Water quality evaluation WQI expert weight fusion method based on weighted extremum fusion algorithm

Abstract

The invention discloses a water quality evaluation WQI expert weight fusion method based on a weighted extremum fusion algorithm, and relates to the technical field of water quality evaluation WQI. The method comprises the following steps: designing a single-factor evaluation algorithm; the importance degree of the water quality index is described by using three dimensions, and three different weighting methods are provided for the three dimensions; an algorithm, namely a weighted extremum fusion algorithm, which can fuse three weighting methods and realize subjective and objective combination is provided; by comprehensively considering the uncertainty degree of the water quality data, the index contribution degree and the regional key degree, the more scientific water quality evaluation method is provided, the method not only can accurately reflect the water quality condition, but also can provide a large number of adjustable parameters so that a user can carry out optimization according to specific requirements, and the user experience is improved. Therefore, water quality evaluation requirements of different regions can be better met, and the overall evaluation effect and practical value are improved.

Description

Technical Field

[0001] The invention relates to the technical field of water quality assessment (WQI), and in particular to a water quality assessment (WQI) expert weight fusion method based on a weighted extreme value fusion algorithm. Background Art

[0002] At present, the worst index evaluation method is usually used to determine the water quality category, but this method is difficult to make full use of the measured results of various water quality indicators, and has the limitation of low information utilization. In many cases, the degree of water quality deterioration between water bodies of the same category is significantly different. How to scientifically and reasonably design a comprehensive index algorithm to describe the water quality status of a water body at a certain moment is an issue that needs to be standardized. At present, the main comprehensive water quality evaluation methods include: pollution index method, fuzzy mathematics method, gray system evaluation method, hierarchical analysis evaluation method, matter element analysis evaluation method, neural network method, water quality identification index method, etc. Among the many evaluation methods, the pollution index method with index empowerment as the core idea is relatively clear and can intuitively reflect the comprehensive evaluation process; although the pollution index method is simple, it still needs to be improved in practical application. my country has a vast land area and many water bodies, and the water body characteristics of various regions are significantly different. For the water quality evaluation of different water bodies, the water quality indicators that need to be considered are different, and the importance of each indicator is also different. Therefore, when constructing a comprehensive water quality evaluation system based on the pollution index method, after screening out the local characteristic water quality indicators, the weight value of each pollution indicator should be given according to the specific local conditions.

[0003] However, the determination of indicator weights not only depends on the quality and instability of the water body's own indicator evaluation, but also on the pollution characteristics of the local water body itself. In many cases, the statistical values ​​of objective water quality data cannot fully explain the importance of water quality indicators. Some important water quality indicators may reflect statistical characteristics that do not meet the requirements of poor quality indicators. In order to improve the evaluation weights of these important water quality indicators in the pollution index method, the present invention achieves this goal by introducing expert intervention, that is, the generation of subjective intervention weights. How to scientifically give experts who understand the local water quality conditions a certain amount of intervention weight while retaining the objective weighting method. In response to the above problems, the inventors proposed a water quality evaluation WQI expert weight fusion method based on a weighted extreme value fusion algorithm to solve the above problems. Summary of the invention

[0004] In order to solve the problem that indicator weights need to be combined with water body characteristics and expert opinions, balance objective data and subjective intervention, and improve evaluation accuracy by scientific weighting; the purpose of the present invention is to provide a water quality evaluation WQI expert weight fusion method based on a weighted extreme value fusion algorithm.

[0005] To solve the above technical problems, the present invention adopts the following technical solutions: a water quality evaluation WQI expert weight fusion method based on a weighted extreme value fusion algorithm. This evaluation method can combine the importance of three different aspects of water quality indicators: indicator contribution, indicator uncertainty, and indicator criticality to the region. At the same time, it allows users to select different model parameters according to the characteristics of the study area to adapt to the evaluation needs.

