CN116957422B - Ecological environment evaluation method based on convolution self-coding and sharp point mutation model - Google Patents
Ecological environment evaluation method based on convolution self-coding and sharp point mutation model Download PDFInfo
- Publication number
- CN116957422B CN116957422B CN202311216232.6A CN202311216232A CN116957422B CN 116957422 B CN116957422 B CN 116957422B CN 202311216232 A CN202311216232 A CN 202311216232A CN 116957422 B CN116957422 B CN 116957422B
- Authority
- CN
- China
- Prior art keywords
- representing
- evaluation
- ecological environment
- value
- index
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000011156 evaluation Methods 0.000 title claims abstract description 130
- 230000035772 mutation Effects 0.000 title claims abstract description 46
- 238000004458 analytical method Methods 0.000 claims abstract description 20
- 230000009467 reduction Effects 0.000 claims abstract description 18
- 238000013210 evaluation model Methods 0.000 claims abstract description 12
- 238000012545 processing Methods 0.000 claims abstract description 8
- 230000000875 corresponding effect Effects 0.000 claims description 42
- 238000000034 method Methods 0.000 claims description 38
- 230000006835 compression Effects 0.000 claims description 12
- 238000007906 compression Methods 0.000 claims description 12
- 230000002596 correlated effect Effects 0.000 claims description 12
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 claims description 8
- 230000010355 oscillation Effects 0.000 claims description 8
- 238000006243 chemical reaction Methods 0.000 claims description 7
- 239000011159 matrix material Substances 0.000 claims description 7
- 238000005457 optimization Methods 0.000 claims description 7
- 238000011176 pooling Methods 0.000 claims description 6
- 230000008569 process Effects 0.000 claims description 6
- 238000012795 verification Methods 0.000 claims description 6
- 238000011160 research Methods 0.000 claims description 4
- 238000009795 derivation Methods 0.000 claims description 3
- 230000010354 integration Effects 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims description 3
- 238000007619 statistical method Methods 0.000 claims description 3
- 238000013519 translation Methods 0.000 claims description 3
- 238000013507 mapping Methods 0.000 abstract description 2
- 230000006870 function Effects 0.000 description 31
- 238000013528 artificial neural network Methods 0.000 description 4
- 238000012549 training Methods 0.000 description 3
- 230000002159 abnormal effect Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000011002 quantification Methods 0.000 description 2
- 230000007704 transition Effects 0.000 description 2
- 238000013459 approach Methods 0.000 description 1
- 230000001174 ascending effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000012854 evaluation process Methods 0.000 description 1
- 238000000513 principal component analysis Methods 0.000 description 1
- 238000012847 principal component analysis method Methods 0.000 description 1
- 230000000630 rising effect Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/26—Government or public services
Abstract
The invention discloses an ecological environment evaluation method based on a convolution self-coding and sharp point mutation model, which relates to the technical field of ecological environment and comprises the following steps: establishing an evaluation index system; normalizing each evaluation index; analyzing the processed evaluation index set through a pearson correlation coefficient, and reserving the evaluation index with the correlation lower than a threshold value to obtain a truncated evaluation index set; inputting the pruned evaluation index set to a convolution self-encoder to obtain a dimension reduction analysis result; based on the dimension reduction analysis result, constructing an ecological environment evaluation model by utilizing a point mutation theory; and inputting the dimension reduction analysis result of the index to be evaluated into an ecological environment evaluation model to finish the evaluation. The invention fuses the convolution self-encoder and the mutation theory, enhances the mapping capability when processing the high-dimensional nonlinear characteristic data, improves the objectivity of acquiring the index weight, solves the uncertainty of the prior art when making a target decision, and improves the accuracy of the ecological environment evaluation result.
Description
Technical Field
The invention relates to the technical field of ecological environment, in particular to an ecological environment evaluation method based on convolution self-coding and a sharp point mutation model.
Background
Ecological environmental assessment has become an important scientific approach to leading-edge scientific questions and interpretation of human relationships in the current sustainable development field. The current situation of the ecological environment and the dynamic change process thereof are objectively analyzed and quantitatively evaluated, and the method has important significance for protecting the regional ecological environment, planning land utilization and formulating sustainable development strategies. At present, scholars at home and abroad develop a great deal of ecological environment evaluation work and develop a plurality of quantification methods. The methods commonly used at present are a hierarchical analysis method, an entropy value method, a principal component analysis method, a gray correlation method, a fuzzy decision analysis method, an artificial neural network and the like. Although these evaluation methods are successfully applied to evaluation research of ecological environment, they have poor mapping ability when processing high-dimensional nonlinear feature data and are difficult to objectively and effectively acquire index weights. For example, the relative importance of the unmeasured elements can be measured through comparison by the analytic hierarchy process, but the judgment of the importance of the elements needs human intervention, so that the evaluation result has larger subjectivity. The gray correlation method has low requirements on the parameters of the ecosystem, is suitable for the fragile ecosystem which is not unified, but the determination of the resolution coefficient has larger subjectivity. The principal component analysis is a linear conversion method and requires data to conform to normal distribution, but in the evaluation of a ecological environment, many index data do not conform to normal distribution, which may cause the evaluation result to be inconsistent with the actual situation. The artificial neural network needs a large amount of sample data in the evaluation process, is sensitive to abnormal values and noise, and can influence the accuracy of an evaluation result if a large amount of abnormal values and noise exist in the data.
