A green building design scheme evaluation system based on a multi-objective optimization algorithm

Abstract

The present application belongs to the technical field of green building design evaluation technology and multi-objective optimization algorithm combination technology, and provides a green building design scheme evaluation system based on a multi-objective optimization algorithm, which comprises: screening quantifiable and optimizable green building design targets and identifying strong conflicts between the targets; if there is a strong conflict, a judgment module is used to test the anti-interference ability of the selected optimal solution set obtained by multi-objective optimization to weight fluctuation; if the solution set has poor anti-interference ability, a sensitivity analysis is used to locate the weight sensitive target; then, in combination with subjective weighting, objective weighting and sensitive target characteristics, reliable benchmark weights are generated; finally, the priority of the selected scheme is output by considering the comprehensive performance and stability of the scheme through priority calculation, thereby solving the problems of subjective weight setting and poor scheme anti-interference in the existing green building design evaluation, significantly improving the scientificity and reliability of the evaluation results, and providing precise support for the decision of the green building design scheme.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of green building design evaluation technology and multi-objective optimization algorithm combination technology, and in particular to a green building design scheme evaluation system based on a multi-objective optimization algorithm. BACKGROUND

[0002] Green building design needs to consider environmental (low carbon emission, energy efficiency), economic (initial construction cost, life cycle cost), performance (indoor comfort, air quality) and other multi-dimensional objectives, and multi-objective optimization algorithm becomes the core tool for design scheme screening. However, the following key problems exist in the existing green building design scheme evaluation process:

[0003] Objective conflict identification is fuzzy: the existing evaluation often directly uses multi-objective optimization, but does not accurately identify whether there is a strong conflict between the objectives, if there is no conflict between the objectives (such as operation energy consumption and low carbon emission being positively correlated), single-objective optimization can meet the demand, and multi-objective optimization increases the complexity of calculation, if there is a strong conflict between the objectives (such as low carbon emission and initial construction cost being negatively correlated), it is easy to lead to deviation of the optimization direction without identification.

[0004] Weight setting is subjective and sensitive: the weight setting in multi-objective optimization depends on expert experience or subjective judgment of the owner, and lacks objective basis; at the same time, under the condition of strong conflict between objectives, small changes in weight will cause the sorting of the selected optimal solution set to fluctuate greatly (i.e. weight sensitive), the existing evaluation does not test the anti-interference of the solution set, leading to the optimal scheme changing and the evaluation result being unreliable due to the weight error.

[0005] The scheme selection ignores stability: the existing scheme priority ranking only focuses on the comprehensive performance under the benchmark weight, and does not consider the stability of the scheme when the weight fluctuates, and although some schemes have good performance, they are sensitive to the weight, and in actual application, they are easy to fail due to weight error, and are difficult to guide actual design decision.

[0006] Therefore, the present application provides a green building design scheme evaluation system based on a multi-objective optimization algorithm. SUMMARY

[0007] In order to make up for the deficiencies of the prior art and solve at least one technical problem proposed in the background art.

[0008] The technical scheme adopted by the present application to solve its technical problems is: a green building design scheme evaluation system based on a multi-objective optimization algorithm, comprising the following modules:

[0009] An optimization conflict judgment module: performing correlation analysis on the to-be-optimized objectives of the green building design scheme, and judging whether there is an optimization strong conflict between the to-be-optimized objectives according to the analysis result;

[0010] The anti-interference judgment module: if there is, the weight sensitivity of the candidate optimal solution set obtained by the multi-objective optimization algorithm is analyzed to determine whether the candidate optimal solution set has anti-interference;

[0011] The sensitivity analysis module: if the candidate optimal solution set does not have anti-interference, the stability of each candidate optimal solution in the candidate optimal solution set and the stability of each optimization target are analyzed to obtain the sensitivity coefficient of each optimization target;

[0012] The reference weight calculation module: the optimization target is respectively subjected to subjective weighting and objective weighting, and the subjective-objective fusion coefficient is calculated according to the sensitivity coefficient of the optimization target, and the reference weight of the optimization target is obtained by combining the weighting method.

[0013] The priority calculation module: the weighted comprehensive score of each candidate optimal solution in the candidate optimal solution set under the reference weight is calculated, and the priority of the candidate optimal solution is obtained in combination with the stability of each candidate optimal solution.

[0014] Further, the judgment method of whether there is an optimization strong conflict between the optimization targets is:

[0015] The Pearson correlation coefficient r between any two optimization targets is calculated, and |r| is compared with the preset correlation coefficient, if there is any two optimization targets between which |r| is greater than the preset correlation coefficient, and r<0, then there is an optimization strong conflict between the optimization targets.