[0006] To achieve the above content, the present invention combines the fixed weight algorithm that integrates seasonal decomposition and entropy weight method with the variable weight algorithm of real-time pollution contribution rate to construct a comprehensive variable weight algorithm that integrates the expert system empowerment system - weighted extreme value fusion method, and applies it to the comprehensive water quality evaluation system to achieve the combination of subjective and objective water quality evaluation, which can be achieved by the following steps:

[0007] 1. Six-point single factor scoring of water quality

[0008] The single factor evaluation method adopted in the present invention is based on the range of various water indicators specified in the "Surface Water Environmental Quality Standard" (GB3838-2002), and performs different mappings according to the characteristics of the indicators, and finally converts all indicator values ​​into a six-point evaluation value.

[0009] (1) For the basic standard limits of surface water environmental quality standards other than pH and water temperature specified in the Surface Water Environmental Quality Standards (such as permanganate index (COD Mn ), total nitrogen (TN), etc.), the present invention uses linear interpolation to map indicators, and the score values ​​corresponding to each indicator node are shown in Table 1:

[0010] Table 1 Mapping relationship between water quality categories and corresponding single factor scores

[0011]

[0012] (2) For indicators not specified in the Surface Water Environmental Quality Standard (such as chlorophyll a, etc.), the present invention adopts a specific problem-by-specific analysis approach. The calculation of the chlorophyll a single factor score is described in detail in the specific implementation method section.

[0013] (3) For the pH index, the pH evaluation curve designed by the present invention is an "Ω"-shaped curve with high sides and low center. According to the "Surface Water Environmental Quality Standards", the pH value range of normal water bodies should be 6 to 9, and the most suitable pH range for nitrogen fixation and nitrification in water bodies is a slightly alkaline pH range of 7.5 to 8.5. Therefore, this study intends to use a segmented S-shaped curve as the evaluation curve, and the selected S-shaped curve expression is Where ai and bi are adjustable parameters; when x=6, 7.5, 8, 8.5, 9, y=3, 0.5, 0, 0.5, 3, parameter Ai , B i , C i Obtained by solving the system of equations.

[0014] 2. "Fixed Weight"

[0015] The seasonal decomposition model used in this invention is the STL model proposed by RBCleveland in 1990 [1]. The average daily series of each indicator in the past few years are input into the seasonal regression model, and the three outputs of trend, seasonal fluctuation and residual are obtained. The absolute value series of the sum of seasonal fluctuation and residual is input into the entropy weight method model, and the information entropy of each indicator is obtained. The entropy weight method formula is as follows:

[0016]

[0017] The p in the above information entropy formula is ij That is, the probability of the event in the i-th row and j-th column of the matrix, k is a constant, usually taken as

[0018] e i is the entropy value of the ith indicator, and n is the number of sample periods.

[0019] Finally, the information entropy is normalized and enhanced to obtain the "fixed weight" part. The enhancement formula is as follows:

[0020]

[0021] in is the weight vector output by the entropy weight method at time i, m is the number of indicators, e is the entropy weight, and η is the enhancement coefficient.

[0022] 3. "Variable weight"

[0023] The normalization method is used to scale the single factor evaluation value of the water quality index at each moment to the interval [0, 1]. The formula of the normalization method is described as follows:

[0024]

[0025] in, The weight vector output by the normalization method at the i-th moment, c i is the contribution value of the ith indicator, and m is the number of indicators.

[0026] The output result is input into the water quality evaluation system as a "variable weight" result.

[0027] 4. “Expert Weight”

[0028] In the present invention, the importance of water quality indicators is divided into 10 levels, of which 1 to 3 are secondary indicators, 4 to 6 are general indicators, and 7 to 10 are important indicators. Experts who understand the characteristics of local water bodies score the importance of each water quality indicator, and use a visual UI to output the expert scores to the water quality evaluation system, and finally generate subjective weights through normalization. The visual UI is attached. Figure 1 .