Disclosure of Invention
Aiming at the defects in the prior art, the ecological environment evaluation method based on the convolution self-coding and the sharp point mutation model solves the problem of low accuracy of the evaluation result of the ecological environment in the prior art.
In order to achieve the aim of the invention, the invention adopts the following technical scheme:
the ecological environment evaluation method based on the convolution self-coding and sharp point mutation model comprises the following steps:
s1, establishing an evaluation index system; normalizing each evaluation index to obtain a processed evaluation index set;
s2, analyzing the processed evaluation index set through a Pearson correlation coefficient, and reserving the evaluation index with the correlation lower than a threshold value to obtain a pruned evaluation index set;
s3, inputting the pruned evaluation index set into a convolution self-encoder to obtain a dimension reduction analysis result of the artificial influence factors and the natural influence factors; based on the dimension reduction analysis result of the convolution self-encoder, constructing an ecological environment evaluation model by utilizing a sharp point mutation theory;
s4, inputting the dimension reduction analysis result of the index to be evaluated into an ecological environment evaluation model to obtain an ecological environment evaluation result, and completing evaluation.
Further, the formula of the normalization processing in step S1 is as follows:
wherein, the two formulas are respectively used for positive correlation evaluation indexes and negative correlation evaluation indexes,representing a normalized value indicative of the evaluation index, < + >>A true value indicative of the positive correlation evaluation index, < >>Represents the minimum value in the same positive correlation evaluation index,/->Represents the maximum value in the same positive correlation evaluation index, < >>A true value representing the negative correlation evaluation index,represents the minimum value in the same negative correlation evaluation index, < ->Represents the maximum value of the same negative correlation evaluation index.
Further, the formula of the correlation coefficient of step S2 is as follows:
wherein,representing the correlation coefficient>Representing a summation function>Representing sample data->Is used for the control of the variable value of (a),representing sample data->Average value of>Representing sample data->Variable value of>Representing sample data->Average value of (2);
when (when)When the two corresponding evaluation indexes are extremely weakly correlated or uncorrelated;
when (when)When the two corresponding evaluation indexes are indicated to be weakly correlated;
when (when)When the two corresponding evaluation indexes are related in a medium degree;
when (when)When the two corresponding evaluation indexes are shown to be strongly correlated;
when (when)When the two corresponding evaluation indexes are extremely strongly correlated.
Further, the threshold of step S2 is set to 0.6.
Further, the evaluation index set in step S2 includes a natural influence factor index and an artificial influence factor index; the convolutional self-encoder in step S3 comprises an encoder and a decoder, the encoder comprising a convolutional layer and a pooling layer; the decoder includes a deconvolution layer.
Further, the specific steps of step S3 are as follows:
s3-1, respectively inputting natural influence factor indexes and artificial influence factor indexes into a convolution self-encoder, and performing dimension reduction processing through the encoder to obtain corresponding initial compression characteristics; reducing the bit number occupied by each pixel by the two initial compression characteristics through a quantizer to obtain corresponding final compression characteristics; inputting the two final compression characteristics to a decoder to obtain corresponding reconstruction data, namely a natural influence comprehensive index and an artificial influence comprehensive index;
s3-2, performing function fitting and statistical analysis on the natural influence comprehensive index and the artificial influence comprehensive index to respectively obtain critical values of oscillation points of the natural influence comprehensive index and the artificial influence comprehensive index, wherein the occurrence of the critical values is larger than a threshold valueAnd->And a critical value for the amplitude of oscillation which tends to be gentle or discontinuous +.>And->；
S3-3, establishing a balanced curved surface equation according to the point mutation theory, wherein the equation is as follows:
wherein,、/>and->Respectively representing the original three-dimensional coordinate system of the balance curved surface,/>、/>representing the coefficients;
s3-4, according to a coordinate conversion formula:
original three-dimensional coordinate system of balanced curved surfaceConversion to a new three-dimensional coordinate System->The method comprises the steps of carrying out a first treatment on the surface of the Wherein (1)>Representing the amount of spatial translation, +.>Representing the angle of rotation of the original three-dimensional coordinate around the Z axis;
s3-5, according to the formula:
performing function fitting analysis on the natural influence comprehensive index and the artificial influence comprehensive index to obtain a corresponding fitting curve; conducting derivation and integration on the two fitting curves to obtain corresponding mutation positions; according to the mutation position, MATLAB and the approximation idea are adopted to obtain、/>、/>、/>、/>、/>Is a value of (2); wherein (1)>Representing a maximum function;
s3-6, according to the formula:
constructing an ecological environment evaluation model, and calculating a corresponding ecological environment evaluation value based on the parameter values obtained in the step S3-5The method comprises the steps of carrying out a first treatment on the surface of the Wherein the ecological environment evaluation value->The judgment standard of (2) is as follows:
as an ecological environment evaluation valueWhen the ecological environment condition is optimal;
as an ecological environment evaluation valueAnd when the method is used, a natural breakpoint method is adopted for grading research, and the grades are respectively good, medium, poor and poor.
