-th evaluation object and the negative ideal solution, which indicates the gap between the evaluation object and the worst solution.

[0155] (3.3) Calculate the closeness of each indicator to the optimal solution (T)

[0156]

[0157] Where T is the closeness between the ith evaluation object and the optimal solution, reflecting the relative superiority of the evaluation object among all evaluation objects.

[0158] 1.22. Construct a gradient boosting tree model between the original index, the optimized index and the constructed wetland performance value, use the training set to train the two models respectively, predict the test sets of the two trained models, and obtain the prediction results;

[0159] Construct a gradient boosting tree model between the original index, the optimized index and the closeness, use the training set to train the two models respectively, predict the test set of the two trained models, and get the prediction results. Verify the results by comparing the mean square error and R2 between the original index model and the optimized index model.

[0160] (I): Divide the data set

[0161] Original data: {X orig , Y orig}

[0162] When it contains both qualitative and quantitative data, X orig is the quantized data matrix, Y orig is a weighted score. When only quantitative data are included, X orig is the original water quality index matrix, Y orig is the original closeness value.

[0163] Optimization data: {X new , Y orig}, X new is the optimization indicator matrix.

[0164] Training set: {X train , Y train}.

[0165] Test set: {X test , Y test}.

[0166] (II): Training the Gradient Boosting Tree Model

[0167] 2.1. Training a model based on raw data

[0168] (1) Initialize the model:

[0169]

[0170] In the formula, is the predicted value based on the original model, is the predicted value based on the original model, N orig is the number of samples in the original dataset, y orig，i is the target value of the i-th sample in the original dataset.

[0171] (2) Training iteration process:

[0172] Compute the residual:

[0173] In the formula, is the residual of the i-th sample in the m-th round, y orig，iis the weighted score or raw closeness value of the ith sample, x orig，i is the feature vector of the i-th sample, is the predicted value of the model in the m-1th round for the i-th sample.

[0174] Training a decision tree:

[0175] In the formula, is the loss function of decision tree training, usually the sum of squares of residuals, is the squared residual of the ith sample.

[0176] Update the model:

[0177] In the formula, is the model prediction value of the mth round, is the model prediction value of the m-1th round, η is the learning rate, which controls the contribution of each tree to the model, is the predicted value of the mth tree for the input x.

[0178] The final model's predictions are:

[0179] In the formula, is the final model prediction based on the original data, B is the total number of trees, is the prediction function of the bth tree, X orig，train is the original quantized data in the training set.

[0180] 2.2. Training a model based on optimized data

[0181] (1) Initialize the model

[0182]

[0183] In the formula, is the predicted value based on the optimization model, is the predicted value based on the optimization model, N new is the number of samples in the optimized dataset.

[0184] (2) Training iteration process:

[0185] Compute the residual:

[0186] In the formula, is the residual of the i-th sample in the m-th round, x new，i is the feature vector of the i-th sample,

[0187] is the predicted value of the model in the m-1th round for the i-th sample.

[0188] Training a decision tree:

[0189] In the formula, is the loss function of decision tree training, usually the sum of squares of residuals, is the squared residual of the ith sample.

[0190] Update the model:

[0191] In the formula, is the model prediction value of the mth round, is the model prediction value of the m-1th round, η is the learning rate, which controls the contribution of each tree to the model, is the predicted value of the mth tree for the input x.

[0192] The final model's predictions are:

[0193] In the formula, is the final model prediction value based on the optimized data, B is the total number of trees, is the prediction function of the bth tree, X new，train is the optimization indicator in the training set.

[0194] (III) Use the trained model to predict the test set

[0195] Make predictions on the original data test set:

[0196]

[0197] In the formula, is the predicted value based on the original model on the test set, X test is the original metric in the test set.

[0198] Make predictions on the optimized data test set:

[0199]

[0200] In the formula, is the predicted value of the model based on the optimization index on the test set, X new，test is the optimization metric in the test set.

[0201] (IV): Calculation of evaluation indicators

[0202] Mean Square Error (MSE)

[0203]

[0204] Y test，i is the true value of the i-th sample in the test set, is the predicted value of the i-th sample in the test set based on the original model, is the predicted value of the i-th sample in the test set based on the optimized model.

[0205] Coefficient of determination (R2)

[0206]

[0207] 1.23. The results were verified by comparing the mean square error and determination coefficient between the original indicator model and the optimized indicator model.

[0208] If MSE new Lower than MSE orig at the same time Higher than This means that the optimized indicator can effectively replace the original indicator.

[0209] Step 2: construct a logistic regression model whose input data is the optimization index and the evaluation value, obtain a simulation equation with the optimization index as the independent variable and the evaluation value as the dependent variable; and verify the accuracy of the simulation equation;

[0210] The logistic regression model is used to construct the models between the optimization index and the weighted score and the closeness, and the equation expression with the optimization index as the independent variable and the weighted score and the closeness as the dependent variable is obtained. The accuracy is verified by linear fitting. If the R2 of the linear fitting is greater than 0.8, it means that the simulation equation can be used.