[0029] 5. Weighted extreme value fusion algorithm

[0030] The calculation formula of the algorithm is as follows:

[0031]

[0032]

[0033]

[0034]

[0035]

[0036] In the formula: i represents the current moment, j represents the current indicator component, n represents the number of samples in the time period, and m represents the number of indicators; is the final output weight vector at the i-th moment, is the unnormalized weight vector; ENT, CTB, and EXP represent the three algorithms of entropy weight method, pollution rate contribution method, and expert system method, respectively. is the weight vector output by the three weighting methods, are the weighted values ​​of the three weight vectors respectively, is the Hadamard product of the matrix, and the result is a matrix consisting of the product of elements by elements; additional terms middle The vector is an extreme value vector, and each of its components takes the maximum value of the corresponding index components of the three methods ( Representation vector The maximum value of each component), The vector is the weighted value of the extreme value vector, and each component has three values, which are λ1, λ2, and λ3 respectively when the corresponding indicator components are ENT, CTB, and EXP.

[0037] The weights of the three different angles are input into the model constructed based on the weighted extreme value fusion algorithm formula, and the weight proportions of the three parts of "fixed weight", "variable weight" and "expert weight" at each moment are calculated.

[0038] 6. Weight fusion and comprehensive scoring

[0039] Combined with the above weighted extreme value model output With the single factor evaluation output x ij (i represents the current moment, j represents the current indicator component), and the six-point water quality score at each moment is calculated using the following formula:

[0040]

[0041] Where: I i is a six-point rating, ω ij 、x ij are the weight output and single indicator evaluation of the jth indicator at the i-th moment, respectively, and m is the number of indicators. Convert it into a percentage WQI, the formula is as follows:

[0042]

[0043] Where: ε is the transformation parameter. When its value is (0, 1), the WQI~I curve is a convex function. Conversely, when its value is (1, +∞), the curve is a concave function.

[0044] Preferably, all parameters involved in the present invention are: η, λ1, λ2, λ3, ε, a1, a2, b1, b2. Their meanings are: entropy weight method enhancement coefficient, "fixed weight" basic weighting, "variable weight" basic weighting, "expert weight" basic weighting, "fixed weight" extreme value weighting, "variable weight" extreme value weighting, "expert weight" extreme value weighting, six-point system to percentage system conversion parameter, pH score curve left half parameter 1, pH score curve right half parameter 1, pH score curve left half parameter 2, pH score curve right half parameter 2. These parameters can be modified according to the actual situation of the study area.

[0045] Compared with the prior art, the present invention has the following beneficial effects:

[0046] 1. The present invention provides a more scientific water quality evaluation method by comprehensively considering the uncertainty of water quality data, the contribution of indicators and the criticality of regions. This method can not only accurately reflect the water quality status, but also be flexibly adjusted according to the characteristics of different regions to ensure the practicality and reliability of the evaluation results. In addition, a large number of adjustable parameters are provided so that users can optimize according to specific needs, thereby better adapting to the water quality evaluation requirements of different regions and improving the overall evaluation effect and practical value;

[0047] 2. The algorithm involved in the present invention is highly flexible and can be implemented in a variety of programming languages, which facilitates seamless integration with existing online water quality evaluation systems. This adaptability means that users do not need to rebuild the system, thereby significantly reducing deployment costs. At the same time, the adjustable parameters in the algorithm have clear physical meanings, allowing users to make reasonable adjustments based on the actual water quality conditions in the region. This feature not only facilitates the formulation of unified standards for large-scale applications, but also lowers the application threshold, which helps promote the widespread popularization and application of this technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0049] Figure 1 This is a visual UI diagram of the expert system of the present invention;

[0050] Figure 2 It is a schematic diagram of the single factor evaluation results of 7 water quality indicators at two stations from January to June 2024 of the present invention;

[0051] Figure 3 This is a schematic diagram of seasonal STL analysis results of Mapengwan Station of the present invention;