Further, the convolutional self-encoder adopts an Adam algorithm to perform parameter optimization; optimizing weight matrixes and bias parameters of all convolution layers of the convolution self-encoder by adopting a mean square error loss function; the formula of the mean square error loss function is as follows:
wherein,mean square error +.>Representing predicted values +.>Indicating target value,/->The amount of data is represented and,representing the summation function.
Further, the Adam algorithm corresponds to the formula:
wherein,representing the current time step->Representing the objective function to be optimized, < >>Gradient representing objective function ∈ ->And->Respectively representing model parameters and step size, +.>And->Representing a first moment estimate of the gradient and a second moment estimate of the gradient, respectively, < >>And->First moment estimates representing gradients, respectively +.>And second moment estimation of gradient +.>Exponential decay rate, +.>Representing a constant->And->Representing a constant.
Further, the formulas corresponding to the encoder and decoder of step S3-1 are as follows:
wherein,indicate->Hidden features extracted by convolution kernel, +.>Represents the pooling function of the encoder, +.>Indicating->Weight matrix of convolution kernels, +.>Representing natural influence factor index and artificial influence factor index, < ->Indicating->Bias of the convolution kernels ∈>Representing reconstructed data,/->Representing all hidden feature sets, ++>Representing a summation function>Representing the%>Weight matrix of convolution kernels, +.>Representing the%>Bias of the convolution kernels ∈>Representing the upsampling function of the decoder.
Further, the mathematical approximation idea of step S3-5 is specifically as follows:
setting parametersA constant value is given to any constant value, and the parameter +.>、/>、/>、/>、/>Is a value of (2); according to the parameters->、/>、/>、/>、/>And->Obtaining a balanced curved surface equation and a bifurcation curve equation; selecting a known ecological environment point far away from the bifurcation curve for verification; if 90% of the verification points are distributed around the bifurcation curve, the parameter +.>And if the given value meets the solving requirement, continuing to assign the solution.
The beneficial effects of the invention are as follows: the method effectively fuses the convolution self-encoder and the mutation theory, ensures the high-dimensional nonlinear characteristics of the ecological environment evaluation index system, overcomes the artificial subjectivity of the determination of the weight of the ecological environment evaluation index, improves the objectivity, and solves the problem of lower accuracy of the ecological environment evaluation result caused by the uncertainty of the traditional mutation theory application mode and the dependence problem of independent parameter adjustment on each index layer.
Drawings
FIG. 1 is a flow chart of the present invention;
FIG. 2 is a convolutional self-encoder training flow chart of the present invention;
FIG. 3 is a coordinate transformation diagram of the theory of sharp point mutation;
FIG. 4 is a graph of a function fit analysis of the natural influencing factor dimension reduction results;
FIG. 5 is a graph of a function fit analysis of the dimension reduction results of artificial influencing factors.
Detailed Description
The following description of the embodiments of the present invention is provided to facilitate understanding of the present invention by those skilled in the art, but it should be understood that the present invention is not limited to the scope of the embodiments, and all the inventions which make use of the inventive concept are protected by the spirit and scope of the present invention as defined and defined in the appended claims to those skilled in the art.
As shown in fig. 1, an ecological environment evaluation method based on convolution self-coding and sharp point mutation models comprises the following steps:
s1, establishing an evaluation index system; normalizing each evaluation index to obtain a processed evaluation index set;
s2, analyzing the processed evaluation index set through a Pearson correlation coefficient, and reserving the evaluation index with the correlation lower than a threshold value to obtain a pruned evaluation index set;
s3, inputting the pruned evaluation index set into a convolution self-encoder to obtain a dimension reduction analysis result of the artificial influence factors and the natural influence factors; based on the dimension reduction analysis result of the convolution self-encoder, constructing an ecological environment evaluation model by utilizing a sharp point mutation theory;
s4, inputting the dimension reduction analysis result of the index to be evaluated into an ecological environment evaluation model to obtain an ecological environment evaluation result, and completing evaluation.
The formula of the normalization processing in step S1 is as follows:
wherein, the two formulas are respectively used for positive correlation evaluation indexes and negative correlation evaluation indexes,representing a normalized value indicative of the evaluation index, < + >>A true value indicative of the positive correlation evaluation index, < >>Represents the minimum value in the same positive correlation evaluation index,/->Represents the maximum value in the same positive correlation evaluation index, < >>A true value representing the negative correlation evaluation index,represents the minimum value in the same negative correlation evaluation index, < ->Represents the maximum value of the same negative correlation evaluation index.