[0211] Step 1: Build the dataset

[0212] Data matrix X:

[0213]

[0214] In the formula, n is the optimized index and m is the number of samples.

[0215] Label vector y: represents the weighted score or closeness, ranging from 0 to 1.

[0216]

[0217] Step 2: Divide the dataset into training and testing sets, with 80% of the data used for training and 20% for testing.

[0218] Training set:

[0219]

[0220] Test set:

[0221]

[0222] Step 3: Train a logistic regression model

[0223] Using training set X train and train Train the logistic regression model. The goal of the model is to find the parameter β so that the predicted value is consistent with the true value to the greatest extent. The hypothesis function of the logistic regression model is:

[0224]

[0225] is the hypothesis function of the logistic regression model, X n is the nth eigenvalue.

[0226] Step 4: Model prediction

[0227] Use the parameters β obtained from the training set to test the data set X test Make predictions and get the predicted value of each test sample The prediction formula is:

[0228]

[0229] Step 5: Model Evaluation

[0230] Use the predicted values ​​on the test set and the true weighted score or closeness y test The evaluation was carried out and the explanatory power of the model was evaluated by the coefficient of determination R2, which was calculated as follows:

[0231]

[0232] In the formula, Yes test If R2 is greater than 0.8, it means that the predicted value can fit the true value well;

[0233] A simulation equation that can easily evaluate the performance of artificial wetlands was developed and verified.

[0234] 5.1 The simulation equation is as follows:

[0235] F=β0+β1X1+β2X2+…+β n X n

[0236] Where β0 is the intercept, X1, X2...X n Represents the optimization index, β1, β2...β n is the parameter corresponding to the optimization indicator. F is the calculated weighted score or closeness.

[0237] 5.2 Perform a linear fit between the calculated value and the original value of the simulation equation to obtain the determination coefficient R2 value. If R2 is greater than 0.8, it means that the simulation equation can be used to simplify the calculation of artificial wetland performance evaluation.

[0238]

[0239] In the formula, y i is the original weighted score or closeness value, is the calculated value of the simulation equation, is the mean of the original closeness.

[0240] Implementation example: Obtaining experimental data on the performance of a product or result:

[0241] The data in Table 1 are the comparison between the closeness values ​​calculated by the entropy weight method combined with TOPSIS when there are only quantitative data and the closeness values ​​calculated by the simulation equation. According to Table 1, it can be concluded that the determination coefficient R2 of the closeness values ​​calculated by the two methods is 0.88, which is greater than the specified value of 0.8. Therefore, it can be shown that the simulation equation can replace the entropy weight method to calculate the closeness value.

[0242] Table 1

[0243]

[0244]

[0245]

[0246] Step 3: Calculate the performance estimation value based on the simulation equation after verification; and evaluate the performance of the artificial wetland ecosystem based on the evaluation value.

[0247] The present invention combines correlation analysis with the data to reduce the dimension by principal component analysis, extracts the eigenvector with larger coefficient value in the principal component as an important indicator, and analyzes the indicators with stronger correlation with closeness or weighted score in the data from multiple angles, so as to achieve the purpose of calculating the screening indicators instead of the original indicators. And the rationality of the selected optimization indicators is verified in the gradient boosting tree model. This is different from the previous technology, which only uses principal component analysis to reduce the dimension of the data set without considering the influence of the eigenvector coefficient value on the principal component. The correlation analysis used at the same time also further strengthens the reliability and persuasiveness of the screening results of the indicators. In addition, the present invention uses a logistic regression model to construct a linear simulation equation of artificial wetland indicators and closeness or weighted score, which greatly reduces the complexity of evaluating the performance of artificial wetlands, and can obtain results that are more consistent with the original values. The uniqueness of the present invention is that the indicators required for the performance evaluation of artificial wetlands are reduced, which means that the same or even better results can be obtained with less data, and the model calculation results are highly accurate, which greatly reduces the complexity of the performance evaluation of artificial wetlands.

[0248] Compared with the content of the background technology, the present invention also uses the hierarchical analysis method, but further screens and verifies various indicators of artificial wetlands, simplifies the number of indicators used in the performance evaluation of artificial wetlands, and can also achieve the same results as those obtained by the original evaluation method. And on the basis of simplifying the indicators, the performance of artificial wetlands is modeled, rather than just performing preliminary calculations through the fuzzy membership function method.

[0249] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this article.

[0250] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0251] In the several embodiments provided herein, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, or it can be an electrical, mechanical or other form of connection.

[0252] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the embodiments of this article.

[0253] In addition, each functional unit in each embodiment of this invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above integrated unit may be implemented in the form of hardware or in the form of software functional unit.

[0254] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this article is essentially or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of this article. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program code.

[0255] Specific embodiments are used in this article to illustrate the principles and implementation methods of this article. The description of the above embodiments is only used to help understand the methods and core ideas of this article. At the same time, for general technicians in this field, according to the ideas of this article, there will be changes in the specific implementation methods and application scopes. In summary, the content of this specification should not be understood as a limitation on this article.

