CN105160448A - Electricity transaction monitoring risk index evaluation method based on unascertained rational numbers - Google Patents

Abstract

The invention relates to the field of electricity transactions, in particular to an electricity transaction monitoring risk index evaluation method based on unascertained rational numbers. The method comprises the steps of firstly, performing integrated credibility modeling for electricity market experts, obtaining integrated credibility of the experts participating in the marking; secondly, the experts marking secondary indexes by reference to index values of the secondary indexes; thirdly, based on the credibility and marks given by the experts, obtaining marks, which belong to unascertained rational numbers, of the secondary indexes; and fourthly, combining the weights of the secondary indexes and corresponding marks, calculating the marks of primary indexes that the second indexes belong, and calculating the mark of a total goal on the analogy of this. The invention has the beneficial effects of being capable of electricity transaction monitoring risk index evaluation by using the unascertained rational numbers, acquiring a variety of possible values of the evaluation results and corresponding credibility, and achieving risk index quantification and effective risk monitoring.

Description

A kind of electricity transaction supervision risk index Evaluation Method based on Uncertainty number

Technical field

The present invention relates to electricity transaction field, specifically, relate to a kind of electricity transaction supervision risk integrated evaluating method based on Uncertainty number.

Background technology

Along with the foundation in unified interconnecting electric power market, electricity transaction object and mode more complicated, there is larger uncertainty in the market behavior, causes power market transaction supervision risk day by day to increase.In order to avoid power market transaction supervision risk constantly worsens, affect market order and bring more risk to electricity transaction, just in the urgent need to setting up electricity transaction supervision risk appraisement system.

In general, appraisement system is made up of evaluation index, index weights and evaluation method three parts.For the evaluation in electricity market, evaluation index is generally determined in conjunction with actual conditions by electricity market expert, index weights is determined by analytical hierarchy process, and evaluation method to apply maximum be at present exactly fuzzy overall evaluation, but the object of these fuzzy overall evaluations majority is Electricity Market Operation efficiency, Electricity Market Operation rule etc. that determinacy is stronger, relative to having for the evaluation of larger probabilistic power market transaction supervision risk, fuzzy overall evaluation can not objectively respond the uncertainties mathematics in the middle of evaluation completely.

And this uncertainty is embodied in two aspects, first due under the overall situation of current unified interconnecting electric power market, electricity transaction object and mode more complicated, cause the market behavior to there is larger uncertainty, this will cause information source when carrying out electricity transaction supervision risk metrics evaluation to have larger uncertainty; Secondly, in current power marketing supervision risk comprehensive evaluation, to the determination of evaluation index value-at-risk, usually adopt expert estimation method.But the information grasped due to different expert is different, naturally different to the evaluation of same risk index, and this information grasped due to decision maker is not enough to determine the uncertainty in the pure subjective understanding that the time of day of things and quantitative relation also can be brought.In fact, any system simultaneously with status consideration and behavial factor, the information provided is all unascertained rational substantially, to the information with uncertainties mathematics, must consider its uncertainty, and it can not be reduced to comformed information and be processed.Therefore, in the evaluation of current power marketing supervision risk, there is probabilistic objective phenomenon, the present invention proposes a kind of electricity transaction supervision risk integrated evaluating method based on Uncertainty number.

So-called " unascertained rational " is exactly the uncertainty in the pure subjective understanding that brings because information that decision maker grasps is not enough to determine the time of day of things and quantitative relation.Uncertainty in this subjective, understanding, is referred to as uncertainties mathematics.That is, it is due to the subjective and objective condition restriction of decision maker, be familiar with unclear, the information grasped is not enough, be difficult to determine the time of day of things and quantitative relation and the pure subjectivity brought, uncertainty in understanding, it had both been different from the randomness of the things that just will occur for future, was also different from the ambiguity in certain characteristic owing to can not be formed with clear and definite definition and evaluating target to some things.In many instances, the true strength of unascertained rational can represent by a number, and at this moment, unascertained rational just available reliability distribution function F (x) is uniquely determined.Uncertainty number be the most substantially, the simplest, be widely used, unascertained events easy to use, it is the popularization of real number, can delineate accurately and express various objective in " unascertained quantity ", and avoid only representing that information that these amounts produce are omitted and the defect of distortion with a real number.