[0016] Further, the judgment process of whether the candidate optimal solution set has anti-interference is:

[0017] Each weight in the current weight vector is subjected to a small disturbance of ±1% to ±5% to generate a plurality of perturbed weight vectors;

[0018] The consistency coefficient is calculated by analyzing the consistency of the ranking of each candidate optimal solution under the current weight and each group of perturbed weights;

[0019] The consistency coefficients between the current weight and each group of perturbed weights are integrated into a consistency coefficient set;

[0020] The proportion of the number of consistency coefficients greater than the threshold value in the consistency coefficient set is counted to obtain the ranking stability coefficient;

[0021] If the ranking stability coefficient is less than the preset stability coefficient, the candidate optimal solution set does not have anti-interference.

[0022] Further, the consistency analysis method of the ranking of each candidate optimal solution under the current weight and each group of perturbed weights is:

[0023] From all the candidate optimal solutions, all possible unordered candidate optimal solution pairs are generated, i.e. combinations of two candidate optimal solutions without considering the order;

[0024] For each candidate optimal solution pair, consistency analysis is performed on the ordering of the candidate optimal solution pair under the current weight and each group of perturbation weights, and the candidate optimal solution pair is divided into consistent pairs and inconsistent pairs;

[0025] Based on the number C of consistent pairs and the number D of inconsistent pairs, the consistency coefficient τ is calculated according to the formula , where M is the total number of candidate optimal solution pairs, , and n is the number of candidate optimal solutions.

[0026] Further, the calculation method of the sensitivity coefficient of each optimization target is as follows:

[0027] For each group of perturbation weights, the perturbation weight difference of each optimization target is calculated, and finally the independent variable matrix X is formed, where each row corresponds to a group of perturbation weight differences, and each column corresponds to a perturbation weight difference of an optimization target.

[0028] Where the perturbation weight difference is the difference between each group of perturbation weights in the perturbation weight vector and the current weight;

[0029] For each group of perturbation weights, the weighted comprehensive score fluctuation of all candidate optimal solutions is calculated to form the dependent variable matrix Y.

[0030] According to the independent variable matrix X and the dependent variable matrix Y, a multiple linear regression equation is constructed, and the regression coefficient of the independent variable is obtained by least squares fitting.

[0031] The regression coefficient of the independent variable is standardized and then taken as the absolute value, which is the sensitivity coefficient of each optimization target.

[0032] Further, the process of subjectively weighting each optimization target is as follows:

[0033] The importance score of the optimization target is obtained by the 1-9 scale method;

[0034] The importance score of the optimization target is taken as the criterion layer, and a reciprocal judgment matrix is constructed with the green building comprehensive performance as the target layer.

[0035] Each column of the reciprocal judgment matrix is normalized, and the sum of each row vector W' is obtained by summing the normalized reciprocal judgment matrix by row.

[0036] The sum of each row vector W' is obtained.

[0037] For any optimization target, the subjective weight of the optimization target is W i / row vector sum, where Wi is the row vector of the i-th row in the reciprocal judgment matrix.

[0038] Further, the process of respectively objectively weighting the to-be-optimized targets is:

[0039] Obtain a green building project library of the same climate zone and the same building type in a historical period, and each project library corresponds to a quantitative value of the to-be-optimized target;

[0040] Standardize the quantitative value of the to-be-optimized target to obtain a standardized data matrix;

[0041] Calculate the probability distribution of each to-be-optimized target in different projects through a probability distribution formula;

[0042] According to the probability distribution of each to-be-optimized target in different projects, calculate the information entropy of each to-be-optimized target through an information entropy formula;

[0043] For any to-be-optimized target, calculate the difference coefficient through the formula g j =1-H j , and normalize the difference coefficient to obtain the objective weight of the to-be-optimized target, wherein g j is the difference coefficient of the j-th to-be-optimized target, and H j is the objective weight of the j-th to-be-optimized target.

[0044] Further, the calculation method of the subjective-objective fusion coefficient is:

[0045] The subjective-objective fusion coefficient α i is defined as the proportion of objective weighting in the combined weight of the i-th to-be-optimized target, and the corresponding proportion of subjective weighting is 1-α i .

[0046] Standardize the sensitivity coefficient of each to-be-optimized target to the interval [0, 1] to obtain a relative sensitivity coefficient;

[0047] Map the relative sensitivity coefficient to the subjective-objective fusion coefficient using a linear mapping method, and ensure that α i ∈[α min ,α max ], α i =α min +(α max -α min )×S ’ i , wherein [α min ,α max ] is the value range of objective weighting.