[0052] Figure 4 This is a schematic diagram of seasonal STL analysis results of Lingtang Station of the present invention;

[0053] Figure 5 This is a line diagram of the WQI output results of the two sites from January to June 2024 of the present invention. DETAILED DESCRIPTION

[0054] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0055] like Figure 1-5 As shown, the present invention provides a water quality assessment WQI expert weight fusion method based on a weighted extreme value fusion algorithm, comprising the following steps:

[0056] S1. Combined with the national standards "Surface Water Environmental Quality Standard GB3838-2002", "Drinking Natural Mineral Water GB8637-2008", and relevant technical regulations "Evaluation Method and Classification Technical Regulations for Lake (Reservoir) Eutrophication" and other documents, a single factor evaluation is conducted on the real-time series and historical series of regional water quality indicators. Each evaluation time point is output as a six-point single factor score;

[0057] S2. Use seasonal regression model and entropy weight method to process the single factor evaluation results of regional historical series, normalize and enhance the processing results to obtain the "fixed weight" part;

[0058] S3. Use the pollution contribution rate method to process the single factor evaluation results of the regional real-time series to obtain the "variable weight" part;

[0059] S4. Use the normalization method to process the importance score results of each indicator in the region obtained from expert consultation to obtain the "expert weight" part;

[0060] S5. Use the multi-weight fusion algorithm of weighted extreme value fusion to process "fixed weight", "variable weight" and "expert weight" to obtain the final "comprehensive weight" for comprehensive water quality evaluation;

[0061] S6. Combine the calculated “comprehensive weight” with the single factor evaluation results of the regional real-time series to obtain a six-point comprehensive score series;

[0062] S7. Use the comprehensive evaluation curve to convert the six-point score into a 100-point score, and finally obtain the real-time comprehensive water quality score series for the region.

[0063] Table 2 Parameters of the embodiment

[0064]

[0065] 1. Six-point single factor scoring of water quality

[0066] The data sources used in the embodiment are the water quality monitoring station at the water intake of Lingtang Water Plant located in Lingtang Hui Township, Gaoyou Lake West New District, Gaoyou City, on the south side of Gaoyou Lake, and the Mapengwan water quality monitoring station located on the east side of Gaoyou Lake and adjacent to the Qingshuitan water intake of Liyunhe; the selected data time period is the pH value (pH), permanganate index (COD Mn The measured daily average data of seven indicators, including dissolved oxygen (DO), ammonia nitrogen (N-NH3), total nitrogen (TN), total phosphorus (TP), and chlorophyll a (Chl.a). Among them, the data from January 2019 to December 2023 are used for the calculation of fixed weights, and the data from January 2024 to May 2024 are used for the final indicator evaluation.

[0067] The Surface Water Environmental Quality Standard does not specify the chlorophyll a index. The "Eutrophication Evaluation Method and Classification Technical Regulations for Lakes (Reservoirs)" issued by my country's Ministry of Ecology and Environment in 2004 proposed the "Comprehensive Nutrition Status Index Method" for evaluating the eutrophication of lakes and reservoirs: chlorophyll a (Chl. a), total phosphorus (TP), total nitrogen (TN), transparency (SD), permanganate index (COD Mn ) Five indicators that directly reflect the degree of eutrophication of water bodies are converted into nutrient status indexes with the same numerical meaning through different calculation formulas, and the weighted average is used to obtain the comprehensive nutrient status index. (Weight W j The square of the correlation coefficient between the jth index and the chlorophyll a index is r j 2 The specific formula is as follows:

[0068]

[0069] TLI Chl.a =10×(2.5+1.086×lnρ Chl.a )

[0070] TLI TP =10×(9.436+1.624×lnρ TP )

[0071] TLI TN =10×(5.453+1.694×lnρ TN )

[0072] TLI SD =10×(5.118-1.94×lnρ SD )

[0073]

[0074] Where: TLI is the comprehensive nutritional status index. j is the nutritional status index of the jth indicator, r j 2 is the square of the correlation coefficient between the jth index and the chlorophyll a index, W j To calculate the weight of the comprehensive nutritional status index, ρ j is the evaluation value of the jth index (Chl.a unit is mg / m 3 , SD unit is m, and the other indicators unit is mg / L). The correlation and weight table of each indicator and chlorophyll a are shown in Table 3.