Claims

1. A method for evaluating the performance of an artificial wetland ecosystem, characterized in that: The following steps are involved: Step 1: Determine the optimization index for artificial wetland ecosystem assessment and verify that the optimized index can replace the original index; Determining the optimization index for artificial wetland ecosystem assessment includes processing the data of the original index, and selecting the index that best evaluates the performance of the artificial wetland as the optimization index through principal component analysis and correlation analysis; Step 2: construct a logistic regression model whose input data is the optimization index and the evaluation value, obtain a simulation equation with the optimization index as the independent variable and the evaluation value as the dependent variable; and verify the accuracy of the simulation equation; Step 3: Calculate the performance estimation value based on the simulation equation after verification; and evaluate the performance of the artificial wetland ecosystem based on the evaluation value.

2. The method for evaluating the performance of an artificial wetland ecosystem according to claim 1, characterized in that: The index that best evaluates the performance of the artificial wetland is selected as the optimization index through principal component analysis and correlation analysis; include The principal component analysis extracts the main components of the data from the dimensionality reduction, analyzes the weights of the indicators in the main components, and determines the important indicators; The correlation analysis uses the calculation of the Spearman rank correlation coefficient to capture the linear relationship and determine the strongest indicator; The indicators that include both important indicators and the strongest indicators are used as optimization indicators.

3. The method for evaluating the performance of an artificial wetland ecosystem according to claim 1, characterized in that: The verifying that the optimized indicator can replace the original indicator includes: Determine constructed wetland performance values; Construct a gradient boosting tree model between the original index, the optimized index and the constructed wetland performance value, use the training set to train the two models respectively, and predict the test sets of the two trained models to obtain the prediction results; The results were validated by comparing the mean square error and coefficient of determination between the original indicator model and the optimized indicator model.

4. The method for evaluating the performance of an artificial wetland ecosystem according to claim 3, characterized in that: Determining the performance value of the artificial wetland includes: If the artificial wetland ecosystem contains data on both qualitative and quantitative indicators, the weights are calculated using the analytic hierarchy process, and the performance value of the artificial wetland is determined based on the weighted scoring results of the analytic hierarchy process; If the artificial wetland ecosystem only contains data on quantitative indicators, the entropy weight method is used to calculate the weights. Based on the weights, the TOPSIS standardized evaluation matrix is ​​constructed to obtain the values ​​for evaluating the performance of artificial wetland treatment and determine the performance value of the artificial wetland.

5. The method for evaluating the performance of an artificial wetland ecosystem according to claim 4, characterized in that: The qualitative indicators include ecological benefits, social benefits, economic benefits and operation and maintenance; The quantitative indicators include the concentrations of ammonia nitrogen, nitrate nitrogen, nitrite nitrogen, total nitrogen, total phosphorus and chemical oxygen demand in the inlet and outlet water.

6. The method for evaluating the performance of an artificial wetland ecosystem according to claim 1, characterized in that: The constructing of a logistic regression model with the optimization index and the evaluation value as input data, and obtaining a simulation equation with the optimization index as an independent variable and the evaluation value as a dependent variable, comprises: 2.

1. Building a Dataset Data matrix X: In the above formula, n is the optimized index and m is the number of samples; Label vector y: represents the weighted score or closeness, ranging from 0 to 1; 2.

2. Divide the data set into training set and test set, 80% of the data is used for training and 20% of the data is used for testing; Training set: Test set: 2.

3. Training Logistic Regression Model Using training set X train and train Train the logistic regression model; the goal of the model is to find the parameter β so that the predicted value is consistent with the true value to the greatest extent; the hypothesis function of the logistic regression model is: is the hypothesis function of the logistic regression model, X n is the nth eigenvalue; 2.4 Model Prediction Use the parameters β obtained from the training set to test the data set X test Make predictions and get the predicted value of each test sample The prediction formula is:

7. The method for evaluating the performance of an artificial wetland ecosystem according to claim 1, characterized in that: The verification of the accuracy of the simulation equation includes fitting the simulation value and the true value to verify the accuracy of the simulation equation.

8. The method for evaluating the performance of an artificial wetland ecosystem according to claim 7, characterized in that: The accuracy of the simulation equation is verified by fitting the simulation value and the true value. The accuracy of the simulation equation is verified. The simulation equation is as follows: F=β0+β1X1+β2X2+…+β n X n Where β0 is the intercept, X1, X2...X n Represents the optimization index, β1, β2...β n is the parameter corresponding to the optimization index; F is the calculated weighted score or closeness; The calculated value and the original value of the simulation equation were linearly fitted to obtain the determination coefficient R2 value. If R2 is greater than 0.8, it means that the simulation equation can be used to simplify the calculation of artificial wetland performance evaluation; In the formula, y i is the original weighted score or closeness value, is the calculated value of the simulation equation, is the mean of the original closeness.

9. An artificial wetland ecosystem performance evaluation device, characterized in that: It comprises a processor and a memory; the memory is used to store a program; the processor executes the program to implement the method as described in any one of claims 1-8.

10. A computer-readable storage medium, characterized in that: The storage medium stores a program, and the program is executed by a processor to implement the method according to any one of claims 1 to 8.

Images