[0048] Further, the calculation method of the reference weight is:

[0049] reference weight of the to-be-optimized target 基 is: ω 基 = α * ω 客 + (1-α) * ω 主 , wherein, alpha is a subjective-objective fusion coefficient, omega 客 is an objective weight, omega 主 is a subjective weight.

[0050] Further, the priority of the to-be-selected optimal solution is calculated in the following manner:

[0051] For any to-be-selected optimal solution in the to-be-selected optimal solution set, a weighted sum of each to-be-optimized target in the to-be-selected optimal solution is calculated based on the reference weight to obtain a weighted comprehensive score;

[0052] The to-be-selected optimal solutions are arranged in descending order of the weighted comprehensive score to obtain a weighted comprehensive score sequence;

[0053] The coefficient of variation of the weighted comprehensive score of each to-be-selected optimal solution under all perturbation weights is calculated, and the to-be-selected optimal solutions are arranged in ascending order of the coefficient of variation to obtain a coefficient of variation sequence;

[0054] The sequence number of the to-be-selected optimal solution in the weighted comprehensive score sequence and the sequence number in the coefficient of variation sequence score are added to obtain a priority index, and the to-be-selected optimal solutions are arranged in ascending order of the priority index to obtain the priority of the to-be-selected optimal solution.

[0055] The beneficial effects of the present application are as follows: the quantifiable and optimizable targets are screened, and the strong conflicts are accurately identified based on the Pearson correlation coefficient, avoiding the problems of redundant calculation of multi-objective optimization for non-conflict targets or optimization deviation caused by unrecognized conflict targets, ensuring that the optimization direction is focused and efficient, the stability of the solution set to weight fluctuation is quantified and solved through weight perturbation and consistency coefficient calculation, avoiding the use of unreliable solution set sensitive to weight for decision-making, reducing the engineering risk caused by weight setting error, positioning the weight sensitive target through multiple linear regression, providing a targeted basis for subsequent reference weight calculation, using a higher proportion of objective weighting for highly sensitive targets to reduce subjective bias and improve the reliability of weight setting, the reference weight calculation module combines expert experience (subjective weighting), historical data (objective weighting) and sensitive target characteristics (fusion coefficient) to avoid the experience deviation of pure subjective weighting or the scene disconnection of pure objective weighting, and generate a reference weight more suitable for actual needs, the priority calculation module considers the comprehensive performance and anti-interference stability of the scheme through the priority index of the weighted comprehensive score + the coefficient of variation, ensuring that the finally selected scheme has both performance and risk resistance, and providing accurate and reliable support for green building design decision-making. BRIEF DESCRIPTION OF DRAWINGS

[0056] The present application will be further described below with reference to the accompanying drawings.

[0057] Figure 1 is a step flow chart of a green building design scheme evaluation system based on a multi-objective optimization algorithm according to the present application;

[0058] Figure 2 is a logical judgment chart of whether the selected optimal solution set has anti-interference. DETAILED DESCRIPTION

[0059] In order to make the technical means, creative features, purposes and effects of the present application easy to understand, the present application is further described below in conjunction with specific embodiments.

[0060] Please refer to Figure 1 The green building design scheme evaluation system based on a multi-objective optimization algorithm according to the present application comprises the following modules:

[0061] An optimization conflict judgment module: performing correlation analysis on the to-be-optimized targets of the green building design scheme, and judging whether there is an optimization strong conflict between the to-be-optimized targets according to the analysis result;

[0062] The process of judging whether there is an optimization strong conflict between the to-be-optimized targets comprises:

[0063] In combination with the green building evaluation standard and the actual requirements of the project, the optimization targets are preliminarily screened, for example:

[0064] Environmental dimension: low carbon emission, energy efficiency, water resource utilization rate, etc.

[0065] Economic dimension: initial construction cost, whole life cycle cost, etc.

[0066] Performance dimension: indoor comfort, indoor air quality, etc.

[0067] The preliminarily screened optimization targets are subjected to quantifiable and optimizable processing, and the optimization targets that are uncontrollable or cannot be adjusted through the design scheme are eliminated, for example:

[0068] | Target category | To-be-optimized target | Quantitative index | Target attribute |

[0069] | Environment | A. Low carbon emission | Whole life cycle carbon emission | The smaller the better |

[0070] | Economy | B. Initial construction cost | Unit area cost | The smaller the better |

[0071] | Performance | Operating energy consumption | Unit area annual power consumption | The smaller the better |

[0072] calculating a Pearson correlation coefficient r between any two to-be-optimized objectives, and comparing |r| with a preset correlation coefficient, if |r| between any two to-be-optimized objectives is greater than the preset correlation coefficient and r < 0, then there is a strong optimization conflict between the to-be-optimized objectives;