[0075] Table 3 Correlation between some indicators of lakes (reservoirs) in China and chlorophyll a j、r j 2 and W j value

[0076]

[0077] The National Ecological Environment Standard of the People's Republic of China - Technical Specifications for Surface Water Environmental Quality Assessment (Draft for Comments) divides water bodies of different water quality categories and eutrophication levels into different colors as shown in Table 4 (for the convenience of expression, The present invention is based on this and combines the TP, TN, and COD given in the "Surface Water Environmental Quality Standards". Mn The standard value of the single evaluation index corresponding to different Chl.a concentrations was deduced, and the deduction table is shown in Table 5.

[0078] Table 4 Comparison table of water quality categories, eutrophication levels and water quality characterization colors

[0079]

[0080] Table 5 Calculation table of chlorophyll a index values ​​at different evaluation nodes

[0081]

[0082]

[0083] Since the Surface Water Environmental Quality Standard does not provide an indicator standard for water transparency SD, the scoring standard for Chl.a in this paper is the result of the integration of four indicators. Like general indicators, the indicator values ​​between scoring nodes are mapped using linear interpolation.

[0084] The remaining indicators are evaluated according to the conventional single factor evaluation formula mentioned above. The final evaluation results are as follows: Figure 2 shown.

[0085] 2. "Fixed Weight"

[0086] First, a single factor evaluation was conducted on the annual average daily water quality process line from January 2019 to December 2023. Then, the evaluation results of each indicator (six-point score, total series duration of 1 year, step length of 1 day) were seasonally decomposed using the statsmodels package of Python. The model period was set to 24, and finally the trend, seasonal fluctuation and residual series of the target research period were obtained. The results are shown in Figures 3-4 .

[0087] The entropy weight method application written in Python is also used to analyze the absolute value series of seasonal fluctuations and the sum of residuals. After normalization and enhancement, the final calculated weights are shown in Table 6.

[0088] Table 6 Fixed weight model output table

[0089]

[0090]

[0091] 3. "Variable weight"

[0092] The variable weight series of the target evaluation period (January 1 to May 31, 2024) of the study area was calculated using the pollution contribution rate of the normalization method. Some results are shown in Table 7.

[0093] Table 7 Output table of variable weight model for some time periods

[0094]

[0095] 4. “Expert Weight”

[0096] The expert consulted by the present invention is Associate Professor Zhou Mi from the College of Hydrology and Water Resources of Hohai University. The expert scoring results of the 7 indicators are as follows: Figure 1 The expert weights obtained after normalization are shown in Table 8.

[0097] Table 8 Expert weight results table

[0098]

[0099] 5. Weighted extreme value fusion algorithm

[0100] The weighted extreme value algorithm was written in Python, and Excel was used as the data input and output carrier. The output results of the three weight models were input into the model to obtain the comprehensive weight sequence. The output results of some periods are shown in Table 9.

[0101] Table 9 Comprehensive weight output table for some time periods

[0102]

[0103]

[0104] 6. Weight fusion and comprehensive scoring

[0105] Calculate the Hadamard product (the matrix consisting of the product of the corresponding elements) of the single factor evaluation results and the matrix corresponding to the comprehensive weight during the evaluation period (January 1 to May 31, 2024) to obtain the six-point scoring result of the evaluation period, and use the conversion formula described in formula (6) to convert the six-point system into a 100-point WQI. See the comprehensive score for details. Figure 5 .