The formula of the correlation coefficient of step S2 is as follows:
wherein,representing the correlation coefficient>Representing a summation function>Representing sample data->Is used for the control of the variable value of (a),representing sample data->Average value of>Representing sample data->Variable value of>Representing sample data->Average value of (2);
when (when)When the two corresponding evaluation indexes are extremely weakly correlated or uncorrelated;
when (when)When the two corresponding evaluation indexes are indicated to be weakly correlated;
when (when)When the two corresponding evaluation indexes are related in a medium degree;
when (when)When the two corresponding evaluation indexes are shown to be strongly correlated;
when (when)When the two corresponding evaluation indexes are extremely strongly correlated.
The threshold of step S2 is set to 0.6.
The evaluation index set in the step S2 comprises natural influence factor indexes and artificial influence factor indexes; the convolutional self-encoder in step S3 comprises an encoder and a decoder; the encoder comprises a convolution layer and a pooling layer; the decoder includes a deconvolution layer.
The specific steps of the step S3 are as follows:
s3-1, respectively inputting natural influence factor indexes and artificial influence factor indexes into a convolution self-encoder, and performing dimension reduction processing through the encoder to obtain corresponding initial compression characteristics; reducing the bit number occupied by each pixel by the two initial compression characteristics through a quantizer to obtain corresponding final compression characteristics; inputting the two final compression characteristics to a decoder to obtain corresponding reconstruction data, namely a natural influence comprehensive index and an artificial influence comprehensive index;
s3-2, performing function fitting and statistical analysis on the natural influence comprehensive index and the artificial influence comprehensive index to respectively obtain critical values of oscillation points of the natural influence comprehensive index and the artificial influence comprehensive index, wherein the occurrence of the critical values is larger than a threshold valueAnd->And a critical value for the amplitude of oscillation which tends to be gentle or discontinuous +.>And->；
S3-3, establishing a balanced curved surface equation according to the point mutation theory, wherein the equation is as follows:
wherein,、/>and->Original three-dimensional coordinate system respectively representing balanced curved surfaces, +.>、/>Representing the coefficients;
s3-4, according to a coordinate conversion formula:
original three-dimensional coordinate system of balanced curved surfaceConversion to a new three-dimensional coordinate System->The method comprises the steps of carrying out a first treatment on the surface of the Wherein (1)>Representing the amount of spatial translation, +.>Representing the angle of rotation of the original three-dimensional coordinate around the Z axis;
s3-5, according to the formula:
performing function fitting analysis on the natural influence comprehensive index and the artificial influence comprehensive index to obtain a corresponding fitting curve; conducting derivation and integration on the two fitting curves to obtain corresponding mutation positions; according to the mutation position, MATLAB and the approximation idea are adopted to obtain、/>、/>、/>、/>、/>Is a value of (2); wherein (1)>Representing a maximum function;
s3-6, according to the formula:
constructing an ecological environment evaluation model, and calculating a corresponding ecological environment evaluation value based on the parameter values obtained in the step S3-5The method comprises the steps of carrying out a first treatment on the surface of the Wherein the ecological environment evaluation value->The judgment standard of (2) is as follows:
as an ecological environment evaluation valueWhen the ecological environment condition is optimal;
as an ecological environment evaluation valueAnd when the method is used, a natural breakpoint method is adopted for grading research, and the grades are respectively good, medium, poor and poor.
The convolution self-encoder adopts an Adam algorithm to perform parameter optimization; optimizing weight matrixes and bias parameters of all convolution layers of the convolution self-encoder by adopting a mean square error loss function; the formula of the mean square error loss function is as follows:
wherein,mean square error +.>Representing predicted values +.>Indicating target value,/->The amount of data is represented and,representing the summation function.
The Adam algorithm corresponds to the formula:
wherein,representing the current time step->Representing the objective function to be optimized, < >>Gradient representing objective function ∈ ->And->Respectively representing model parameters and step size, +.>And->Respectively represent one of gradientsMoment estimate and second moment estimate of gradient, +.>And->First moment estimates representing gradients, respectively +.>And second moment estimation of gradient +.>Exponential decay rate, +.>Representing a constant->And->Representing a constant.
The formulas corresponding to the encoder and decoder of step S3-1 are as follows:
wherein,indicate->Hidden features extracted by convolution kernel, +.>Represents the pooling function of the encoder, +.>Indicating->Weight matrix of convolution kernels, +.>Representing natural influence factor index and artificial influence factor index, < ->Indicating->Bias of the convolution kernels ∈>Representing reconstructed data,/->Representing all hidden feature sets, ++>Representing a summation function>Representing the%>Weight matrix of convolution kernels, +.>Representing the%>Bias of the convolution kernels ∈>Representing the upsampling function of the decoder.
The specific process of the mathematical approximation idea of step S3-5 is as follows:
setting parametersA constant value is given to any constant value, and the parameter +.>、/>、/>、/>、/>Is a value of (2); according to the parameters->、/>、/>、/>、/>And->Obtaining a balanced curved surface equation and a bifurcation curve equation; selecting a known ecological environment point far away from the bifurcation curve for verification; if 90% of the verification points are distributed around the bifurcation curve, the parameter +.>And if the given value meets the solving requirement, continuing to assign the solution.