[0073] The function of judging whether there is a strong optimization conflict between to-be-optimized objectives is:

[0074] The multi-objective optimization provides effective and necessary optimization objects, and judges whether the multi-objective optimization method needs to be used, if there is no conflict between the objectives, single-objective optimization can be used;

[0075] Anti-interference judgment module: if there is, performing weight sensitivity analysis on the to-be-selected optimal solution set obtained by the multi-objective optimization algorithm to judge whether the to-be-selected optimal solution set has anti-interference;

[0076] Please refer to Figure 2 The process of judging whether the to-be-selected optimal solution set has anti-interference includes:

[0077] The to-be-selected optimal solution set S = {S1, S2,..., Sn} obtained by the multi-objective algorithm, n is the number of solutions, each solution corresponds to a group of to-be-optimized objective performance values; n

[0078] The current weight vector is W0, which contains the weights of each to-be-optimized objective, a small perturbation of ±1% ~ ±5% is performed on each weight in the current weight vector W0 to generate a plurality of perturbed weight vectors;

[0079] For each group of perturbed weights in the perturbed weight vector, the weighted comprehensive score of all to-be-selected optimal solutions is calculated: , wherein p is the number of to-be-optimized objectives, is the weight of the kth to-be-optimized objective in the jth group of perturbed weights, is the normalized performance value of the ith to-be-selected optimal solution on the kth to-be-optimized objective;

[0080] Wherein, the to-be-optimized objectives need to be standardized first, each to-be-optimized objective is standardized to [0, 1]:

[0081] If the to-be-optimized objective is the smaller the better, = (max k - x i ) / (max k - min k ), wherein max k , min k are the maximum value and the minimum value of the to-be-optimized objective x i in the Pareto solution;

[0082] ​If the goal of optimization is to approach T after optimization... =1-|x i -T| / max(|max k -T|,|min k -T|), where, max k min k Let x be the target to be optimized. i The maximum and minimum values ​​in the set of candidate optimal solutions;

[0083] Sort the candidate optimal solutions in descending order of their weighted composite scores to generate a sorted sequence R(ω)=[r1,r2,...,r n ], where r n S is the nth candidate optimal solution n Ranking;

[0084] From all candidate optimal solutions, generate all possible unordered pairs of candidate optimal solutions, i.e., combinations of two candidate optimal solutions regardless of their order. The total number of candidate optimal solution pairs is: ;

[0085] For each pair of candidate optimal solutions (S) i ,S k Let's examine the ranking relationships under the current weight and each perturbation weight:

[0086] Consistent pair: If S under the current weight i Ranked higher than S k And under the perturbation weights S i The ranking is still higher than S k Or both are S i Ranked below S k If the candidate optimal solution pair is a consistent pair, it is denoted as C;

[0087] Inconsistent pairs: If S under the current weight i Ranked higher than S k However, under the perturbation weights, S i Ranked below S k If so, then the candidate optimal solution pair is an inconsistent pair;

[0088] Based on the number of consistent pairs C and inconsistent pairs D, the consistency coefficient is calculated using the following formula: Where M is the total number of pairs of candidate optimal solutions;

[0089] It is understandable that the physical meaning of the consistency coefficient calculation formula is as follows: when all candidate optimal solution pairs are consistent pairs (D=0), τ=1, indicating that the two sets of rankings are completely consistent; when all candidate optimal solution pairs are inconsistent pairs (C=0), τ=-1, indicating that the two sets of rankings are completely opposite; when C=D, τ=0, indicating that the two sets of rankings are unrelated.

[0090] a consistency coefficient of the current weight and each set of perturbation weights is calculated to obtain a consistency coefficient set;

[0091] a proportion of consistency coefficients greater than a threshold value in the consistency coefficient set is counted to obtain a sorting stability coefficient;

[0092] the sorting stability coefficient is compared with a preset stability coefficient, and if the sorting stability coefficient is less than the preset stability coefficient, the set of candidate optimal solutions does not have anti-interference;

[0093] It can be understood that the judgment logic of anti-interference is that if the sorting stability coefficient is greater than the preset stability coefficient, it indicates that each candidate optimal solution in the set of candidate optimal solutions remains stable in most weight fluctuations, and the overall sensitivity is low, otherwise, it indicates that the sorting is easily affected by the weight;

[0094] the role of judging whether the set of candidate optimal solutions has anti-interference is:

[0095] under the premise that the target exists strong conflict, the anti-interference ability of the set of candidate optimal solutions obtained by multi-objective optimization to weight setting error is tested, that is, whether the sorting of the solution is stable when the weight changes slightly, to avoid the case that the optimal solution becomes unreliable when the weight changes slightly;

[0096] the sensitivity analysis module: if the set of candidate optimal solutions does not have anti-interference, the stability of each candidate optimal solution in the set of candidate optimal solutions and the stability of each optimization target are analyzed to obtain the sensitivity coefficient of each optimization target;

[0097] the calculation process of the sensitivity coefficient of each optimization target includes:

[0098] for each set of perturbation weights in the perturbation weight vector, the weighted comprehensive performance score of all candidate optimal solutions is calculated;

[0099] the difference between each set of perturbation weights in the perturbation weight vector and the current weight is the perturbation weight difference;

[0100] for each set of perturbation weights, the perturbation weight difference of the kth optimization target is calculated , wherein, ω 0k is the current weight of the kth optimization target, and finally the independent variable matrix X is formed, each row corresponds to a set of perturbation weight differences, and each column corresponds to a perturbation weight difference of an optimization target;

[0101] for each set of perturbation weights, the weighted comprehensive score fluctuation of all candidate optimal solutions is calculated : , wherein, Score(S i , ω0) is the candidate optimal solution S iThe weighted comprehensive score under each group of disturbance weights forms the dependent variable vector Y finally;

[0102] According to the independent variable matrix X and the dependent variable matrix Y, a multiple linear regression equation Y=β0+β1X1+β2X2+...+β p X p +ε is constructed, where β0 is a constant term, β1~β p is a partial regression coefficient, reflecting the influence direction and absolute intensity of the weight fluctuation of the kth optimization target on the result, and ε is a random error term;

[0103] The partial regression coefficients β1~β p are obtained by fitting the model by the least squares method.

[0104] The partial regression coefficients are standardized to obtain the standardized regression coefficients: , where σ(X k ) is the standard deviation of the kth independent variable, and σ(Y) is the standard deviation of the dependent variable;

[0105] The absolute value of the standardized regression coefficient of each optimization target is the sensitivity coefficient;

[0106] The role of calculating the sensitivity coefficient of each optimization target is:

[0107] When the anti-interference of the selected optimal solution set is poor, find out which target's weight change has the greatest impact on the score of the solution, i.e. locate the sensitive target, and provide a targeted basis for subsequent reasonable weighting (the weight of the high sensitivity target needs to be set more carefully);

[0108] The reference weight calculation module: subjectively and objectively weights the optimization targets, and calculates the subjective-objective fusion coefficient according to the sensitivity coefficient of the optimization target, and obtains the reference weight of the optimization target through the combined weighting method;

[0109] The process of subjectively weighting the optimization targets includes:

[0110] Select a number of cross-disciplinary experts to score the optimization targets according to the 1-9 scale method (1=equally important, 9=extremely important);

[0111] Taking the green building comprehensive performance as the target layer and the optimization target as the criterion layer, a reciprocal judgment matrix A is constructed:

[0112] | Criterion layer | G1 | G2 |... | G n |;

[0113] | G1 | a 11 | a 12 |... | a 1n |;

[0114] |G2|a 21 |a 22 |...|a 2n |;

[0115] |...|...|...|...|...|;

[0116] |G n |a n1 |a n2 |...|a nn |;

[0117] The matrix elements satisfy: diagonal elements a ii =1 (the importance of the target itself and itself is equal), the non-diagonal elements satisfy a ij =1 / a ji (if the importance scale of G i is 3 than G j , the scale of G j is 1 / 3 than G i );

[0118] For the judgment matrix A, the product sum of the matrix A and the normalized column vector is calculated, and the maximum eigenvalue λ max of the matrix is obtained by averaging, that is:

[0119] Each column of the judgment matrix A is normalized, the sum of the normalized matrix is calculated by row, and the vector W' is obtained, and the maximum eigenvalue is: , wherein, A×W' is the product of the matrix A and the vector W', and (A×W')i is the ith element;

[0120] The consistency deviation of the matrix is obtained by the formula ;

[0121] The consistency ratio CR is calculated by combining the average random consistency index RI to eliminate the influence of the order of the matrix, and the formula is: CR=CI / RI, wherein the average random consistency index RI is obtained by looking up the table;

[0122] If CR<0.1, the matrix passes the consistency test (logical contradiction is eliminated);

[0123] The subjective weight is obtained by the eigenvalue method: the sum of the vector W' is calculated, and the subjective weight is ω i组 =W i ' / sum, wherein i is the ith target to be optimized;

[0124] The process of respectively objectively weighting the targets to be optimized includes:

[0125] Acquire the green building project library of the same climate zone and the same building type in the historical period, and each project library corresponds to the quantitative value of the to-be-optimized target;

[0126] Standardize the quantitative value of the to-be-optimized target to obtain a standardized data matrix X';

[0127] Calculate the probability distribution p of the jth target ij : , wherein m is the number of projects;

[0128] Calculate the information entropy H of the jth target j : , wherein 1 / lnm is a normalization factor to ensure that H j ∈[0,1];

[0129] Calculate the difference coefficient g j =1-H j , and normalize the difference coefficient to obtain the objective weight: ;

[0130] The calculation process of the subjective-objective fusion coefficient includes:

[0131] The subjective-objective fusion coefficient a i is defined as the proportion of objective weighting in the combined weight of the ith to-be-optimized target, and the proportion of subjective weighting corresponding to the objective weighting is 1-a i , and the calculation process needs to follow the principle of high sensitivity→high objective proportion:

[0132] Standardize the sensitivity coefficient of each to-be-optimized target to a relative sensitivity coefficient in the interval [0,1];

[0133] Map the relative sensitivity coefficient to the subjective-objective fusion coefficient using the linear mapping method to ensure that a i ∈[a min ,a max ]: a i =a min +(a max -a min )×S ’ i , wherein [a min ,a max ] is the value range of objective weighting, which is determined according to the project requirements;

[0134] It should be noted that the calculation logic of the subjective-objective fusion coefficient is: the higher the sensitivity coefficient of the to-be-optimized target (the greater the influence of the weight fluctuation on the comprehensive score of the solution), the higher the proportion of objective weighting (a ) in the combined weight of the target, and the proportion of subjective weighting (1-a The lower the subjective bias is, the lower the interference of the subjective bias on the sensitive target weight is, and the reliability of the benchmark weight is ensured.

[0135] The benchmark weight of the target to be optimized is ω 基 =α×ω 客 +(1-α)×ω 主 .

[0136] The role of calculating the benchmark weight is:

[0137] The unreasonable problem of weight setting is solved, the pure subjective weighting (expert experience bias) or pure objective weighting (ignoring the actual demand of the project) is avoided, the reliable benchmark weight is generated in combination with the sensitivity coefficient of the target to be optimized, and a fair and reasonable evaluation standard is provided for subsequent priority calculation;

[0138] The priority calculation module: calculating the weighted comprehensive score of each candidate optimal solution in the set of candidate optimal solutions under the benchmark weight, and obtaining the priority of the candidate optimal solution in combination with the stability of each candidate optimal solution;

[0139] The priority calculation process of the scheme includes:

[0140] For any one of the candidate optimal solution in the set of candidate optimal solutions, the weighted sum of each target to be optimized in the candidate optimal solution is calculated based on the benchmark weight, and the weighted comprehensive score is obtained;

[0141] The candidate optimal solution is arranged in descending order according to the weighted comprehensive score, and the weighted comprehensive score sequence is obtained;

[0142] The variation coefficient of the weighted comprehensive score of each candidate optimal solution under all perturbation weights is calculated, the candidate optimal solution is arranged in ascending order according to the variation coefficient, and the variation coefficient sequence is obtained;

[0143] The sequence number of the candidate optimal solution in the weighted comprehensive score sequence and the sequence number in the variation coefficient sequence score are added to obtain the priority index, and the candidate optimal solution is arranged in ascending order according to the priority index, and the priority of the candidate optimal solution is obtained.

[0144] It can be understood that the physical meaning of the priority index is: the smaller the sum of the sequence numbers, the better the overall performance of the solution on the two core requirements of the comprehensive performance under the benchmark weight and the anti-interference stability when the weight fluctuates, and the higher the priority in the candidate set.

[0145] The role of determining the priority of the candidate optimal solution is:

[0146] After the benchmark weight is determined, the two dimensions of comprehensive performance and anti-interference stability are considered, all candidate optimal solutions are sorted, and the final optimal scheme most suitable for actual application is selected.