At present, under unified interconnecting electric power market environment, the research specially for power market transaction supervision risk metrics evaluation does not also have.And in existing electricity market is evaluated, applying maximum is exactly fuzzy overall evaluation, but the object of these fuzzy overall evaluations majority is the Electricity Market Operation efficiency that determinacy is stronger, Electricity Market Operation rule etc., relative to having for the evaluation of larger probabilistic power market transaction supervision risk, fuzzy overall evaluation can not objectively respond the uncertainties mathematics in the middle of evaluation completely, therefore, the present invention is specially for the uncertainty that the evaluation of power market transaction supervision risk just occurs, build a kind of electricity transaction supervision risk index Evaluation Method based on Uncertainty number.

Summary of the invention

The technical problem to be solved in the present invention is:

The first, solve the uncertain problem occurred in current power transaction supervision risk metrics evaluation.In the face of current unified interconnecting electric power market environment, electricity transaction object and mode more complicated, cause the market behavior to there is larger uncertainty, cause information source when carrying out electricity transaction supervision risk metrics evaluation to have larger uncertainty; And in current power marketing supervision risk comprehensive evaluation, the information grasped due to different expert is different, the pure subjectivity that can bring, uncertainty in understanding.How processing the uncertainty occurred in electricity transaction supervision risk metrics evaluation is the first problem that the present invention will solve.

The second, carry out quantizing and defining the level for the supervision risk occurred in electricity transaction process.Along with the foundation in unified interconnecting electric power market, electricity transaction object and mode more complicated, cause power market transaction supervision risk day by day to increase.In order to avoid power market transaction supervision risk constantly worsens, even if also Timeliness coverage risk early warning can be produced after there is risk, power monitoring personnel are and guided to take corresponding measure to carry out effective decision-making, with regard to needing badly, the evaluation of electricity transaction supervision risk index is quantized, be convenient to power monitoring personnel carry out risk identification and judge risk class, realize the effective supervision to risk.

Technical scheme of the present invention is as follows:

Under current unified interconnecting electric power market environment, make every effort to accurate, objective, comprehensive reflection power market transaction supervision risk situation herein, and follow the principles such as susceptibility, representativeness, dynamic and operability, according to electricity market expertise, construct the transaction supervision risk pre-warning indexes system being applicable to unified interconnecting electric power market.As shown in Figure 1, comprise 5 class one-level warning indexs: city's field coordination, trading program, contract and clearing, energy efficiency and operation of power networks, under every class first class index, comprise some two-level index, altogether 16 two-level index.

Based on the electricity transaction supervision risk comprehensive evaluation of Uncertainty number

(1) general thought

First synthetic reliability modeling is carried out to electricity market expert, draw the synthetic reliability of each expert participating in scoring; Secondly by the desired value of each expert with reference to each two-level index, give a mark to each two-level index; Then be worth the mark of each two-level index according to the confidence level of expert with giving a mark, it belongs to Uncertainty number; Finally in conjunction with weight and the corresponding score value of each two-level index, calculate the score of first class index belonging to each two-level index, this is analogized, until calculate the score value of general objective.Wherein, because evaluation procedure relates to the computing of Uncertainty number, the exponent number of the Uncertainty number obtained after computing each time all can raise, and therefore, needs the exponent number rationally reducing Uncertainty number in computation process.

(2) electricity market expert synthetic reliability modeling

In power market transaction supervision risk comprehensive evaluation, to the determination of evaluation index value-at-risk, usually adopt expert estimation method.This is widely used in current field evaluation and a kind of effective method.The feature of this method is simple, convenient.But also there is the not enough problem of expertise and know-how in the method.The expert evaluated is carried out at scene, is generally some people, and certain desired value to be determined, be the concentrated expression of each expert analysis mode.Usual treating method is the desired value of averaging as last, but the know-how of the expert that participation is evaluated, experience, authoritative level are differentiated, some real informations during evaluation are often covered in such process, the desired value obtained is lost biased, the good overall view of expert estimation can not be reflected, the degree of reliability of desired value is reduced.

