CN103530525A - Method for improving risk evaluation accuracy of tailing dam based on reservoir water level - Google Patents

Abstract

The invention relates to a method for improving risk evaluation accuracy of a tailing dam based on the reservoir water level. The method particularly includes the steps that the reservoir water level and risk state information of a dam body are collected and preprocessed in real time; complex variability functional relations which probably exist between the reservoir water level and the risk state information of the dam body are established; a weight fusion method based on the similarity measure is utilized to conduct fusion processing on the multiple functional relations; convexification processing is conducted on the fused function, the risk state value interval of the dam body is obtained through a possibility mean value and the variance, risks of the dam body are evaluated, and danger classes are issued. The method for improving the risk evaluation accuracy of the tailing dam has the advantages of being high in accuracy, wide in engineering application prospect and capable of being conveniently achieved on a system module.

Description

A kind of method that improves the risk assessment accuracy of tailing dam based on reservoir level

Technical field

The present invention relates to a kind of method that improves the risk assessment accuracy of tailing dam based on reservoir level, be specially the method that improves the risk assessment accuracy of tailing dam based on reservoir level.

Background technology

Tailing dam, as a great artificial dangerous matter sources, is all a potential threat near village, enterprise and resident etc.Once tailing dam dam break, not only relates to the production safety in mine self, is more related to periphery and downstream resident's thereof lives and properties safety, very easily causes great, serious accident, the safety and stability of society in serious threat.

Reservoir level is as the most important monitoring index of tailing dam, and its height is having a strong impact on the risk status of dam body.In tailing dam actual motion, reservoir level index is carried out to Real-Time Monitoring, and study its impact on dam body risk status, be the requisite measure of guaranteeing tailing dam safe operation.But the reservoir level information that usually makes to gather due to the limitation of the dynamic of dam structure, the capable and experienced immunity of environment and monitoring system itself etc. has the very large uncertainty such as fuzzy, random.In addition, the risk status of dam body is also dynamic change, and it is complicated and changeable causing the funtcional relationship between reservoir level and dam body risk status.This is complicated and changeable is mainly reflected in two aspects: the first, and at synchronization, may there are a plurality of funtcional relationships between the two, and there is compatibility; The second, over time, funtcional relationship between the two is also changing, rather than single.So utilize single function cannot describe this relation complicated and changeable, thereby cause erroneous judgement, the misjudgement of dam body risk assessment, greatly reduce the accuracy of assessment.Therefore,, by research polytrope funtcional relationship between the two, dam-break accident is controlled to bud needs dam body methods of risk assessment effective, rational, high accuracy.

In the existing dam body risk assessment based on reservoir level, document is illustrated the importance of reservoir level mostly.But on the whole, not to represent relation between the two with single fluid mechanics function, the weights that lay particular emphasis on exactly between reservoir level and other monitoring indexes (as saturation, dam body displacement, dry beach length) are definite, and rarely have and relate to for the research of polytrope relation between reservoir level and dam body risk.At present, have to research and propose by random set and can set up the funtcional relationship between two class sets in system risk assessment, but choosing of its confidence level is limited; Also there is scholar on the basis of random set, proposed the mapping of possibility collection value and solved the limited problem of confidence level.In addition, possibility theory is as a kind of new uncertain inference method, and its possibility distributes and can also characterize well the uncertainty such as random, fuzzy of reservoir level and dam body risk status information, for tailing dam risk assessment provides describing method more accurately.Therefore,, by the Weighted Fusion between the mapping of possibility collection value and each mapping, the accuracy that improves risk assessment has important scientific basis and realistic meaning.

The monitoring of reservoir level is completed by tailing dam on-line monitoring system.On-line monitoring system is mainly by reservoir level monitoring subsystem, saturation monitoring subsystem, dry beach length monitoring subsystem, the compositions such as dam body displacement monitoring subsystem and video monitoring subsystem.Representational research contents relates in each subsystem sensor material and the development of signal acquiring system, the aspects such as early warning of the self check of the safe operation of comprehensive management module, data acquisition module, collection and storage, the transmission of data communication module, the reliability and stability of data processing module and monitoring and warning module.Integrated, long-range, real-time, the automatic analysis and assessment, the intellectuality that realize system are the development trends of tailing dam on-line monitoring system.Undoubted, although the development of these technology has improved the accuracy of dam body risk assessment to a certain extent, not from the at all of problem, do not find the polytrope relation existing between each single index and dam body risk status; Do not utilize polytrope relation to assess dam body, caused erroneous judgement, the misjudgement of risk assessment, greatly reduce the accuracy of assessment, had a strong impact on Evaluated effect.