[0147] The technical scheme and advantages of the embodiment of the application are as follows: the correlation of the to-be-optimized targets of the green building design scheme is analyzed, whether there is an optimization strong conflict between the to-be-optimized targets is judged according to the analysis result, if there is, weight sensitivity analysis is performed on the to-be-selected optimal solution set obtained by the multi-objective optimization algorithm, whether the to-be-selected optimal solution set has anti-interference is judged, if the to-be-selected optimal solution set does not have anti-interference, correlation analysis is performed on the stability of each to-be-selected optimal solution in the to-be-selected optimal solution set and the stability of each to-be-optimized target, the sensitivity coefficient of each to-be-optimized target is obtained, the to-be-optimized targets are respectively subjected to subjective weighting and objective weighting, and the subjective-objective fusion coefficient is calculated according to the sensitivity coefficient of the to-be-optimized target, the reference weight of the to-be-optimized target is obtained through the combined weighting method, the weighted comprehensive score of each to-be-selected optimal solution in the to-be-selected optimal solution set under the reference weight is calculated, and the priority of the to-be-selected optimal solution is obtained in combination with the stability of each to-be-selected optimal solution. The application can screen quantifiable and optimal green building design targets and identify strong conflicts between the targets. If there is a strong conflict, the to-be-selected optimal solution set obtained by the multi-objective optimization is tested for anti-interference ability to weight fluctuation through the judgment module. If the solution set has poor anti-interference, the weight sensitive target is located through sensitivity analysis. In combination with subjective weighting, objective weighting and sensitive target characteristics, reliable reference weight is generated. Finally, the priority of the to-be-selected scheme is output by taking into account the comprehensive performance and stability of the scheme through priority calculation, the problems of subjective weight setting and poor anti-interference of the scheme in the existing green building design evaluation are solved, the scientificity and reliability of the evaluation result are significantly improved, and accurate support is provided for the decision of the green building design scheme.

[0148] The basic principles, main features and advantages of the application are shown and described above. It should be understood by those skilled in the art that the application is not limited by the above embodiments, and the above embodiments and descriptions in the specification are only to illustrate the principles of the application. Without departing from the spirit and scope of the application, various changes and improvements can be made to the application, and these changes and improvements all fall within the scope of the claimed application. The scope of protection of the application is defined by the appended claims and their equivalents.

Claims

1. A green building design scheme evaluation system based on a multi-objective optimization algorithm, characterized in that: The system comprises the following modules: An optimization conflict judgment module: performing correlation analysis on the to-be-optimized targets of the green building design scheme, and judging whether there is an optimization strong conflict between the to-be-optimized targets according to the analysis result; An anti-interference judgment module: if there is, performing weight sensitivity analysis on the to-be-selected optimal solution set obtained by the multi-objective optimization algorithm, and judging whether the to-be-selected optimal solution set has anti-interference; The judgment process of whether the to-be-selected optimal solution set has anti-interference is as follows: Performing a slight perturbation of ±1% to ±5% on each weight in the current weight vector to generate a plurality of perturbed weight vectors; Performing consistency analysis on the ranking of each to-be-selected optimal solution under the current weight and each group of perturbed weights to calculate a consistency coefficient; Integrating the consistency coefficients between the current weight and each group of perturbed weights into a consistency coefficient set; Statistically counting the proportion of consistency coefficients greater than a threshold value in the consistency coefficient set to obtain a ranking stability coefficient; If the ranking stability coefficient is less than a preset stability coefficient, the to-be-selected optimal solution set does not have anti-interference; A sensitivity analysis module: if the to-be-selected optimal solution set does not have anti-interference, performing correlation analysis on the stability of each to-be-selected optimal solution in the to-be-selected optimal solution set and the stability of each to-be-optimized target to obtain a sensitivity coefficient of each to-be-optimized target; The calculation method of the sensitivity coefficient of each to-be-optimized target is as follows: For each group of perturbed weights, calculate the perturbed weight difference of each to-be-optimized target to finally form an independent variable matrix X, each row corresponding to a group of perturbed weight differences and each column corresponding to a perturbed weight difference of a to-be-optimized target; Wherein, the perturbed weight difference is the difference between each group of perturbed weight vectors and the current weight; For each group of perturbed weights, calculate the weighted comprehensive score fluctuation of all to-be-selected optimal solutions to form a dependent variable matrix Y; According to the independent variable matrix X and the dependent variable matrix Y, a multiple linear regression equation is constructed, the model is fitted by the least square method, and the regression coefficient of the independent variable is obtained; The regression coefficient of the independent variable is standardized and then takes the absolute value, which is the sensitivity coefficient of each to-be-optimized target; A baseline weight calculation module: performing subjective weighting and objective weighting on the to-be-optimized targets respectively, and calculating a subjective-objective fusion coefficient according to the sensitivity coefficient of the to-be-optimized target, and obtaining the baseline weight of the to-be-optimized target by combination weighting method; The calculation method of the subjective-objective fusion coefficient is as follows: Subjective-objective fusion coefficient α i The objective weight occupies a proportion in the combination weight of the ith to-be-optimized target, and the corresponding subjective weight occupies a proportion of 1-α i ; The sensitivity coefficients of each target to be optimized are normalized to the interval [0, 1] to obtain the relative sensitivity coefficients S ’ i ; The relative sensitivity coefficient is mapped into the subjective-objective fusion coefficient using a linear mapping method, ensuring that i ∈ [α min ,α max ], α i =α min +(α max -α min )×S ’ i , where [α min ,α max ] is the value range of the objective weighting; The calculation method of the baseline weight is as follows: The reference weight ω of the target to be optimized 基 is: ω 基 = α i × ω 客 + (1- α i )× ω 主 , wherein α i is the subjective-objective fusion coefficient of the i-th target to be optimized, ω 客 is the objective weight, and ω 主 is the subjective weight A priority calculation module: calculating the weighted comprehensive score of each to-be-selected optimal solution in the to-be-selected optimal solution set under the baseline weight, and combining the stability of each to-be-selected optimal solution to obtain the priority of the to-be-selected optimal solution.