In one embodiment of the present invention, the establishment of the evaluation index system is required to follow the following principles:
1. scientificity is combined with operability; establishing an ecological environment evaluation index system must deeply understand the ecological environment problem of the evaluation area, and the index selection and the index system establishment should follow scientific principles; the constructed index system should objectively reflect the essential problems and system characteristics of the ecological environment in the evaluation area, and the feasibility, reliability and the like of index data acquisition and quantification should be considered in index selection.
2. Integrity is combined with dominance; the evaluation indexes are selected by taking natural environment indexes into consideration and also taking artificial influence into consideration, so that the completeness of index selection is ensured; meanwhile, key problems and dominant factors affecting the ecological environment of the evaluation area are considered in important points in index selection, the selected indexes can comprehensively and truly represent the ecological environment condition, and the condition that the evaluation index system is too complicated due to excessive number of evaluation indexes is avoided.
3. Systematically and pertinently combined; as a complete ecological environment system, an evaluation index which can represent the internal structure of the system and reflect the overall function of the system is selected from the system perspective; meanwhile, according to the characteristics of complexity of an ecological system, the stability of the index in a certain period is kept, and the evaluation is convenient.
The mean square error is the mean value of the sum of squares of the differences between the target value and the predicted value, and is the most widely used error analysis method in the current regression loss function; the mean square error loss function is a commonly used loss function and is beneficial to using a gradient descent algorithm, and has the advantages that the function curve is smooth and continuously conductive everywhere, and the gradient is reduced along with the reduction of the error, so that the data convergence is facilitated, and the data convergence to the minimum value can be faster even if the learning rate is fixed. The Adam optimization algorithm is an expansion of the gradient descent optimization algorithm, and is a self-adaptive learning rate method for estimating each weight parameter; the purpose is to speed up the optimization process, such as reducing the number of iterations to an optimal value, or to increase the ability of the optimization algorithm.
Parameters of the convolutional self-coding deep neural network are set as follows: an Adam parameter optimizer is selected, the Batchsize is set to 16, the initial learning rate is 0.005, and the epoch is set to 60, so that the test precision is ensured to be stable at a higher level;is->，/>And->Between 0 and 1.
As shown in fig. 2, the training process of the convolutional self-coding deep neural network is as follows:
will be respectively input and the dimensions areNatural influence factor index and artificial influence factor index of (2) are converted into +.>Picture downsampled to +.>The method comprises the steps of carrying out a first treatment on the surface of the And then convolved with layer 2 (downscaledThe method comprises the steps of carrying out a first treatment on the surface of the Expansion into dimension by data>A one-dimensional array of (a) is provided; the input data size is restored by the downscaling of the full connection layer 1 and the up-sampling of the full connection layer 2, and the input data dimension is +.>The training is completed. In FIG. 2, input represents an Input layer, CONV2D_1 represents a convolution layer 1, CONV2D_2 represents a convolution layer 2, flatten represents data expansion, and Dense_1 and Dense_2 represent a full connection layer 1 and a full connection layer 2, respectively>Representing the number of pixels of a picture,/>Representing dimensions->Representing the length and width of the picture,/->Representing the scale result of convolution layer 2 Dense_2, i.e. +.>。
As shown in fig. 3, regarding the equilibrium surface equation, it is necessary to setFor a fixed value, ->And->To form a linear relationship, the equilibrium curved surface is formed by parallel to the plane +.>And->Is formed by a straight line of (a). Wherein, when->State variable +.>And control variable +.>Positive correlation is presented; when->State variable +.>With control variable->But increases when the control variable +.>When a certain range is reached, the control variable +.>Causes a small increase in the state variable +.>The discontinuity increases drastically.
Natural and artificial impact synthetic index under new coordinate system, and critical value of oscillation point of which occurrence is greater than thresholdAnd->And a critical value for the amplitude of oscillation which tends to be gentle or discontinuous +.>And->Points can be knownThe origin of the coordinate system in an ideal state is also a demarcation point with or without mutation for the critical point of the mutation of the ecological geological environment, so that the plane also passes through the origin of the coordinate system in the ideal state. So get->The coordinate points in the original coordinate system areAnd->The direction vector of the axis is +.>So it is perpendicular to->The plane formula of the axis passing through the original coordinate system origin is as follows:
point to PointSubstituting the above formula, the formula can be obtained:
similarly, the pointAlso, the mutation phenomenon occurs, the coordinate point corresponding to the original coordinate system isAnd is +.>Intersection and abrupt change of axis and odd point set of balanced curved surface, i.e.)>And (5) a dot. From the above, it can be seen that->At->The dots have three roots, respectively +.>、And->Then the formula is obtained:
three roots are present and satisfied. Due toThen the formula is obtained:
let the formula:
the corresponding coefficients are equal and brought into step S3-4And->To solve the formula of->The point is a mutation point of the ecological environment condition, the ecological environment control variable coefficient is increased and slowly increased within a certain range before and after the point, thereby obtaining the ecological environment condition of +.>The point realizes transition to ensure that all data are +.>The point implementation transitions to the formula in step S3-5.