[0106] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A water quality assessment WQI expert weight fusion method based on weighted extreme value fusion algorithm, characterized in that: The following steps are involved: S1. Combined with relevant technical regulations and documents, single factor evaluation is conducted on the real-time and historical series of regional water quality indicators, and each evaluation time point is output as a six-point single factor score; S2. Use seasonal regression model and entropy weight method to process the single factor evaluation results of regional historical series, normalize and enhance the processing results to obtain the "fixed weight" part; S3. Use the pollution contribution rate method to process the single factor evaluation results of the regional real-time series to obtain the "variable weight" part; S4. Use the normalization method to process the importance score results of each indicator in the region obtained from expert consultation to obtain the "expert weight" part; S5. Use the multi-weight fusion algorithm of weighted extreme value fusion to process "fixed weight", "variable weight" and "expert weight" to obtain the final "comprehensive weight" for comprehensive water quality evaluation; S6. Combine the calculated "comprehensive weight" with the single factor evaluation results of the regional real-time series to obtain a six-point comprehensive score series; S7. Use the comprehensive evaluation curve to convert the six-point score into a 100-point score, and finally obtain the real-time comprehensive water quality score series for the region.

2. The water quality assessment WQI expert weight fusion method based on weighted extreme value fusion algorithm as claimed in claim 1, characterized in that: The specific method of single factor evaluation is to divide the indicators into three categories: basic items of surface water environmental quality standards other than pH and water temperature, unspecified indicators, and pH indicators. The first category uses linear interpolation to map indicators; the second category conducts specific analysis in combination with other national standards; the third category uses an "Ω"-shaped curve with high sides and a low middle.

3. The water quality assessment WQI expert weight fusion method based on weighted extreme value fusion algorithm as claimed in claim 1, characterized in that: The steps of calculating the "fixed weight" by the seasonal regression model and the entropy weight method are as follows: inputting the average daily series of each indicator in the historical years into the seasonal regression model, and running to obtain three outputs: trend, seasonal fluctuation and residual; inputting the absolute value series of the sum of seasonal fluctuations and residuals into the entropy weight method model, and running to obtain the information entropy of each indicator; and calculating the "fixed weight" part after normalizing and enhancing the information entropy.

4. The water quality assessment WQI expert weight fusion method based on weighted extreme value fusion algorithm as claimed in claim 1, characterized in that: The steps of calculating the "variable weight" by the pollution contribution rate method are: using the normalization method to proportionally scale all pollution indicators at a certain moment to the interval [0, 1]; and inputting the scaled result as the "variable weight" into the entire system.

5. The water quality assessment WQI expert weight fusion method based on weighted extreme value fusion algorithm as claimed in claim 1, characterized in that: The calculation steps of the "expert weight" are as follows: the importance of water quality indicators is divided into 10 levels: 1 to 3 are secondary indicators, 4 to 6 are general indicators, and 7 to 10 are important indicators; experts who understand the characteristics of local water bodies score the importance of each water quality indicator, and use a visual UI to output the expert scores to the water quality evaluation system; The scores are normalized and input into the entire system as "expert weights".

6. The water quality assessment WQI expert weight fusion method based on weighted extreme value fusion algorithm as claimed in claim 1, characterized in that: The specific implementation steps of the multi-weight fusion algorithm of weighted extreme value fusion are: determine the values ​​of each parameter in the algorithm according to the specific requirements of the research target area; calculate the weight ratio of the three parts of "fixed weight", "variable weight" and "expert weight" at each moment according to the weighted extreme value fusion algorithm formula described in the invention content later; calculate the six-point comprehensive score at each moment by combining the above weight ratios and the single factor evaluation results at each moment.

7. The water quality assessment WQI expert weight fusion method based on weighted extreme value fusion algorithm as claimed in claim 1, characterized in that: pass The six-point rating is converted into a comprehensive evaluation of the 100-point rating, where I i is the six-point evaluation at time i, S i is the percentage evaluation at time i, and ε is the conversion parameter.

Images