Considering above situation can find out, is badly in need of a kind of method and is specifically designed to polytrope relation and the impact of reservoir level on dam body solving between reservoir level and dam body risk status, to improve the risk assessment ability of tailing dam on-line monitoring system.

Summary of the invention

The present invention cannot effectively describe because of single function the low problem of risk assessment accuracy that the complex relationship between reservoir level and dam body risk status causes in order to solve in existing tailing dam risk assessment, and a kind of method that improves the risk assessment accuracy of tailing dam based on reservoir level is provided.

The present invention adopts following technical scheme to realize: a kind of method that improves the risk assessment accuracy of tailing dam based on reservoir level, comprises the following steps:

Utilize the reservoir level monitoring subsystem of tailing dam to carry out Real-time Collection to reservoir level monitoring information x ', and reservoir level monitoring information x ' is converted into gaussian random fuzzy variable

conversion method is: continuous recording reservoir level monitoring information x ' in same time interval of delta t, carry out k time, each recorded information 20～30 times, the reservoir level monitoring information x ' of each record is designated as one group, every group of reservoir level monitoring information carried out to standardization and obtain reservoir level standardized information x, calculate the average M of every group of reservoir level standardized information x _kand variances sigma _k, build reservoir level Gaussian random variable x _k～N (M _k,

), by average M _kbe converted into Triangular Fuzzy Number

triangular Fuzzy Number probability distribution function is

and b _k=c _k, a _kfor minimum average, d _kfor maximum average, b _k, c _kfor the maximum possibility average of quantity, by reservoir level Gaussian random variable x _kchange into gaussian random fuzzy variable ${\mathrm{\π}}_{{\stackrel{\stackrel{\‾}{~}}{\mathrm{\ξ}}}_k}\left(x_k\right)=\underset{M_k}{\sup }\left\{{\mathrm{\π}}_{{\tilde{M}}_k}\left(s\right)\right|x_k~N(M_k,{\mathrm{\σ}}_k^2)\},$ S is Triangular Fuzzy Number value space;

Utilize expert system to analyze the reservoir level monitoring information x ' gathering, obtain dam body risk status information y _k, and be translated into trapezoidal Random-fuzzy variable

conversion method is: by expert system, according to the derailment criteria of dam body, utilize analytical hierarchy process pair to pass judgment on the dam body state of k group reservoir level monitoring information x ' in the same time period, obtain the k group dam body risk status information y that k group reservoir level monitoring information is corresponding _k, k is organized to dam body risk status information y _kbe converted into trapezoidal Random-fuzzy variable

be that dam body risk status information is trapezoidal Random-fuzzy variable

wherein, a ' _k, b ' _k, c ' _k, d ' _kby expert system, determined;

Set up gaussian random fuzzy variable (x _k) and trapezoidal Random-fuzzy variable (y _k) between polytrope function, k is organized to gaussian random fuzzy variable

(x _k) and k organize trapezoidal Random-fuzzy variable (y _k) with possibility composition rule, synthesize respectively, obtain reservoir level monitoring composite signal

and dam body risk status composite signal (x)

(y), in formula, ∨, for getting macrooperation symbol, then determines reservoir level monitoring composite signal

and dam body risk status composite signal (x)

(y) the shadow that falls:

λ ∈ [0,1] and A in formula _λbe respectively reservoir level monitoring composite signal

(x) confidence level and the shadow that falls; α ∈ [0,1] and B _αbe respectively dam body risk status composite signal

(y) confidence level and the shadow that falls, utilize extension Principle to set up reservoir level monitoring composite signal

and dam body risk status composite signal (x)

(y) combine to fall shadow: in formula, ∪ is union symbol, and Y is dam body risk status information y _kvalue space,

for reservoir level monitoring composite signal

(x) the shadow A that falls _λbenefit, obtain the reservoir level monitoring composite signal after synthetic

and dam body risk status composite signal (x)

(y) the Copula relation between

(x, y); When reservoir level standardized information x is not unique in value space X, its value changes, and uses fuzzy set x ₀∈ X carrys out library representation water level standardized information x, will

(x ₀, y) carry out standardization, obtain reservoir level monitoring composite signal (x) with dam body risk status composite signal

(y) polytrope funtcional relationship: π _ζ(x ₀, y)=(1-A (x ₀)) ∨ A (x ₀) B (y), in formula