The ecological environment evaluation model is constructed by the following formula:
will beSet as unknown number, < >>Andthe formula of step S3-6 is obtained by using the Cardano formula as a constant.
As shown in FIG. 4, the state variable ecological environment conditions of the natural influencing factors corresponding to the mutation point 1Is at a value ofThe ecological environment state of the state variable of the natural influencing factors corresponding to the mutation point 2>Is at a value ofThe fitted curve between the mutation point 1 and the mutation point 2 is in an ascending trend, and the derivative curve is stably changed. As shown in FIG. 5, the state variable ecological environment condition of the artificial influence factor corresponding to mutation point 1 +.>Is at a value ofThe ecological environment state of the artificial influence factor corresponding to the mutation point 2>Is at a value ofThe fitting curve and the area curve between the mutation point 1 and the mutation point 2 have sharp rising trend.
In conclusion, the convolution self-encoder and the mutation theory are effectively fused, so that the high-dimensional nonlinear characteristics of the ecological environment evaluation index system are ensured, the artificial subjectivity of the determination of the weight of the ecological environment evaluation index is overcome, the objectivity is improved, and the problem of lower accuracy of the ecological environment evaluation result caused by the uncertainty of the traditional mutation theory application mode and the dependence of independent parameter adjustment on each index layer is solved.
Claims (8)
1. An ecological environment evaluation method based on convolution self-coding and sharp point mutation models is characterized by comprising the following steps of: the method comprises the following steps:
s1, establishing an evaluation index system; normalizing each evaluation index to obtain a processed evaluation index set;
s2, analyzing the processed evaluation index set through a Pearson correlation coefficient, and reserving the evaluation index with the correlation lower than a threshold value to obtain a pruned evaluation index set;
s3, inputting the pruned evaluation index set into a convolution self-encoder to obtain a dimension reduction analysis result of the artificial influence factors and the natural influence factors; based on the dimension reduction analysis result of the convolution self-encoder, constructing an ecological environment evaluation model by utilizing a sharp point mutation theory;
s4, inputting a dimension reduction analysis result of the index to be evaluated into an ecological environment evaluation model to obtain an ecological environment evaluation result, and completing evaluation;
s3-1, respectively inputting natural influence factor indexes and artificial influence factor indexes into a convolution self-encoder, and performing dimension reduction processing through the encoder to obtain corresponding initial compression characteristics; reducing the bit number occupied by each pixel by the two initial compression characteristics through a quantizer to obtain corresponding final compression characteristics; inputting the two final compression characteristics to a decoder to obtain corresponding reconstruction data, namely a natural influence comprehensive index and an artificial influence comprehensive index;
s3-2, performing function fitting and statistical analysis on the natural influence comprehensive index and the artificial influence comprehensive index to respectively obtain critical values of oscillation points of the natural influence comprehensive index and the artificial influence comprehensive index, wherein the occurrence of the critical values is larger than a threshold valueAnd->And a critical value for the amplitude of oscillation which tends to be gentle or discontinuous +.>And->；
S3-3, establishing a balanced curved surface equation according to the point mutation theory, wherein the equation is as follows:
wherein,、/>and->Original three-dimensional coordinate system respectively representing balanced curved surfaces, +.>、/>Representing the coefficients;
s3-4, according to a coordinate conversion formula:
original three-dimensional coordinate system of balanced curved surfaceConversion to a new three-dimensional coordinate System->The method comprises the steps of carrying out a first treatment on the surface of the Wherein,representing the amount of spatial translation, +.>Representing the angle of rotation of the original three-dimensional coordinate around the Z axis;
s3-5, according to the formula:
performing function fitting analysis on the natural influence comprehensive index and the artificial influence comprehensive index to obtain corresponding fittingA curve; conducting derivation and integration on the two fitting curves to obtain corresponding mutation positions; according to the mutation position, MATLAB and the approximation idea are adopted to obtain、/>、/>、/>、/>、/>Is a value of (2); wherein (1)>Representing a maximum function;
s3-6, according to the formula:
constructing an ecological environment evaluation model, and calculating a corresponding ecological environment evaluation value based on the parameter values obtained in the step S3-5The method comprises the steps of carrying out a first treatment on the surface of the Wherein the ecological environment evaluation value->The judgment standard of (2) is as follows:
as an ecological environment evaluation valueWhen then theThe ecological environment condition is excellent;
as an ecological environment evaluation valueWhen the method is used, a natural breakpoint method is adopted for grading research, and the grades are good, medium, poor and poor respectively;
the corresponding formulas of the encoder and the decoder in the step S3-1 are as follows:
wherein,indicate->Hidden features extracted by convolution kernel, +.>Represents the pooling function of the encoder, +.>Indicating->Weight matrix of convolution kernels, +.>Representing natural influence factor index and artificial influence factor index, < ->Indicating->Bias of the convolution kernels ∈>Representing reconstructed data,/->Representing all hidden feature sets, ++>Representing a summation function>Representing the%>Weight matrix of convolution kernels, +.>Representing the%>Bias of the convolution kernels ∈>Representing the upsampling function of the decoder.