2. The green building design scheme evaluation system based on a multi-objective optimization algorithm according to claim 1, characterized in that: The judgment method of whether there is an optimization strong conflict between the to-be-optimized targets is as follows: Calculate the Pearson correlation coefficient r between any two to-be-optimized targets, and compare |r| with a preset correlation coefficient, if there is |r| greater than the preset correlation coefficient between any two to-be-optimized targets, and r<0, then there is an optimization strong conflict between the to-be-optimized targets.

3. The green building design scheme evaluation system based on a multi-objective optimization algorithm of claim 1, characterized in that: the consistency analysis of the order of each candidate optimal solution under the current weight and each group of disturbance weights is performed in the following manner: all possible unordered candidate optimal solution pairs are generated from all candidate optimal solutions, i.e. combinations of two candidate optimal solutions without considering the order; for each candidate optimal solution pair, the consistency analysis of the order of the candidate optimal solution pair under the current weight and each group of disturbance weights is performed, and the candidate optimal solution pair is divided into consistent pairs and inconsistent pairs; Based on the number of consistent pairs and inconsistent pairs C, D, the consistency coefficient τ is calculated according to the formula , wherein M is the total number of pairs of optimal solutions to be selected, , and n is the number of optimal solutions to be selected.

4. The green building design scheme evaluation system based on a multi-objective optimization algorithm of claim 1, characterized in that: the process of subjectively weighting each optimization target is as follows: the importance score of each optimization target is obtained by using the 1-9 scale method; a reciprocal judgment matrix is constructed with the green building comprehensive performance as the target layer and the importance score of each optimization target as the criterion layer; each column of the reciprocal judgment matrix is normalized, and the sum of each row vector W' of the normalized reciprocal judgment matrix is obtained; the sum of each row vector W' is obtained; For any one of the to-be-optimized targets, the subjective weight of the to-be-optimized target is W i ’ / row vector sum, wherein W i ’ is the row vector of the i-th row in the reciprocal judgment matrix.

5. The green building design scheme evaluation system based on a multi-objective optimization algorithm of claim 4, characterized in that: the process of objectively weighting each optimization target is as follows: a green building project library of the same climate zone and the same building type in the historical period is obtained, and each project library corresponds to the quantified value of each optimization target; the quantified value of each optimization target is standardized to obtain a standardized data matrix; the probability distribution of each optimization target in different projects is calculated by using the probability distribution formula; the information entropy of each optimization target is calculated by using the information entropy formula according to the probability distribution of each optimization target in different projects; For any one to be optimized target, the difference coefficient is calculated by formula g j =1-H j , and the objective weight of the to-be-optimized target is obtained by normalizing the difference coefficient, wherein g j is the difference coefficient of the jth to-be-optimized target, and H j is the information entropy of the jth to-be-optimized target.

6. The green building design scheme evaluation system based on a multi-objective optimization algorithm of claim 1, characterized in that: the calculation method of the priority of the candidate optimal solution is as follows: for any candidate optimal solution in the candidate optimal solution set, the weighted sum of each optimization target in the candidate optimal solution is calculated based on the benchmark weight to obtain a weighted comprehensive score; the candidate optimal solutions are arranged in descending order of the weighted comprehensive score to obtain a weighted comprehensive score sequence; the coefficient of variation of the weighted comprehensive score of each candidate optimal solution under all disturbance weights is calculated, and the candidate optimal solutions are arranged in ascending order of the coefficient of variation to obtain a coefficient of variation sequence; the priority index is obtained by adding the sequence number of the candidate optimal solution in the weighted comprehensive score sequence to the sequence number in the coefficient of variation sequence score, and the candidate optimal solutions are arranged in ascending order of the priority index to obtain the priority of the candidate optimal solution.

Images