$A\left(x_0\right)=\left\{{\mathrm{\π}}_{\stackrel{\stackrel{\‾}{~}}{\mathrm{\ξ}}}\left(x_0\right)\right|x_0\∈\stackrel{\‾}{\widetilde{\mathrm{\ξ}}}(\mathrm{\λ},\mathrm{\α})\}=\{{\mathrm{\π}}_{\stackrel{\‾}{\widetilde{\mathrm{\ξ}}}}\left(x_0\right)|x_0\∈A_{\mathrm{\λ}}\},$

$B\left(y\right)=\left\{{\mathrm{\π}}_{\stackrel{\‾}{\widetilde{\mathrm{\η}}}}\left(y\right)\right|y\∈\stackrel{\‾}{\widetilde{\mathrm{\η}}}(\mathrm{\λ},\mathrm{\α})\}=\{{\mathrm{\π}}_{\stackrel{\‾}{\widetilde{\mathrm{\η}}}}\left(y\right)|y\∈B_{\mathrm{\λ}}\};$

The Weighted Fusion method of utilization based on similarity measure is to above-mentioned polytrope function π _ζ(x ₀, the function in y) carries out fusion treatment, and fusion method is: when λ, α value are different, the reservoir level monitoring composite signal obtaining

(x) with dam body risk status composite signal

(y) plausibility function between is also different, from polytrope function π _ζ(x ₀, choose n plausibility function π in y) ₁, π ₂..., π _n, calculate in n function the similarity measure between function between two, $S({\mathrm{\π}}_i,{\mathrm{\π}}_j)=(1-|x_{{\mathrm{\π}}_i}-x_{{\mathrm{\π}}_i}\left|\right)\×\frac{\min (P_l\left({\mathrm{\π}}_i\right),P_l\left({\mathrm{\π}}_j\right))}{\max (P_l\left({\mathrm{\π}}_i\right),P_l\left({\mathrm{\π}}_j\right))}\×\frac{\min (A\left({\mathrm{\π}}_i\right),A\left({\mathrm{\π}}_j\right))+\min (w_{{\mathrm{\π}}_i},w_{{\mathrm{\π}}_j})}{\max (A\left({\mathrm{\π}}_i\right),A\left({\mathrm{\π}}_j\right))+\max (w_{{\mathrm{\π}}_i},w_{{\mathrm{\π}}_j})}$ , in formula, i, j=1,2 ..., n, and i ≠ j,

be respectively function π _iand π _jcenter of gravity; P _l(π _i), A (π _i) be respectively function π _igirth and area, P _l(π _j), A (π _j) be respectively function π _jgirth and area;

be respectively function π _iand π _jmaximal value, set up similarity measure S (π _i, π _j) similarity matrix D, according to similarity matrix D, calculate the degree of belief β of each plausibility function _i, using it as i plausibility function π _iweights, be weighted fusion, obtain fusion function ${\mathrm{\&pi;}}_F(x_0,y)=\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{\mathrm{\&beta;}}_i{\mathrm{\&pi;}}_i({\mathrm{<figure-callout\; id="0"\; label="x"\; filenames="131024173912.png,131024173924.png"\; state="\{\{state\}\}">x</figure-callout>}}_0,y);$

To fusion function π _f(x ₀, y) carry out convexification processing, calculate possibility average M and possibility variances sigma that convexification is processed rear function ², providing the risk status value interval [M-σ, M+ σ] of dam body, the danger classes method according to dividing tailing dam, draws dam body danger classes.

First the present invention has monitored k group reservoir level monitoring information x ', for unified dimension, is carried out standardization, tries to achieve the average M of reservoir level standardized information x _kand variances sigma _k, build reservoir level Gaussian random variable

special must be by average M _kbe converted into Triangular Fuzzy Number by reservoir level standardized information x, it is gaussian random fuzzy variable ${\mathrm{\&pi;}}_{{\stackrel{\stackrel{\&OverBar;}{~}}{\mathrm{\&xi;}}}_k}\left(x_k\right)=\underset{M_k}{\sup }\left\{{\mathrm{\&mu;}}_{{\tilde{M}}_k}\left(s\right)\right|x_k~N(M_k,{\mathrm{\&sigma;}}_k^2)\},$ Recycling expert system is analyzed the reservoir level monitoring information x ' gathering, and obtains dam body risk status information y _k, and be translated into trapezoidal Random-fuzzy variable

by the gaussian random fuzzy variable obtaining ${\mathrm{\&pi;}}_{{\stackrel{\stackrel{\&OverBar;}{~}}{\mathrm{\&xi;}}}_k}\left(x_k\right)=\underset{M_k}{\sup }\left\{{\mathrm{\&mu;}}_{{\tilde{M}}_k}\left(s\right)\right|x_k~N(M_k,{\mathrm{\&sigma;}}_k^2)\}$ With trapezoidal Random-fuzzy variable

polytrope function π between synthetic reservoir level and dam body risk status _ζ(x ₀, y)=(1-A (x ₀)) ∨ A (x ₀) B (y), and then calculate polytrope function π _ζ(x ₀, y)=(1-A (x ₀)) ∨ A (x ₀) similarity between function between two in B (y), calculate the degree of belief β of each function _i, to polytrope function π _ζ(x ₀, y)=(1-A (x ₀)) ∨ A (x ₀) function in B (y) is weighted fusion, obtains fusion function

after again fusion function convexification being processed, obtain possibility average M and possibility variances sigma that convexification is processed rear function ², provide the risk status value interval [M-σ, M+ σ] of dam body, just can, according to the danger classes method of dividing tailing dam, draw the risk class of tailing dam, reservoir level monitoring information and dam body risk status information are changed into respectively to triangle Random-fuzzy variable and trapezoidal stochastic variable, the randomness of information has not only been described, and its ambiguity described, make the description of information more accurate, for the foundation of mapping relations between the two provides information more accurately, utilize the mapping of possibility collection value to set up the polytrope funtcional relationship between reservoir level monitoring information and dam body risk status information, can reflect from different sides the risk status of dam body, not only solve single mapping relation and cannot describe the problem of this complex relationship, and improved the accuracy of dam body risk assessment, the polytrope function that the present invention tries to achieve between reservoir level and dam body risk status comprises the funtcional relationship between a plurality of reservoir levels and dam body risk status, the fusion function that these funtcional relationship Weighted Fusions are tried to achieve has just comprised the dynamic relation between reservoir level and dam body risk status, this fusion function just can give expression to the complexity concerns between reservoir level and dam body risk status, interval [the M-σ of risk status value drawing thus, M+ σ] just can be more comprehensive, reflect accurately the risk status of dam body.

Expert system is comprised of multidigit tailing dam risk assessment expert, every expert utilizes analytical hierarchy process to draw tailing dam risk status information according to the derailment criteria of generally acknowledged tailing dam, the tailing dam risk status information that every expert draws is in predictable scope, and the trapezoidal Random-fuzzy variable therefore being transformed by tailing dam risk status information is definite variable.

A kind of above-mentioned method that improves the risk assessment accuracy of tailing dam based on reservoir level, according to different λ and α value from polytrope function π _ζ(x ₀, choose n function π in y) ₁, π ₂..., π _ntime, λ and α are 0.75～0.95 choosing value, and the polytrope function of selecting is reliable, and then the risk status value finally drawing interval [M-σ, M+ σ] is more accurately with comprehensive.

By " modified rule of Nine ", by the dangerous grade classification of tailing dam, be five regions as following table:

Hazard level	Dangerous values (W)	Danger classes
			Fairly good risk, can accept	＜0.09	A
General dangerous, should be noted that	0.09～0.27	B
			Significantly dangerous, need rectification	0.27～0.54	C

Highly dangerous, need rectify and improve immediately	0.54～0.99	D
			Be in extreme danger, stop storehouse rectification	＞0.99	E

When reservoir level monitoring information is fuzzy set [0.68,0.72], utilize the present invention and hydrodynamic methods to carry out risk assessment to dam body, result is compared as following table:

Part is subordinate to factor Q _irefer to that the risk status value interval of dam body is positioned at the ratio of the risk status value burst length of each danger classes length of an interval degree and dam body,

in formula; The risk status value length of an interval degree that L (W) is dam body; L _ifor dam body risk status value interval is positioned at the length of i danger classes, as seen from the above table, when reservoir level monitoring information is assessed by fluid mechanics method, the risk status of dam body is C grade, exists significantly dangerously, needs rectification; And utilize the assessment result of gained of the present invention to be: the possibility that the possibility that belongs to C grade is 15.38%, belong to D grade is 84.62%, according to part, be subordinate to danger classes that the maximal value of the factor belongs to as the risk status of dam body, that is to say, when reservoir level is fuzzy set [0.68,0.72] time, in D level, there is highly dangerous in the danger classes of dam body, should rectify and improve immediately; Therefore the present invention has important superiority describing aspect the dynamic change of dam body and monitoring information uncertain, and the advantage such as this methods of risk assessment has reasonable, high accuracy, future in engineering applications are wide, has very important significance for the safety management of tailing dam.

The method of the risk assessment accuracy of raising tailing dam of the present invention based on reservoir level not only has the advantages such as practical, accuracy is high, future in engineering applications is wide, but also provide describing method more accurately for reservoir level and dam body risk status information, can easily be embedded in tailing dam on-line monitoring system, meet integrated, the automatic analysis of system and the demand of assessment, the risk assessment of dam body is had to very high using value.

Accompanying drawing explanation

Fig. 1 is process flow diagram of the present invention.

Fig. 2 is for using gaussian random fuzzy variable

the distribution plan of reservoir level monitoring information after 3 groups of standardization that represent.

Fig. 3 is with trapezoidal Random-fuzzy variable

the distribution plan of dam body risk status information after 3 groups of standardizations that represent.

Reservoir level monitoring information after being as the criterion of Fig. 4

(x) with the rear dam body risk status information of standardization

(y) the polytrope functional arrangement between.

Fig. 5 is the reservoir level that obtains according to fluid mechanics method and the functional arrangement between dam body risk status information.

Fig. 6 is the reservoir level that obtains of the present invention and the functional arrangement between dam body risk status information.

Embodiment

A method for the risk assessment accuracy of tailing dam based on reservoir level, comprises the following steps:

Utilize the reservoir level monitoring subsystem of tailing dam to carry out Real-time Collection to reservoir level monitoring information x ', and reservoir level monitoring information x ' is converted into gaussian random fuzzy variable

conversion method is: continuous recording reservoir level monitoring information x ' in same time interval of delta t, carry out k time, each recorded information 20～30 times, the reservoir level monitoring information x ' of each record is designated as one group, every group of reservoir level monitoring information carried out to standardization and obtain reservoir level standardized information x, calculate the average M of every group of reservoir level standardized information x _kand variances sigma _k, build reservoir level Gaussian random variable

by average M _kbe converted into Triangular Fuzzy Number

triangular Fuzzy Number probability distribution function is

and b _k=c _k, a _kfor minimum average, d _kfor maximum average, b _k, c _kfor the maximum possibility average of quantity, by reservoir level Gaussian random variable x _kchange into gaussian random fuzzy variable ${\mathrm{\&pi;}}_{{\stackrel{\stackrel{\&OverBar;}{~}}{\mathrm{\&xi;}}}_k}\left(x_k\right)=\underset{M_k}{\sup }\left\{{\mathrm{\&pi;}}_{{\tilde{M}}_k}\left(s\right)\right|x_k~N(M_k,{\mathrm{\&sigma;}}_k^2)\},$ S is Triangular Fuzzy Number value space;

Utilize expert system to analyze the reservoir level monitoring information x ' gathering, obtain dam body risk status information y _k, and be translated into trapezoidal Random-fuzzy variable

conversion method is: by expert system, according to the derailment criteria of dam body, utilize analytical hierarchy process pair to pass judgment on the dam body state of k group reservoir level monitoring information x ' in the same time period, obtain the k group dam body risk status information y that k group reservoir level monitoring information is corresponding _k, k is organized to dam body risk status information y _kbe converted into trapezoidal Random-fuzzy variable

be that dam body risk status information is trapezoidal Random-fuzzy variable

wherein, a ' _k, b ' _k, c ' _k, d ' _kby expert system, determined;

Set up gaussian random fuzzy variable

(x _k) and trapezoidal Random-fuzzy variable

(y _k) between polytrope function, k is organized to gaussian random fuzzy variable

(x _k) and k organize trapezoidal Random-fuzzy variable (y _k) with possibility composition rule, synthesize respectively, obtain reservoir level monitoring composite signal and dam body risk status composite signal (x) (y),

in formula, ∨, for getting macrooperation symbol, then determines reservoir level monitoring composite signal

and dam body risk status composite signal (x)

(y) the shadow that falls:

λ ∈ [0,1] and A in formula _λbe respectively reservoir level monitoring composite signal

(x) confidence level and the shadow that falls; α ∈ [0,1] and B _αbe respectively dam body risk status composite signal

(y) confidence level and the shadow that falls, utilize extension Principle to set up reservoir level monitoring composite signal

and dam body risk status composite signal (x)

(y) combine to fall shadow:

in formula, ∪ is union symbol, and Y is dam body risk status information y _kvalue space,

for reservoir level monitoring composite signal

(x) the shadow A that falls _λbenefit, obtain the reservoir level monitoring composite signal after synthetic

and dam body risk status composite signal (x)

(y) Copula between is related to π ' _ζ(x, y); When reservoir level standardized information x is not unique in value space X, its value changes, and uses fuzzy set x ₀∈ X carrys out library representation water level standardized information x, by π ' _ζ(x ₀, y) carry out standardization, obtain reservoir level monitoring composite signal

(x) with dam body risk status composite signal

(y) polytrope funtcional relationship: π _ζ(x ₀, y)=(1-A (x ₀)) ∨ A (x ₀) B (y), in formula

$A\left(x_0\right)=\left\{{\mathrm{\&pi;}}_{\stackrel{\stackrel{\&OverBar;}{~}}{\mathrm{\&xi;}}}\left(x_0\right)\right|x_0\&Element;\stackrel{\&OverBar;}{\widetilde{\mathrm{\&xi;}}}(\mathrm{\&lambda;},\mathrm{\&alpha;})\}=\{{\mathrm{\&pi;}}_{\stackrel{\&OverBar;}{\widetilde{\mathrm{\&xi;}}}}\left(x_0\right)|x_0\&Element;A_{\mathrm{\&lambda;}}\},$

$B\left(y\right)=\left\{{\mathrm{\&pi;}}_{\stackrel{\&OverBar;}{\widetilde{\mathrm{\&eta;}}}}\left(y\right)\right|y\&Element;\stackrel{\&OverBar;}{\widetilde{\mathrm{\&eta;}}}(\mathrm{\&lambda;},\mathrm{\&alpha;})\}=\{{\mathrm{\&pi;}}_{\stackrel{\&OverBar;}{\widetilde{\mathrm{\&eta;}}}}\left(y\right)|y\&Element;B_{\mathrm{\&lambda;}}\};$

The Weighted Fusion method of utilization based on similarity measure is to above-mentioned polytrope function π _ζ(x ₀, the function in y) carries out fusion treatment, and fusion method is: when λ, α value are different, the reservoir level monitoring composite signal obtaining

(x) with dam body risk status composite signal

(y) plausibility function between is also different, from polytrope function π _ζ(x ₀, choose n plausibility function π in y) ₁, π ₂..., π _n, calculate in n function the similarity measure between function between two, $S({\mathrm{\&pi;}}_i,{\mathrm{\&pi;}}_j)=(1-|x_{{\mathrm{\&pi;}}_i}-x_{{\mathrm{\&pi;}}_i}\left|\right)\&times;\frac{\min (P_l\left({\mathrm{\&pi;}}_i\right),P_l\left({\mathrm{\&pi;}}_j\right))}{\max (P_l\left({\mathrm{\&pi;}}_i\right),P_l\left({\mathrm{\&pi;}}_j\right))}\&times;\frac{\min (A\left({\mathrm{\&pi;}}_i\right),A\left({\mathrm{\&pi;}}_j\right))+\min (w_{{\mathrm{\&pi;}}_i},w_{{\mathrm{\&pi;}}_j})}{\max (A\left({\mathrm{\&pi;}}_i\right),A\left({\mathrm{\&pi;}}_j\right))+\max (w_{{\mathrm{\&pi;}}_i},w_{{\mathrm{\&pi;}}_j})}$ , in formula, i, j=1,2 ..., n, and i ≠ j, be respectively function π _iand π _jcenter of gravity; P _l(π _i), A (π _i) be respectively function π _igirth and area, P _l(π _j), A (π _j) be respectively function π _jgirth and area; be respectively function π _iand π _jmaximal value, set up similarity measure S (π _i, π _j) similarity matrix D, according to similarity matrix D, calculate the degree of belief β of each plausibility function _i, using it as i plausibility function π _iweights, be weighted fusion, obtain fusion function ${\mathrm{\&pi;}}_F(x_0,y)=\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{\mathrm{\&beta;}}_i{\mathrm{\&pi;}}_i({\mathrm{<figure-callout\; id="0"\; label="x"\; filenames="131024173912.png,131024173924.png"\; state="\{\{state\}\}">x</figure-callout>}}_0,y);$

To fusion function π _f(x ₀, y) carry out convexification processing, calculate possibility average M and possibility variances sigma that convexification is processed rear function ², providing the risk status value interval [M-σ, M+ σ] of dam body, the danger classes method according to dividing tailing dam, draws dam body danger classes.

A kind of above-mentioned method that improves the risk assessment accuracy of tailing dam based on reservoir level, according to different λ and α value from polytrope function π _ζ(x ₀, choose n function π in y) ₁, π ₂..., π _ntime, λ and α are 0.75～0.95 choosing value.

During concrete enforcement, λ and α should not be lower than 0.6, and generally 0.75～0.95 choosing value, the polytrope funtcional relationship drawing is reliable; To fusion function π _f(x ₀, y) carry out convexification processing, according to following rule treatments: for any interval y ∈ [y _p, y _p+1], if π _f(x ₀, y _p)≤π _f(x ₀, y _p+1), work as π _f(x ₀, y)≤π _f(x ₀, y _p) time π _f(x ₀, y)=π _f(x ₀, y _p); If π _f(x ₀, y _p)>=π _f(x ₀, y _p+1), work as π _f(x ₀, y)≤π _f(x ₀, y _p+1) time π _f(x ₀, y)=π _f(x ₀, y _p+1); In formula, y _p, y _p+1for the corresponding horizontal ordinate of maximum value, and p=1,2 ..., r-1, r is maximum value number.Calculate convexification and process possibility average M and the possibility variances sigma of rear function ², $\{\begin{array}{c}M={\&Integral;}_0^1(a_1\left(\mathrm{\&mu;}\right)+a_2\left(\mathrm{\&mu;}\right))\mathrm{\&mu;d\&mu;}\\ {\mathrm{\&sigma;}}^2={\&Integral;}_0^1\frac{1}{6}[a_1\left(\mathrm{\&mu;}\right)+a_2\left(\mathrm{\&mu;}\right){]}^2\mathrm{\&mu;d\&mu;}\end{array};$ In formula, μ ∈ [0,1] is the confidence level of the rear function of convexification processing; a ₁(μ), a ₂(μ) be respectively two end points interval under μ confidence level.

Claims (2)

1. a method that improves the risk assessment accuracy of tailing dam based on reservoir level, is characterized in that comprising the following steps:

Utilize the reservoir level monitoring subsystem of tailing dam to carry out Real-time Collection to reservoir level monitoring information x ', and reservoir level monitoring information x ' is converted into gaussian random fuzzy variable conversion method is: continuous recording reservoir level monitoring information x ' in same time interval of delta t, carry out k time, each recorded information 20～30 times, the reservoir level monitoring information x ' of each record is designated as one group, every group of reservoir level monitoring information carried out to standardization and obtain reservoir level standardized information x, calculate the average M of every group of reservoir level standardized information x _kand variances sigma _k, build reservoir level Gaussian random variable x _k～N (M _k,

), by average M _kbe converted into Triangular Fuzzy Number

triangular Fuzzy Number probability distribution function is and b _k=c _k, a _kfor minimum average, d _kfor maximum average, b _k, c _kfor the maximum possibility average of quantity, by reservoir level Gaussian random variable x _kchange into gaussian random fuzzy variable ${\mathrm{\&pi;}}_{{\stackrel{\&OverBar;}{\widetilde{\mathrm{\&xi;}}}}_k}\left(x_k\right)=\underset{M_k}{\sup }\left\{{\mathrm{\&pi;}}_{{\tilde{M}}_k}\left(s\right)\right|x_k~N(M_k,{\mathrm{\&sigma;}}_k^2)\},$ S is Triangular Fuzzy Number value space;

Utilize expert system to analyze the reservoir level monitoring information x ' gathering, obtain dam body risk status information y _k, and be translated into trapezoidal Random-fuzzy variable

conversion method is: by expert system, according to the derailment criteria of dam body, utilize analytical hierarchy process pair to pass judgment on the dam body state of k group reservoir level monitoring information x ' in the same time period, obtain the k group dam body risk status information y that k group reservoir level monitoring information is corresponding _k, k is organized to dam body risk status information y _kbe converted into trapezoidal Random-fuzzy variable

be that dam body risk status information is trapezoidal Random-fuzzy variable wherein, a ' _k, b ' _k, c ' _k, d ' _kby expert system, determined;

Set up gaussian random fuzzy variable with trapezoidal Random-fuzzy variable

(y _k) between polytrope function, k is organized to gaussian random fuzzy variable organize trapezoidal Random-fuzzy variable with k

with possibility composition rule, synthesize respectively, obtain reservoir level monitoring composite signal

with dam body risk status composite signal

in formula, ∨, for getting macrooperation symbol, then determines reservoir level monitoring composite signal with dam body risk status composite signal

the shadow that falls:

λ ∈ [0,1] and A in formula _λbe respectively reservoir level monitoring composite signal

confidence level and the shadow that falls; α ∈ [0,1] and B _αbe respectively dam body risk status composite signal

confidence level and the shadow that falls, utilize extension Principle to set up reservoir level monitoring composite signal

with dam body risk status composite signal

combine to fall shadow:

in formula, ∪ is union symbol, and Y is dam body risk status information y _kvalue space,

for reservoir level monitoring composite signal shadow A falls _λbenefit, obtain the reservoir level monitoring composite signal after synthetic

with dam body risk status composite signal

between Copula relation

when reservoir level standardized information x is not unique in value space X, its value changes, and uses fuzzy set x ₀∈ X carrys out library representation water level standardized information x, will (x ₀, y) carry out standardization, obtain reservoir level monitoring composite signal

(x) with dam body risk status composite signal

(y) polytrope funtcional relationship: π _ζ(x ₀, y)=(1-A (x ₀)) ∨ A (x ₀) B (y), in formula

$A\left(x_0\right)=\left\{{\mathrm{\&pi;}}_{\stackrel{\stackrel{\&OverBar;}{~}}{\mathrm{\&xi;}}}\left(x_0\right)\right|x_0\&Element;\stackrel{\&OverBar;}{\widetilde{\mathrm{\&xi;}}}(\mathrm{\&lambda;},\mathrm{\&alpha;})\}=\{{\mathrm{\&pi;}}_{\stackrel{\&OverBar;}{\widetilde{\mathrm{\&xi;}}}}\left(x_0\right)|x_0\&Element;A_{\mathrm{\&lambda;}}\},$

$B\left(y\right)=\left\{{\mathrm{\&pi;}}_{\stackrel{\&OverBar;}{\widetilde{\mathrm{\&eta;}}}}\left(y\right)\right|y\&Element;\stackrel{\&OverBar;}{\widetilde{\mathrm{\&eta;}}}(\mathrm{\&lambda;},\mathrm{\&alpha;})\}=\{{\mathrm{\&pi;}}_{\stackrel{\&OverBar;}{\widetilde{\mathrm{\&eta;}}}}\left(y\right)|y\&Element;B_{\mathrm{\&lambda;}}\};$

The Weighted Fusion method of utilization based on similarity measure is to above-mentioned polytrope function π _ζ(x ₀, the function in y) carries out fusion treatment, and fusion method is: when λ, α value are different, the reservoir level monitoring composite signal obtaining

(x) with dam body risk status composite signal

(y) plausibility function between is also different, from polytrope function π _ζ(x ₀, choose n plausibility function π in y) ₁, π ₂..., π _n, calculate in n function the similarity measure between function between two, $S({\mathrm{\&pi;}}_i,{\mathrm{\&pi;}}_j)=(1-|x_{{\mathrm{\&pi;}}_i}-x_{{\mathrm{\&pi;}}_i}\left|\right)\&times;\frac{\min (P_l\left({\mathrm{\&pi;}}_i\right),P_l\left({\mathrm{\&pi;}}_j\right))}{\max (P_l\left({\mathrm{\&pi;}}_i\right),P_l\left({\mathrm{\&pi;}}_j\right))}\&times;\frac{\min (A\left({\mathrm{\&pi;}}_i\right),A\left({\mathrm{\&pi;}}_j\right))+\min (w_{{\mathrm{\&pi;}}_i},w_{{\mathrm{\&pi;}}_j})}{\max (A\left({\mathrm{\&pi;}}_i\right),A\left({\mathrm{\&pi;}}_j\right))+\max (w_{{\mathrm{\&pi;}}_i},w_{{\mathrm{\&pi;}}_j})}$ , in formula, i, j=1,2 ..., n, and i ≠ j, be respectively function π _iand π _jcenter of gravity; P _l(π _i), A (π _i) be respectively function π _igirth and area, P _l(π _j), A (π _j) be respectively function π _jgirth and area;

be respectively function π _iand π _jmaximal value, set up similarity measure S (π _i, π _j) similarity matrix D, according to similarity matrix D, calculate the degree of belief β of each plausibility function _i, using it as i plausibility function π _iweights, be weighted fusion, obtain fusion function ${\mathrm{\&pi;}}_F(x_0,y)=\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{\mathrm{\&beta;}}_i{\mathrm{\&pi;}}_i(x_0,y);$

To fusion function π _f(x ₀, y) carry out convexification processing, calculate possibility average M and possibility variances sigma that convexification is processed rear function ², providing the risk status value interval [M-σ, M+ σ] of dam body, the danger classes method according to dividing tailing dam, draws dam body danger classes.

2. a kind of method that improves the risk assessment accuracy of tailing dam based on reservoir level according to claim 1, is characterized in that according to different λ and α value from polytrope function π _ζ(x ₀, choose n function π in y) ₁, π ₂..., π _ntime, λ and α are 0.75～0.95 choosing value.

Images