2. The ecological environment evaluation method based on convolution self-coding and sharp point mutation models according to claim 1, wherein the method comprises the following steps: the formula of the normalization processing in the step S1 is as follows:
wherein, the two formulas are respectively used for positive correlation evaluation indexes and negative correlation evaluation indexes,representing a normalized value indicative of the evaluation index, < + >>A true value indicative of the positive correlation evaluation index, < >>Represents the minimum value in the same positive correlation evaluation index,represents the maximum value in the same positive correlation evaluation index, < >>A true value indicative of the negative correlation evaluation index, < >>Represents the minimum value in the same negative correlation evaluation index, < ->Represents the maximum value of the same negative correlation evaluation index.
3. The ecological environment evaluation method based on convolution self-coding and sharp point mutation models according to claim 1, wherein the method comprises the following steps: the formula of the correlation coefficient of the step S2 is as follows:
wherein,representing the correlation coefficient>Representing a summation function>Representing sample data->Variable value of>Representing sample data->Average value of>Representing sample data->Variable value of>Representing sample data->Average value of (2);
when (when)When the two corresponding evaluation indexes are extremely weakly correlated or uncorrelated;
when (when)When the two corresponding evaluation indexes are indicated to be weakly correlated;
when (when)When the two corresponding evaluation indexes are related in a medium degree;
when (when)When the two corresponding evaluation indexes are shown to be strongly correlated;
when (when)When the two corresponding evaluation indexes are extremely strongly correlated.
4. A method for evaluating an ecological environment based on a convolution self-coding and sharp point mutation model according to claim 3, wherein the method comprises the following steps: the threshold value of the step S2 is set to 0.6.
5. The ecological environment evaluation method based on convolution self-coding and sharp point mutation models according to claim 1, wherein the method comprises the following steps: the evaluation index set in the step S2 comprises natural influence factor indexes and artificial influence factor indexes; the convolutional self-encoder in the step S3 comprises an encoder and a decoder, wherein the encoder comprises a convolutional layer and a pooling layer; the decoder includes a deconvolution layer.
6. The ecological environment evaluation method based on convolution self-coding and sharp point mutation models according to claim 1, wherein the method comprises the following steps: the convolutional self-encoder adopts an Adam algorithm to perform parameter optimization; the weight matrix and bias parameters of each convolution layer of the convolution self-encoder are optimized by adopting a mean square error loss function; the formula of the mean square error loss function is as follows:
wherein,mean square error +.>Representing predicted values +.>Indicating target value,/->Representing data volume,/->Representing the summation function.
7. The ecological environment evaluation method based on the convolution self-coding and sharp point mutation model according to claim 6, wherein the method comprises the following steps: the Adam algorithm corresponds to the following formula:
wherein,representing the current time step->Representing the objective function to be optimized, < >>The gradient of the objective function is represented,and->Respectively representing model parameters and step size, +.>And->Representing a first moment estimate of the gradient and a second moment estimate of the gradient, respectively, < >>And->First moment estimates representing gradients, respectively +.>And second moment estimation of gradient +.>Exponential decay rate, +.>Representing a constant->And->Representing a constant.
8. The ecological environment evaluation method based on convolution self-coding and sharp point mutation models according to claim 1, wherein the method comprises the following steps: the specific process of the mathematical approximation idea of the step S3-5 is as follows:
setting parametersA constant value is given to any constant value, and the parameter +.>、/>、/>、/>、/>Is a value of (2); according to parameters、/>、/>、/>、/>And->Obtaining a balanced curved surface equation and a bifurcation curve equation; selecting a known ecological environment point far away from the bifurcation curve for verification; if 90% of the verification points are distributed around the bifurcation curve, the parameter +.>And if the given value meets the solving requirement, continuing to assign the solution.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311216232.6A CN116957422B (en) | 2023-09-20 | 2023-09-20 | Ecological environment evaluation method based on convolution self-coding and sharp point mutation model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311216232.6A CN116957422B (en) | 2023-09-20 | 2023-09-20 | Ecological environment evaluation method based on convolution self-coding and sharp point mutation model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116957422A CN116957422A (en) | 2023-10-27 |
| CN116957422B true CN116957422B (en) | 2024-01-02 |
Family
ID=88458707