The confidence level of expert, be also the authority of expert people's degree of trusting expert in other words, represent with α (0≤α≤1), α=1 represents that certain expert is the most credible, and α=0 represents that certain expert is least worth believing.In power market transaction supervision risk comprehensive evaluation, the confidence level of expert can be determined according to the academic title of expert, educational background, length of service three aspects.Evaluation operation is in table 1, and confidence level evaluation criteria is in table 2.In the sorted table of table 2, represent expert's credibility with the quantity scale of ten point system, 10 points the highest, and 1 point minimum, and the meaning of numeric representation is therebetween successively decreased.

Table 1 expert reliability evaluation form

Project	Academic title	Educational background	Length of service	Evaluation score	Confidence level
						Expert

Table 2 expert reliability evaluation criteria sorted table

Use ε _i(i=1,2,3) are respectively the evaluation score in expert academic title, educational background, the length of service, then the confidence level β of expert _ican represent with formula (1).

${\mathrm{\β}}_i=\frac{\underset{i=1}{\overset{3}{\mathrm{\Σ}}}{\mathrm{\ϵ}}_i}{30}---\left(1\right)$

β _ivalue more close to 1, represent that this expert is the most credible, the judgement done is more accurate, otherwise, β _iless, represent that this expert is least worth believing.

If the expert participating in power market transaction supervision risk comprehensive evaluation is: B ₁, B ₂, ^, B _nconfidence level be respectively: order:

${\mathrm{\α}}_i=\frac{{\stackrel{\‾}{\mathrm{\α}}}_i}{\underset{k=1}{\overset{n}{\mathrm{\Σ}}}{\stackrel{\‾}{\mathrm{\α}}}_k},(i=\mathrm{1,2},^,n)---\left(2\right)$

Claim α _ifor expert B _iabout expert group B ₁, B ₂, ^, B _nsynthetic reliability, referred to as expert B _icomprehensive reliability.Claim

$\mathrm{\α}=\frac{1}{n}({\stackrel{\‾}{\mathrm{\α}}}_1+{\stackrel{\‾}{\mathrm{\α}}}_2+^+{\stackrel{\‾}{\mathrm{\α}}}_n)---\left(3\right)$

For expert group B ₁, B ₂, ^, B _nsynthetic reliability.α in formula represents the authoritative degree evaluating expert group.

(3) uncertainty of electricity market expert opinion quantizes

Because the cognitive behavior of expert has uncertainties mathematics, available Uncertainty number represents the evaluation information of expert, is called that the uncertainty of expert opinion quantizes.

If A is an evaluation object in power market transaction supervision risk comprehensive evaluation, A ₁, A ₂, ^, A _mfor m the index of A.N expert is had to participate in evaluating, if expert group B to A ₁, B ₂, ^, B _nexpert with centesimal system to the m of A index factor assessment marking, obtain expert estimation table, in table 3.Wherein, to factor A _i, expert estimation is respectively C _i1, C _i2, ^, C _in, expert's synthetic reliability is α ₁, α ₂, ^, α _n, existing by score set (C _i1, C _i2, ^, C _in) in score value be arranged as by getting a mode time identical: C _ij1, C _ij2, ^, C _ijkso, for factor A _iuncertainty number f can be obtained _i(x):

Wherein, i=1,2, ^, m; α _ij1, α _ij2, ^, α _ijkbe marking value be respectively C _ij1, C _ij2, ^, C _ijkthose expert's synthetic reliability sums, due to factor A _iweight be w _i, so the unascertained quantify value of evaluation object A is:

$C=\underset{i=1}{\overset{m}{\mathrm{\Σ}}}{w}_i{f}_i\left(x\right)---\left(5\right)$

Table 3 expert estimation table

Title C is the uncertainty quantification value of the expert opinion of evaluation object A.Obvious 0≤E (C)≤100, when total confidence level α=1 of unascertained quantify value C, E (C) is real number; The expert that all participations that and if only if are evaluated is to evaluation index factor A _iall play 100 timesharing, E (C)=100, and if only if all expert is to institute factor of evaluation A _iall play 0 timesharing, E (C)=0.

(4) depression of order of high-order Uncertainty number

Owing to relating to the computing of Uncertainty number in evaluation procedure, the exponent number of the Uncertainty number obtained after computing each time all can raise, and supposes that 4 three rank Uncertainty number do additive operation, then Uncertainty number exponent number is as a result up to n=3 ⁴=81.On of such a size interval, get the value of 81 non-zero confidence levels, and the confidence level of these values is all very little, from application point, such result has not had too many actual value, and calculated amount is very large.So in calculating process, the exponent number rationally reducing Uncertainty number is necessary.

If high-order Uncertainty number P (formula 6) is the result of Uncertainty number once-through operation, by x _ineighborhood [x _i, x _i+ δ] in value merge according to a certain method, can exponent number be reduced, different according to different actual conditions δ value.

First, from i=1, x is located at ₁neighborhood [x _i, x _i+ δ] in value have x ₁, x ₂, ^, x _k, corresponding confidence level is respectively α ₁, α ₂, ^, α _k, replace x by formula 7 ₁, x ₂, ^, x _k, replace α by formula 8 ₁, α ₂, ^, α _k, formula is:

and $0<\underset{i=1}{\overset{m}{\mathrm{\&Sigma;}}}{\mathrm{\&alpha;}}_i=\mathrm{\&alpha;}<\mathrm{1,0}{<\mathrm{\&alpha;}}_i\&le;1---\left(6\right)$

${\stackrel{\&OverBar;}{x}}_i=\frac{1}{\underset{k=1}{\overset{k}{\mathrm{\&Sigma;}}}{\mathrm{\&alpha;}}_i}\underset{i=1}{\overset{k}{\mathrm{\&Sigma;}}}{x}_i{\mathrm{\&alpha;}}_i---\left(7\right)$

${\stackrel{\&OverBar;}{\mathrm{\&alpha;}}}_i=\underset{i=1}{\overset{k}{\mathrm{\&Sigma;}}}{\mathrm{\&alpha;}}_i---\left(8\right)$

So just k-1 rank are fallen in Uncertainty number P, then remaining value has been continued in the same way to merge, by that analogy.After depression of order, Uncertainty number P can be expressed as

wherein $0<{\stackrel{\&OverBar;}{\mathrm{\&alpha;}}}_i<1,i=\mathrm{1,2},^,m,\underset{i=1}{\overset{k}{\mathrm{\&Sigma;}}}{\stackrel{\&OverBar;}{\mathrm{\&alpha;}}}_i=1---\left(9\right)$

Beneficial effect of the present invention is:

The present invention is relative to prior art, the uncertain problem occurred in electricity transaction supervision risk metrics evaluation can be processed, Uncertainty number is used to carry out electricity transaction supervision risk metrics evaluation, the various possibility value of evaluation result and corresponding confidence level can be obtained, realize risk indicator to quantize, and then the degree of risk of various index can be provided, be convenient to electricity transaction supervisor and carry out risk identification and judge risk class, aid decision making, realizes the effective supervision to risk.

Accompanying drawing explanation

Fig. 1 unifies interconnecting electric power marketing supervision risk pre-warning indexes system figure.

Fig. 2 is method flow diagram of the present invention.

Embodiment

Electricity transaction supervision risk index system: under current unified interconnecting electric power market environment, make every effort to accurate, objective, comprehensive reflection power market transaction supervision risk situation herein, and follow the principles such as susceptibility, representativeness, dynamic and operability, according to electricity market expertise, construct the transaction supervision risk pre-warning indexes system being applicable to unified interconnecting electric power market.Comprise 5 class one-level warning indexs: city's field coordination, trading program, contract and clearing, energy efficiency and operation of power networks, under every class first class index, comprise some two-level index, altogether 16 two-level index.Respectively, comprise under city's field coordination index: extra-high voltage transactions balances adjustment ratio, transregional electricity trade proportion transprovincially, transaction conclusion of the business electricity accounts for total electricity ratio, and transregional transactions balances transprovincially lacks adjustment ratio, the transregional utilization factor transprovincially of clean energy resource; Comprise under trading program index: base power generation completion rate of the plan, base power generation plan performs balanced rate, and unit utilizes hourage to hang upside down; Comprise under contract and clearing index: Filing of contract rate, clearing completion rate, clearing promptness rate; Comprise under energy efficiency index: the transregional rate of transprovincially dissolving of clean energy resource, transaction coal conservation year-on-year growth rate, flue dust CER year-on-year growth rate; Transregional passway for transmitting electricity transprovincially heavy duty ratio is comprised under operation of power networks index.

Below for first class index trading program subordinate two-level index, evaluation procedure is described:

For convenience of description, suppose have 3 electricity market experts to participate in evaluating, through the modeling of expert's synthetic reliability, the synthetic reliability obtaining three experts is respectively: expert 1-0.5, expert 2-0.3, expert 3-0.2; Each expert gives a mark in table 4 to trading program subordinate two-level index:

Table 4 trading program subordinate two-level index expert estimation table

	Expert 1	Expert 2	Expert 3
				A 1: base power generation completion rate of the plan	75	80	78
A 2: base power generation plan performs balanced rate	65	60	66
				A 3: unit utilizes hourage to hang upside down	40	50	35

Then the score value F Uncertainty number of above-mentioned each index is expressed as follows:

The weight of each index is defined as w={w by analytical hierarchy process ₁, w ₂, w ₃}={ 0.2,0.4,0.4}, what calculate first class index trading program belonging to above-mentioned 3 two-level index according to formula (5) must be divided into 27 rank Uncertainty number, get δ=10, draw first class index trading program evaluation result through suitable depression of order, Uncertainty number represents in a tabular form, in table 5:

Table 5 first class index trading program evaluation result

Score probable value	40.43	52.33	61.63	75.10
					Confidence level	0.105	0.270	0.365	0.260

The evaluation scoring event of first class index trading program as can be seen from Table 5, wherein 60 points (score value is higher, degree of risk is larger) more than possibility be 62.5%, illustrate in the power market transaction process in this stage, first class index trading program has certain risk, and relevant persons in charge should strengthen the supervision to trading program, makes early warning response for this index, and propose counter-measure, in order to avoid larger risk occurs.

Claims (1)

1., based on an electricity transaction supervision risk index Evaluation Method for Uncertainty number, it is characterized in that, described method step is as follows:

Described electricity transaction supervision risk index system, comprises 5 class one-level warning indexs: state of market, trading program, contract and clearing, energy efficiency and operation of power networks; Described electricity transaction supervision risk index system, some two-level index are comprised under every class first class index, 16 class two-level index altogether, be respectively, comprise under city's field coordination index: extra-high voltage transactions balances adjustment ratio, transregional electricity trade proportion transprovincially, transaction conclusion of the business electricity accounts for total electricity ratio, transregional transactions balances transprovincially lacks adjustment ratio, the transregional utilization factor transprovincially of clean energy resource; Comprise under trading program index: base power generation completion rate of the plan, base power generation plan performs balanced rate, and unit utilizes hourage to hang upside down; Comprise under contract and clearing index: Filing of contract rate, clearing completion rate, clearing promptness rate; Comprise under energy efficiency index: the transregional rate of transprovincially dissolving of clean energy resource, transaction coal conservation year-on-year growth rate, flue dust CER year-on-year growth rate; Transregional passway for transmitting electricity transprovincially heavy duty ratio is comprised under operation of power networks index;

Concrete steps are as follows:

(1) electricity market expert synthetic reliability modeling

The confidence level of expert, be also the authority of expert people's degree of trusting expert in other words, represent with α (0≤α≤1), α=1 represents that certain expert is the most credible, and α=0 represents that certain expert is least worth believing; In power market transaction supervision risk comprehensive evaluation, the confidence level of expert can be determined according to the academic title of expert, educational background, length of service three aspects; Evaluation operation is in table 1, and confidence level evaluation criteria is in table 2; In the sorted table of table 2, represent expert's credibility with the quantity scale of ten point system, 10 points the highest, and 1 point minimum, and the meaning of numeric representation is therebetween successively decreased;

Table 1 expert reliability evaluation form

Table 2 expert reliability evaluation criteria sorted table

Use ε _i(i=1,2,3) are respectively the evaluation score in expert academic title, educational background, the length of service, then the confidence level β of expert _irepresent with formula (1);

${\mathrm{\&beta;}}_i=\frac{\underset{i=1}{\overset{3}{\mathrm{\&Sigma;}}}{\mathrm{\&epsiv;}}_i}{30}---\left(1\right)$

β _ivalue more close to 1, represent that this expert is the most credible, the judgement done is more accurate, otherwise, β _iless, represent that this expert is least worth believing;

If the expert participating in power market transaction supervision risk comprehensive evaluation is: B ₁, B ₂, ∧, B _nconfidence level be respectively: order:

Claim α _ifor expert B _iabout expert group B ₁, B ₂, ∧, B _nsynthetic reliability, referred to as expert B _icomprehensive reliability; Claim

For expert group B ₁, B ₂, ∧, B _nsynthetic reliability; α in formula represents the authoritative degree evaluating expert group;

(2) uncertainty of electricity market expert opinion quantizes

Because the cognitive behavior of expert has uncertainties mathematics, available Uncertainty number represents the evaluation information of expert, is called that the uncertainty of expert opinion quantizes;

If A is an evaluation object in power market transaction supervision risk comprehensive evaluation, A ₁, A ₂, ∧, A _mfor m the index of A; N expert is had to participate in evaluating, if expert group B to A ₁, B ₂, ∧, B _nexpert with centesimal system to the m of A index factor assessment marking, obtain expert estimation table, in table 3; Wherein, to factor A _i, expert estimation is respectively C _i1, C _i2, ∧, C _in, expert's synthetic reliability is α ₁, α ₂, ∧, α _n, existing by score set (C _i1, C _i2, ∧, C _in) in score value be arranged as by getting a mode time identical: C _ij1, C _ij2, ∧, C _ijkso, for factor A _iuncertainty number f can be obtained _i(x):

Wherein, i=1,2, ∧, m; α _ij1, α _ij2, ∧, α _ijkbe marking value be respectively C _ij1, C _ij2, ∧, C _ijkthose expert's synthetic reliability sums, due to factor A _iweight be w _i, so the unascertained quantify value of evaluation object A is:

$C=\underset{i=1}{\overset{m}{\mathrm{\&Sigma;}}}{w}_i{f}_i\left(x\right)---\left(5\right)$

Table 3 expert estimation table

Title C is the uncertainty quantification value of the expert opinion of evaluation object A; Obvious 0≤E (C)≤100, when total confidence level α=1 of unascertained quantify value C, E (C) is real number; The expert that all participations that and if only if are evaluated is to evaluation index factor A _iall play 100 timesharing, E (C)=100, and if only if all expert is to institute factor of evaluation A _iall play 0 timesharing, E (C)=0.

(3) depression of order of high-order Uncertainty number

If high-order Uncertainty number P (formula 6) is the result of Uncertainty number once-through operation, by x _ineighborhood [x _i, x _i+ δ] in value merge according to a certain method, can exponent number be reduced, different according to different actual conditions δ value;

First, from i=1, x is located at ₁neighborhood [x _i, x _i+ δ] in value have x ₁, x ₂, ∧, x _k, corresponding confidence level is respectively α ₁, α ₂, ∧, α _k, replace x by formula 7 ₁, x ₂, ∧, x _k, replace α by formula 8 ₁, α ₂, ∧, α _k, formula is:

${\stackrel{\&OverBar;}{x}}_i=\frac{1}{\underset{k=1}{\overset{k}{\mathrm{\&Sigma;}}}{\mathrm{\&alpha;}}_i}\underset{i=1}{\overset{k}{\mathrm{\&Sigma;}}}{x}_i{\mathrm{\&alpha;}}_i---\left(7\right)$

${\stackrel{\&OverBar;}{\mathrm{\&alpha;}}}_i=\underset{i=1}{\overset{k}{\mathrm{\&Sigma;}}}{\mathrm{\&alpha;}}_i---\left(8\right)$

So just k-1 rank are fallen in Uncertainty number P, then remaining value has been continued in the same way to merge, by that analogy; After depression of order, Uncertainty number P can be expressed as