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311216232.6A Active CN116957422B (en) | 2023-09-20 | 2023-09-20 | Ecological environment evaluation method based on convolution self-coding and sharp point mutation model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116957422B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111275230A (en) * | 2018-12-05 | 2020-06-12 | 辽宁工程技术大学 | Safety evaluation method for coal mine spontaneous combustion coal seam |
| CN113469269A (en) * | 2021-07-16 | 2021-10-01 | 上海电力大学 | Residual convolution self-coding wind-solar-charged scene generation method based on multi-channel fusion |
| CN113553907A (en) * | 2021-06-21 | 2021-10-26 | 山东建筑大学 | Forest ecological environment condition evaluation method based on remote sensing technology |
| CN114023449A (en) * | 2021-11-05 | 2022-02-08 | 中山大学 | Diabetes risk early warning method and system based on depth self-encoder |
| WO2023151776A1 (en) * | 2022-02-08 | 2023-08-17 | Telefonaktiebolaget Lm Ericsson (Publ) | Method and apparatus for detecting changes in an enviroment |
-
2023
- 2023-09-20 CN CN202311216232.6A patent/CN116957422B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111275230A (en) * | 2018-12-05 | 2020-06-12 | 辽宁工程技术大学 | Safety evaluation method for coal mine spontaneous combustion coal seam |
| CN113553907A (en) * | 2021-06-21 | 2021-10-26 | 山东建筑大学 | Forest ecological environment condition evaluation method based on remote sensing technology |
| CN113469269A (en) * | 2021-07-16 | 2021-10-01 | 上海电力大学 | Residual convolution self-coding wind-solar-charged scene generation method based on multi-channel fusion |
| CN114023449A (en) * | 2021-11-05 | 2022-02-08 | 中山大学 | Diabetes risk early warning method and system based on depth self-encoder |
| WO2023151776A1 (en) * | 2022-02-08 | 2023-08-17 | Telefonaktiebolaget Lm Ericsson (Publ) | Method and apparatus for detecting changes in an enviroment |
Non-Patent Citations (4)
| Title |
|---|
| Xiaofei Sun,etc.A Coupling Method for Eco-Geological Environmental Safety Assessment in Mining Areas Using PCA and Catastrophe Theory.《 Natural Resources Research》.2020,第4133-4148页. * |
| Yongjian Wang,etc.A temporal clustering method fusing deep convolutional autoencoders and dimensionality reduction methods and its application in air quality visualization.《Chemometrics and Intelligent Laboratory Systems》.2022,第227卷正文第1-6页. * |
| 基于突变级数法的绿色矿山环境度评价;汪谋;《中国优秀硕士学位论文全文数据库 工程科技Ⅰ辑(月刊)》(第4期);第B027-659页 * |
| 攀西矿集区生态地质环境安全评估预警;孙小飞;《中国优秀硕士学位论文全文数据库 工程科技Ⅰ辑(月刊)》(第2期);第B027-1601页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116957422A (en) | 2023-10-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| GB2547816B (en) | Actually-measured marine environment data assimilation method based on sequence recursive filtering three-dimensional variation | |
| CN110361778B (en) | Seismic data reconstruction method based on generation countermeasure network | |
| CN111199270B (en) | Regional wave height forecasting method and terminal based on deep learning | |
| CN111860982A (en) | Wind power plant short-term wind power prediction method based on VMD-FCM-GRU | |
| CN107463993B (en) | Medium-and-long-term runoff forecasting method based on mutual information-kernel principal component analysis-Elman network | |
| CN112836610A (en) | Land use change and carbon reserve quantitative estimation method based on remote sensing data | |
| CN112229403B (en) | Method for improving ocean gravity reconstruction accuracy based on three-dimensional correction principle of ground level | |
| CN116150897A (en) | Machine tool spindle performance evaluation method and system based on digital twin | |
| CN107239477B (en) | Geographic data support vector regression method fusing spatial correlation | |
| CN111125885A (en) | ASF correction table construction method based on improved kriging interpolation algorithm | |
| CN112862178A (en) | Ocean station water level space-time prediction method and device based on deep learning | |
| CN112257021A (en) | Selectable kriging spatial interpolation rainfall estimation method | |
| Li et al. | Application of a semivariogram based on a deep neural network to Ordinary Kriging interpolation of elevation data | |
| CN112581626B (en) | Complex curved surface measurement system based on non-parametric and multi-attention force mechanism | |
| Mossad et al. | Determination and prediction of standardized precipitation index (SPI) using TRMM data in arid ecosystems | |
| CN116957422B (en) | Ecological environment evaluation method based on convolution self-coding and sharp point mutation model | |
| CN116310851B (en) | Remote sensing image change detection method | |
| CN112949944A (en) | Underground water level intelligent prediction method and system based on space-time characteristics | |
| CN114155354B (en) | Method and device for reconstructing capacitance tomography based on graph convolution network | |
| CN110503142B (en) | Multi-source remote sensing product hierarchy Bayesian fusion method based on space-time random effect addition process model | |
| CN110794469B (en) | Gravity inversion method based on minimum geological feature unit constraint | |
| Wunderlich et al. | Estimation of channelized Hotelling observer performance with known class means or known difference of class means | |
| CN113011086A (en) | Estimation method of forest biomass based on GA-SVR algorithm | |
| CN116780524B (en) | Industrial enterprise short-term load prediction method based on LSTM deep learning | |
| CN116680988A (en) | Porous medium permeability prediction method based on Transformer network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |